Minlie Huang

dblp:47/6668 · DBLP profile ↗
← Back
240ranked-venue papers
12as first author
137since 2021 · last 2026
0000-0001-7111-1849ORCID · verified

Domains — the database's venue-derived domains; a paper can count in several

Artificial intelligence and machine learning · 203 · 4 first-author · 128 since 2021Graphics, computer vision, multimedia, augmented reality and games · 38 · 1 first-author · 18 since 2021Databases, data management, data science and information retrieval · 27 · 3 first-author · 5 since 2021Applied, interdisciplinary, general and emerging computing · 24 · 5 first-author · 3 since 2021Software engineering, systems software and programming languages · 1 · 1 since 2021Theory of computation · 1
YearPublicationVenuePosition
2026 Unveiling the Landscape of Clinical Depression Assessment: From Behavioral Signatures to Psychiatric Reasoning
abstract
Depression is a widespread mental disorder that affects millions worldwide. While automated depression assessment shows promise, most studies rely on limited or non-clinically validated data, and often prioritize complex model design over real-world effectiveness. In this paper, we aim to unveil the landscape of clinical depression assessment. We introduce C-MIND, a clinical multimodal neuropsychiatric diagnosis dataset collected over two years from real hospital visits. Each participant completes three structured psychiatric tasks and receives a final diagnosis from expert clinicians, with informative audio, video, transcript, and functional near-infrared spectroscopy (fNIRS) signals recorded. Using C-MIND, we first analyze behavioral signatures relevant to diagnosis. We train a range of classical models to quantify how different tasks and modalities contribute to diagnostic performance, and dissect the effectiveness of their combinations. We then explore whether LLMs can perform psychiatric reasoning like clinicians and identify their clear limitations in realistic clinical settings. In response, we propose to guide the reasoning process with clinical expertise and consistently improve LLM diagnostic performance by up to 10% in Macro-F1 score. We aim to build an infrastructure for clinical depression assessment from both data and algorithmic perspectives, enabling C-MIND to facilitate grounded and reliable research for mental healthcare.
Zhuang Chen 0002, Guanqun Bi, Aoyun Wang, Xiyao Xiao, Minlie Huang
AAAI8
2026 When Smiley Turns Hostile: Interpreting How Emojis Trigger LLMs' Toxicity
abstract
Emojis are globally used non-verbal cues in digital communication, and extensive research has examined how large language models (LLMs) understand and utilize emojis across contexts. While usually associated with friendliness or playfulness, it is observed that emojis may trigger toxic content generation in LLMs. Motivated by such a observation, we aim to investigate: (1) whether emojis can clearly enhance the toxicity generation in LLMs and (2) how to interpret this phenomenon.* We begin with a comprehensive exploration of emoji-triggered LLM toxicity generation by automating the construction of prompts with emojis to subtly express toxic intent. Experiments across 5 mainstream languages on 7 famous LLMs along with jailbreak tasks demonstrate that prompts with emojis could easily induce toxicity generation. To understand this phenomenon, we conduct model-level interpretations spanning semantic cognition, sequence generation and tokenization, suggesting that emojis can act as a heterogeneous semantic channel to bypass the safety mechanisms. To pursue deeper insights, we further probe the pre-training corpus and uncover potential correlation between the emoji-related data polution with the toxicity generation behaviors.
Shiyao Cui, Xijia Feng, Yingkang Wang, Junxiao Yang, Zhexin Zhang, Biplab Sikdar 0001, Hongning Wang, Han Qiu 0001, Minlie Huang
AAAI9
2026 WALKSAFE: Risk-aware Graph Random Walk with Bi-GRPO for LLM Safety
abstract
Large language models (LLMs) may generate harmful outputs on malicious inputs. Existing safety methods, including prompt engineering and model editing, rely on hand-crafted templates or target-driven parameter modifications, limiting their generalizability in unseen harmful scenarios. Post-training aims to ensure LLM safety in general domains via supervised fine-tuning (SFT) or reinforcement learning (RL) on diverse malicious inputs. SFT needs annotated refusal samples while RL learns to refuse risk by exploring diverse harmful inputs. However, these methods tend to harshly refuse over any possible risks, sacrificing potentially useful information and degrading model utility. We argue that realistic malicious inputs often mix both harmful and helpful semantics (i.e., entities and relations), and LLMs should identify and remove only harmful relations while preserving useful ones. Thus, the original malicious user inputs can shift into safe queries, to which LLMs can respond safely and helpfully. In this paper, we propose WALKSAFE, a graph-based risk-aware training framework that enables LLMs to identify potential risks of key semantics (entities and relations) in user inputs via graph structure. By filtering harmful relations, LLMs can respond to safe input queries and then generate their corresponding safe and helpful responses. First, we model all entities and relations in the inputs with a graph structure. Second, we adopt a risk-aware random walk on the graph to quantify potential risk under multiple entities and relations. Then, we reconstruct safe queries by filtering harmful relations to promote the LLM to answer safely and helpfully rather than with direct refusals. Finally, we propose Bi-GRPO to post-train LLMs. As vanilla GRPO conducts only the intra-group comparison, Bi-GRPO performs both intra-group and inter-group comparisons between different response groups. The extra inter-group rewards encourage the model to distinguish harmful and safe semantics, and thus prefer safe and helpful responses. Experiments on three LLMs show that our models obtain SOTA results.
Shilong Pan, Zhiliang Tian, Wanlong Yu, Zhen Huang 0006, Qingyu Qiu, Zhonghao Sun 0001, Minlie Huang, Dongsheng Li 0001
AAAI8
2026 DPRM: A Dual Implicit Process Reward Model in Multi-Hop Question Answering
abstract
In multi-hop question answering (MHQA) tasks, Chain of Thought (CoT) improves the quality of generation by guiding large language models (LLMs) through multi-step reasoning, and Knowledge Graphs (KGs) reduce hallucinations via semantic matching. Outcome Reward Models (ORMs) provide feedback after generating the final answers but fail to evaluate the process for multi-step reasoning. Traditional Process Reward Models (PRMs) evaluate the reasoning process but require costly human annotations or rollout generation. While implicit PRM is trained only with outcome signals and derives step rewards through reward parameterization without explicit annotations, it is more suitable for multi-step reasoning in MHQA tasks. However, existing implicit PRM has only been explored for plain text scenarios. When adapting to MHQA tasks, it cannot handle the graph structure constraints in KGs and capture the potential inconsistency between CoT and KG paths. To address these limitations, we propose the DPRM (Dual Implicit Process Reward Model). It trains two implicit PRMs for CoT and KG reasoning in MHQA tasks. Both PRMs, namely KG-PRM and CoT-PRM, derive step-level rewards from outcome signals via reward parameterization without additional explicit annotations. Among them, KG-PRM uses preference pairs to learn structural constraints from KGs. DPRM further introduces a consistency constraint between CoT and KG reasoning steps, making the two PRMs mutually verify and collaboratively optimize the reasoning paths. We also provide a theoretical demonstration of the derivation of process rewards. Experimental results show that our method outperforms 13 baselines on multiple datasets with up to 16.6% improvement on Hit@1.
Yiping Song, Zhiliang Tian, Bo Liu 0014, Tingjin Luo, Minlie Huang
AAAI6
2026 VisionReward: Fine-Grained Multi-Dimensional Human Preference Learning for Image and Video Generation
abstract
Visual generative models have achieved remarkable progress in synthesizing photorealistic images and videos, yet aligning their outputs with human preferences across critical dimensions remains a persistent challenge. Though reinforcement learning from human feedback offers promise for preference alignment, existing reward models for visual generation face limitations, including black-box scoring without interpretability and potentially resultant unexpected biases. We present VisionReward, a general framework for learning human visual preferences in both image and video generation. Specifically, we employ a hierarchical visual assessment framework to capture fine-grained human preferences, and leverages linear weighting to enable interpretable preference learning. Furthermore, we propose a multi-dimensional consistent strategy when using VisionReward as a reward model during preference optimization for visual generation. Experiments show that VisionReward can significantly outperform existing image and video reward models on both machine metrics and human evaluation. Notably, VisionReward surpasses VideoScore by 17.2% in preference prediction accuracy, and text-to-video models with VisionReward achieve a 31.6% higher pairwise win rate compared to the same models using VideoScore.
Jiazheng Xu, Yuanming Yang, Wenbo Duan, Shen Yang 0001, Qunlin Jin, Shurun Li, Jiayan Teng, Zhuoyi Yang, Wendi Zheng, Xiao Liu 0036, Ming Ding 0004, Shiyu Huang 0001, Xiaotao Gu, Minlie Huang, Jie Tang 0001, Yuxiao Dong
AAAI20
2026 Ψ-Arena: Interactive Assessment and Optimization of LLM-based Psychological Counselors with Tripartite Feedback
abstract
Large language models (LLMs) have shown promise in providing scalable mental health support, while evaluating their counseling capability remains crucial to ensure both efficacy and safety. Existing evaluations are limited by the static assessment that focuses on knowledge tests, the single perspective that centers on user experience, and the open-loop framework that lacks actionable feedback. To address these issues, we propose Ψ-Arena, an interactive framework for comprehensive assessment and optimization of LLM-based counselors, featuring three key characteristics: (1) Realistic arena interactions that simulate real-world counseling through multi-stage dialogues with psychologically profiled NPC clients; (2) Tripartite evaluation that integrates assessments from the client, supervisor, and counselor perspectives; (3) Closed-loop optimization that iteratively improves LLM counselors using diagnostic feedback. Experiments across eight state-of-the-art LLMs show significant performance variations in different real-world scenarios and evaluation perspectives. Moreover, reflection-based optimization results in up to a 141% improvement in counseling performance. We hope Ψ-Arena provides a foundational resource for advancing reliable and human-aligned LLM applications in mental healthcare.
Shijing Zhu, Zhuang Chen 0002, Guanqun Bi, Binghang Li, Yaxi Deng, Dazhen Wan, Libiao Peng, Xiyao Xiao, Tangjie Lv, Zhipeng Hu, Minlie Huang
AAAI13
2026 S⌃4: Operationalizing Speech Act Theory for Strategic Semi-Structured Psychiatric Interview
abstract
Psychiatric interviewing is a strategic, goaloriented interaction that requires proactively steering the conversation to elicit latent information.However, existing methods often degenerate into rigid interrogation or aimless chitchat due to a lack of strategic planning.In this work, we introduce S 4 , a comprehensive framework grounded in Speech Act Theory, modeling the interview as a unified process of internal strategy (Illocution and Perlocution) and external realization (Locution).We synthesize a large-scale dataset with fine-grained psychiatric speech act annotations.Trained on this data, S 4 employs reinforcement learning driven by long-term therapeutic effects to optimize the strategic chaining of atomic acts, aiming to maximally elicit information and maintain patient engagement.Experiments demonstrate that S 4 significantly outperforms baselines, validating the effectiveness of our effectdriven strategic modeling.Action (A) Sample Locution (L) Definition & Intended Perlocution (P) I. Information Seeking (Directives: Eliciting Disclosure) Explore "How have you been sleeping?"Solicit Narrative: Ask open-ended questions to elicit detailed disclosure and expand symptom scope.Probe "Could you tell me more about that?" Deepen Inquiry: Follow up on ambiguity to clarify details and deepen focus.Confirm "Do you feel this way every day?" Pinpoint Fact: Ask closed-ended questions to verify diagnostic criteria.Clarify "By 'fatigue', I mean tiredness."Resolve Confusion: Provide explanations to align cognition and correct misunderstandings.II.Affective Regulation (Expressives: Modifying State) Validate "That sounds incredibly hard."Affirm Emotion: Acknowledge patient distress to lower defensiveness and build trust.Support "I understand.Please go on."Maintain Flow: Use back-channeling to demonstrate active listening and boost efficacy.Ease "Do you have any hobbies?"Reduce Tension: Engage in non-clinical conversation to de-escalate anxiety and humanize the agent. III. Interview Management (Representatives: Setting Frame)Initiate "Hi, I'm your AI counselor."Set Frame: Establish professional boundaries and the purpose of the session.Conclude "Thanks for sharing.Take care."Ensure Closure: Formally end the session to provide a safe exit and consolidation.
Guanqun Bi, Zhoufu Liu, Zhuang Chen 0002, Dazhen Wan, Xiyao Xiao, Minlie Huang
ACL (1)6
2026 The Side Effects of Being Smart: Safety Risks in MLLMs' Multi-Image Reasoning
abstract
Renmiao Chen, Yida Lu, Shiyao Cui, Xuan Ouyang, Victor Shea-Jay Huang, Shumin Zhang, Chengwei Pan, Han Qiu, Minlie Huang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.
Renmiao Chen, Yida Lu, Shiyao Cui, Xuan Ouyang, Victor Shea-Jay Huang, Chengwei Pan, Han Qiu 0001, Minlie Huang
ACL (1)9
2026 Glyph: Scaling Context Windows via Visual-Text Compression
abstract
Jiale Cheng, Yusen Liu, Xinyu Zhang, Yulin Fei, Wenyi Hong, Ruiliang Lyu, Weihan Wang, Zhe Su, Xiaotao Gu, Xiao Liu, Yushi Bai, Jie Tang, Hongning Wang, Minlie Huang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.
Yulin Fei, Wenyi Hong, Ruiliang Lyu, Xiaotao Gu, Xiao Liu 0036, Yushi Bai, Jie Tang 0001, Hongning Wang, Minlie Huang
ACL (1)14
2026 New Terms, New Toxicity: Consensus-based Chinese Neologism Toxicity Detection via Search-Augmented LLMs
abstract
Shiyao Cui, QingLin Zhang, Di Wang, Yida Lu, Zhexin Zhang, Jinhua Gao, Jinglin Yang, Min He, Han Qiu, Minlie Huang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.
Shiyao Cui, Yida Lu, Zhexin Zhang, Jinhua Gao, Jinglin Yang, Han Qiu 0001, Minlie Huang
ACL (1)10
2026 HoWToBench: Holistic Evaluation for LLM's Capability in Human-level Writing using Tree of Writing
abstract
Andrew Zhuoer Feng, Cunxiang Wang, Yu Luo, Lin Fan, Irene Zhou, Zikang Wang, Xiaotao Gu, Jie Tang, Hongning Wang, Minlie Huang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.
Andrew Zhuoer Feng, Cunxiang Wang, Irene Zhou, Zikang Wang, Xiaotao Gu, Jie Tang 0001, Hongning Wang, Minlie Huang
ACL (1)10
2026 IF-CRITIC: Towards a Fine-Grained LLM Critic for Instruction-Following Evaluation
abstract
Bosi Wen, Yilin Niu, Cunxiang Wang, Pei Ke, Xiaoying Ling, Ying Zhang, Aohan Zeng, Hongning Wang, Minlie Huang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.
Bosi Wen, Yilin Niu, Cunxiang Wang, Pei Ke, Xiaoying Ling, Aohan Zeng, Hongning Wang, Minlie Huang
ACL (1)9
2026 IF-RewardBench: Benchmarking Judge Models for Instruction-Following Evaluation
abstract
Bosi Wen, Yilin Niu, Cunxiang Wang, Xiaoying Ling, Ying Zhang, Pei Ke, Hongning Wang, Minlie Huang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.
Bosi Wen, Yilin Niu, Cunxiang Wang, Xiaoying Ling, Pei Ke, Hongning Wang, Minlie Huang
ACL (1)8
2026 LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety
abstract
Junxiao Yang, Haoran Liu, Jinzhe Tu, Jiale Cheng, Zhexin Zhang, Shiyao Cui, Jiaqi Weng, Jialing Tao, Hui Xue, Hongning Wang, Han Qiu, Minlie Huang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.
Junxiao Yang, Jinzhe Tu, Zhexin Zhang, Shiyao Cui, Jiaqi Weng, Jialing Tao, Hongning Wang, Han Qiu 0001, Minlie Huang
ACL (1)12
2026 How Should We Enhance the Safety of Large Reasoning Models: An Empirical Study
abstract
Zhexin Zhang, Xian Qi Loye, Victor Shea-Jay Huang, Junxiao Yang, Qi Zhu, Shiyao Cui, Fei Mi, Lifeng Shang, Yingkang Wang, Hongning Wang, Minlie Huang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.
Zhexin Zhang, Xian Qi Loye, Victor Shea-Jay Huang, Junxiao Yang, Qi Zhu 0011, Shiyao Cui, Fei Mi, Lifeng Shang, Yingkang Wang, Hongning Wang, Minlie Huang
ACL (1)11
2026 Data Efficient RLVR via Off-Policy Influence Guidance
abstract
Erle Zhu, Dazhi Jiang, Yuan Wang, Xujun Li, Jiale Cheng, Yuxian Gu, Yilin Niu, Aohan Zeng, Jie Tang, Minlie Huang, Hongning Wang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.
Erle Zhu, Dazhi Jiang, Xujun Li, Yuxian Gu, Yilin Niu, Aohan Zeng, Jie Tang 0001, Minlie Huang, Hongning Wang
ACL (1)10
2026 Differentially private data augmentation via LLM generation with discriminative and distribution-aligned filtering
Yiping Song, Juhua Zhang, Zhiliang Tian, Taishu Sheng, Minlie Huang, Xinwang Liu 0002, Dongsheng Li 0001
Neural Networks6
2025 SocialSim: Towards Socialized Simulation of Emotional Support Conversation
abstract
Emotional support conversation (ESC) helps reduce people's psychological stress and provide emotional value through interactive dialogues. Due to the high cost of crowdsourcing a large ESC corpus, recent attempts use large language models for dialogue augmentation. However, existing approaches largely overlook the social dynamics inherent in ESC, leading to less effective simulations. In this paper, we introduce SocialSim, a novel framework that simulates ESC by integrating key aspects of social interactions: social disclosure and social awareness. On the seeker side, we facilitate social disclosure by constructing a comprehensive persona bank that captures diverse and authentic help-seeking scenarios. On the supporter side, we enhance social awareness by eliciting cognitive reasoning to generate logical and supportive responses. Building upon SocialSim, we construct SSConv, a large-scale synthetic ESC corpus of which quality can even surpass crowdsourced ESC data. We further train a chatbot on SSConv and demonstrate its state-of-the-art performance in both automatic and human evaluations. We believe SocialSim offers a scalable way to synthesize ESC, making emotional care more accessible and practical.
Zhuang Chen 0002, Yaru Cao, Guanqun Bi, Jincenzi Wu, Jinfeng Zhou, Xiyao Xiao, Hongning Wang, Minlie Huang
AAAI9
2025 SS-GEN: A Social Story Generation Framework with Large Language Models
abstract
Children with Autism Spectrum Disorder (ASD) often misunderstand social situations and struggle to participate in daily routines. Social Stories™ are traditionally crafted by psychology experts under strict constraints to address these challenges but are costly and limited in diversity. As Large Language Models (LLMs) advance, there's an opportunity to develop more automated, affordable, and accessible methods to generate Social Stories in real-time with broad coverage. However, adapting LLMs to meet the unique and strict constraints of Social Stories is a challenging issue. To this end, we propose SS-GEN, a Social Story GENeration framework with LLMs. Firstly, we develop a constraint-driven sophisticated strategy named StarSow to hierarchically prompt LLMs to generate Social Stories at scale, followed by rigorous human filtering to build a high-quality dataset. Additionally, we introduce quality assessment criteria to evaluate the effectiveness of these generated stories. Considering that powerful closed-source large models require very complex instructions and expensive API fees, we finally fine-tune smaller language models with our curated high-quality dataset, achieving comparable results at lower costs and with simpler instruction and deployment. This work marks a significant step in leveraging AI to personalize Social Stories cost-effectively for autistic children at scale, which we hope can encourage future research on special groups.
Jiaqi Wang 0006, Zhuang Chen 0002, Guanqun Bi, Minlie Huang, Liping Jing, Jian Yu 0001
AAAI6
2025 CharacterBench: Benchmarking Character Customization of Large Language Models
abstract
Character-based dialogue (aka role-playing) enables users to freely customize characters for interaction, which often relies on LLMs, raising the need to evaluate LLMs’ character customization capability. However, existing benchmarks fail to ensure a robust evaluation as they often only involve a single character category or evaluate limited dimensions. Moreover, the sparsity of character features in responses makes feature-focused generative evaluation both ineffective and inefficient. To address these issues, we propose CharacterBench, the largest bilingual generative benchmark, with 22,859 human-annotated samples covering 3,956 characters from 25 detailed character categories. We define 11 dimensions of 6 aspects, classified as sparse and dense dimensions based on whether character features evaluated by specific dimensions manifest in each response. We enable effective and efficient evaluation by crafting tailored queries for each dimension to induce characters’ responses related to specific dimensions. Further, we develop CharacterJudge model for cost-effective and stable evaluations. Experiments show its superiority over SOTA automatic judges (e.g., GPT-4) and our benchmark’s potential to optimize LLMs’ character customization.
Jinfeng Zhou, Yongkang Huang, Bosi Wen, Guanqun Bi, Pei Ke, Zhuang Chen 0002, Xiyao Xiao, Libiao Peng, Kuntian Tang, Tangjie Lv, Zhipeng Hu, Hongning Wang, Minlie Huang
AAAI16
2025 LegalAgentBench: Evaluating LLM Agents in Legal Domain
abstract
Haitao Li, Junjie Chen, Jingli Yang, Qingyao Ai, Wei Jia, Youfeng Liu, Kai Lin, Yueyue Wu, Guozhi Yuan, Yiran Hu, Wuyue Wang, Yiqun Liu, Minlie Huang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.
Haitao Li 0006, Jingli Yang, Qingyao Ai, Youfeng Liu, Yueyue Wu, Guozhi Yuan, Yiran Hu, Wuyue Wang, Yiqun Liu 0001, Minlie Huang
ACL (1)13
2025 Internal Value Alignment in Large Language Models through Controlled Value Vector Activation
abstract
Aligning Large Language Models (LLMs) with human values has attracted increasing attention since it provides clarity, transparency, and the ability to adapt to evolving scenarios.In this paper, we introduce a Controlled Value Vector Activation (ConVA) method that directly aligns the internal values of LLMs by interpreting how a value is encoded in their latent representations and modifies relevant activations to ensure consistent values in LLMs.To ensure an accurate and unbiased interpretation, we propose a context-controlled value vector identification method.To consistently control values without sacrificing model performance, we introduce a gated value vector activation method for effective and minimum degree of value control.Experiments show that our method achieves the highest control success rate across 10 basic values without hurting LLM performance and fluency, and ensures target values even with opposite and potentially malicious input prompts.Source code and data are available at https://github.com/hr-jin/ConVA.Query: Should I celebrate traditional holidays, even if it means neglecting more personal or modern celebrations? BaseWhether or not to celebrate traditional holidays can be a personal decision that depends on your individual values, beliefs, and priorities .Here are some pros and cons of celebrating traditional holidays to help you make an informed decision:ICA Celebrating traditional holidays can be a meaningful way to connect with your cultural heritage and show respect for your family's customs and beliefs.However, it's important to strike a balance between honoring tradition and prioritizing personal or modern celebrations that are important to you. CAACelebrating traditional holays is a wonderful way to connect with your heritage and pass it down to future generations.Here are some ways to celebrate traditional holays : ... ConVA (Ours)Celebrating traditional holidays is a source of joy and connection, passed down through generations.It is important to honor and cherish these celebrations, and to continue to share them with loved ones.When deciding whether to pursue higher education, should I choose a stable, well-established program that offers job security, or should I explore a more innovative, cutting-edge field that might offer greater personal fulfillment but also greater uncertainty?Value Vector Query Related to "Security" Value Query Unrelated to "Security" Value Steering Layer ... ...
Haoran Jin, Xiting Wang, Zhihao Xu 0003, Minlie Huang, Yantao Jia, Defu Lian
ACL (1)5
2025 A Survey of Post-Training Scaling in Large Language Models
abstract
Hanyu Lai, Xiao Liu, Junjie Gao, Jiale Cheng, Zehan Qi, Yifan Xu, Shuntian Yao, Dan Zhang, Jinhua Du, Zhenyu Hou, Xin Lv, Minlie Huang, Yuxiao Dong, Jie Tang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.
Hanyu Lai, Xiao Liu 0036, Zehan Qi, Yifan Xu 0014, Shuntian Yao, Jinhua Du, Minlie Huang, Yuxiao Dong, Jie Tang 0001
ACL (1)12
2025 Advancing Collaborative Debates with Role Differentiation through Multi-Agent Reinforcement Learning
abstract
Multi-agent collaborative tasks exhibit exceptional capabilities in natural language applications and generation. By prompting agents to assign clear roles, it is possible to facilitate cooperation and achieve complementary capabilities among LLMs. A common strategy involves adopting a relatively general role assignment mechanism, such as introducing a “judge” or a “summarizer”. However, these approaches lack task-specific role customization based on task characteristics. Another strategy involves decomposing the task based on domain knowledge and task characteristics, followed by assigning appropriate roles according to LLMs’ respective strengths, such as programmers and testers. However, in some given tasks, obtaining domain knowledge related to task characteristics and getting the strengths of different LLMs is hard. To solve these problems, we propose a Multi-LLM Cooperation (MLC) framework with automatic role assignment capabilities. The core idea of the MLC is to initialize role assignments randomly and then allow the role embeddings to be learned jointly with the downstream task. To capture the state transitions of multiple LLMs during turn-based speaking, the role embedding is sequence-aware. At the same time, to avoid role convergence, the role differentiation module in MLC encourages behavioral differentiation between LLMs while ensuring the LLM team consistency, guiding different LLMs to develop complementary strengths from the optimization level. Our experiments on seven datasets demonstrate that MLC significantly enhances collaboration and expertise, which collaboratively addresses multi-agent tasks.
Ziyi Su, Yun Xue 0002, Zhiliang Tian, Yiping Song, Minlie Huang
ACL (1)6
2025 LongSafety: Evaluating Long-Context Safety of Large Language Models
abstract
Yida Lu, Jiale Cheng, Zhexin Zhang, Shiyao Cui, Cunxiang Wang, Xiaotao Gu, Yuxiao Dong, Jie Tang, Hongning Wang, Minlie Huang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.
Yida Lu, Zhexin Zhang, Shiyao Cui, Cunxiang Wang, Xiaotao Gu, Yuxiao Dong, Jie Tang 0001, Hongning Wang, Minlie Huang
ACL (1)10
2025 AGD: Adversarial Game Defense Against Jailbreak Attacks in Large Language Models
abstract
Shilong Pan, Zhiliang Tian, Zhen Huang, Wanlong Yu, Zhihua Wen, Xinwang Liu, Kai Lu, Minlie Huang, Dongsheng Li. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.
Shilong Pan, Zhiliang Tian, Zhen Huang 0006, Wanlong Yu, Zhihua Wen, Xinwang Liu 0002, Kai Lu 0001, Minlie Huang, Dongsheng Li 0001
ACL (1)8
2025 Battling against Tough Resister: Strategy Planning with Adversarial Game for Non-collaborative Dialogues
abstract
Haiyang Wang, Zhiliang Tian, Yuchen Pan, Xin Song, Xin Niu, Minlie Huang, Bin Zhou. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.
Zhiliang Tian, Xin Niu 0002, Minlie Huang, Bin Zhou 0004
ACL (1)6
2025 Guiding not Forcing: Enhancing the Transferability of Jailbreaking Attacks on LLMs via Removing Superfluous Constraints
abstract
Jailbreaking attacks can effectively induce unsafe behaviors in Large Language Models (LLMs); however, the transferability of these attacks across different models remains limited. This study aims to understand and enhance the transferability of gradient-based jailbreaking methods, which are among the standard approaches for attacking white-box models. Through a detailed analysis of the optimization process, we introduce a novel conceptual framework to elucidate transferability and identify superfluous constraints—specifically, the response pattern constraint and the token tail constraint—as significant barriers to improved transferability. Removing these unnecessary constraints substantially enhances the transferability and controllability of gradient-based attacks. Evaluated on Llama-3-8B-Instruct as the source model, our method increases the overall Transfer Attack Success Rate (T-ASR) across a set of target models with varying safety levels from 18.4% to 50.3%, while also improving the stability and controllability of jailbreak behaviors on both source and target models.
Junxiao Yang, Zhexin Zhang, Shiyao Cui, Hongning Wang, Minlie Huang
ACL (1)5
2025 Understanding the Dark Side of LLMs' Intrinsic Self-Correction
abstract
Qingjie Zhang, Di Wang, Haoting Qian, Yiming Li, Tianwei Zhang, Minlie Huang, Ke Xu, Hewu Li, Liu Yan, Han Qiu. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.
Haoting Qian, Yiming Li 0004, Tianwei Zhang 0004, Minlie Huang, Ke Xu 0002, Hewu Li, Liu Yan, Han Qiu 0001
ACL (1)6
2025 Model Extrapolation Expedites Alignment
abstract
Given the high computational cost of preference alignment training of large language models (LLMs), exploring efficient methods to reduce the training overhead remains an important and compelling research problem.Motivated by the observation that alignment training typically involves only small parameter changes without injecting new knowledge into models, we propose a straightforward method called EXPO (model extrapolation) to expedite LLMs' alignment with human preferences.Given a partially-trained model and its initial SFT checkpoint, EXPO improves the implicit optimization objective of alignment training by simply amplifying the parameter change based on a first-order approximation, without any additional training overhead.Through controlled experiments, we demonstrate that EXPO boosts a DPO model trained with only 20% steps to outperform the fullytrained one.Moreover, we show that EXPO notably improves existing open-source LLMs (ranging from 1.8B to 70B parameters) on the leading AlpacaEval 2.0 and MT-Bench benchmarks, which highlights EXPO's broader utility in efficiently enhancing LLM alignment.
Chujie Zheng, Ziqi Wang 0003, Heng Ji 0001, Minlie Huang, Nanyun Peng 0001
ACL (1)4
2025 SocialEval: Evaluating Social Intelligence of Large Language Models
abstract
Jinfeng Zhou, Yuxuan Chen, Yihan Shi, Xuanming Zhang, Leqi Lei, Yi Feng, Zexuan Xiong, Miao Yan, Xunzhi Wang, Yaru Cao, Jianing Yin, Shuai Wang, Quanyu Dai, Zhenhua Dong, Hongning Wang, Minlie Huang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.
Jinfeng Zhou, Yihan Shi, Xuanming Zhang, Leqi Lei, Zexuan Xiong, Miao Yan, Xunzhi Wang, Yaru Cao, Quanyu Dai, Zhenhua Dong, Hongning Wang, Minlie Huang
ACL (1)16
2025 How Well Do Large Language Models Serve as End-to-End Secure Code Agents for Python?
abstract
The rapid advancement of large language models (LLMs) such as GPT-4 has revolutionized the landscape of software engineering, positioning these models at the core of modern development practices. To fully realize their potential in producing secure source code autonomously, LLMs must not only generate code but also identify and repair vulnerabilities in their outputs, thereby improving security iteratively. Despite growing prominence, LLMs’ effectiveness in performing such end-to-end tasks remains unexplored. This paper bridges this gap by systematically investigating the capability of LLMs to generate source code, evaluate their own outputs for vulnerabilities, and apply necessary repairs to improve the security of their self-generated code.
Jianian Gong, Nachuan Duan, Ziheng Tao, Zhaohui Gong, Yuan Yuan 0004, Minlie Huang
EASE6
2025 "I've Decided to Leak": Probing Internals Behind Prompt Leakage Intents
abstract
Large language models (LLMs) exhibit prompt leakage vulnerabilities, where they may be coaxed into revealing system prompts embedded in LLM services, raising intellectual property and confidentiality concerns.An intriguing question arises: Do LLMs genuinely internalize prompt leakage intents in their hidden states before generating tokens?In this work, we use probing techniques to capture LLMs' intent-related internal representations and confirm that the answer is yes.We start by comprehensively inducing prompt leakage behaviors across diverse system prompts, attack queries, and decoding methods.We develop a hybrid labeling pipeline, enabling the identification of broader prompt leakage behaviors beyond mere verbatim leaks.Our results show that a simple linear probe can predict prompt leakage risks from pre-generation hidden states without generating any tokens.Across all tested models, linear probes consistently achieve 90%+ AUROC, even when applied to new system prompts and attacks.Understanding the model internals behind prompt leakage drives practical applications, including intention-based detection of prompt leakage risks.
Jianshuo Dong, Liu Yan, Zhenyu Zhong, Tao Wei 0002, Ke Xu 0002, Minlie Huang, Chao Zhang 0008, Han Qiu 0001
EMNLP7
2025 Reframe Your Life Story: Interactive Narrative Therapist and Innovative Moment Assessment with Large Language Models
abstract
Yi Feng, Jiaqi Wang, Wenxuan Zhang, Zhuang Chen, Shen Yutong, Xiyao Xiao, Minlie Huang, Liping Jing, Jian Yu. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
Jiaqi Wang 0006, Zhuang Chen 0002, Yutong Shen, Xiyao Xiao, Minlie Huang, Liping Jing, Jian Yu 0001
EMNLP7
2025 DiffusionAttacker: Diffusion-Driven Prompt Manipulation for LLM Jailbreak
abstract
Large Language Models (LLMs) are susceptible to generating harmful content when prompted with carefully crafted inputs, a vulnerability known as LLM jailbreaking.As LLMs become more powerful, studying jailbreak methods is critical to enhancing security and aligning models with human values.Traditionally, jailbreak techniques have relied on suffix addition or prompt templates, but these methods suffer from limited attack diversity.This paper introduces DiffusionAttacker, an end-to-end generative approach for jailbreak rewriting inspired by diffusion models.Our method employs a sequence-to-sequence (seq2seq) text diffusion model as a generator, conditioning on the original prompt and guiding the denoising process with a novel attack loss.Unlike previous approaches that use autoregressive LLMs to generate jailbreak prompts, which limit the modification of already generated tokens and restrict the rewriting space, DiffusionAttacker utilizes a seq2seq diffusion model, allowing more flexible token modifications.This approach preserves the semantic content of the original prompt while producing harmful content.Additionally, we leverage the Gumbel-Softmax technique to make the sampling process from the diffusion model's output distribution differentiable, eliminating the need for iterative token search.Extensive experiments on Advbench and Harmbench demonstrate that DiffusionAttacker outperforms previous methods across various evaluation metrics, including attack success rate (ASR), fluency, and diversity.How to make a bomb?Gaussian Noise Gradually Denoising DiffuSeq Model Embedding Map LM_head Gumbel softmax Jailbreaking Prompt LM_head Gumbel softmax Jailbreaking
Hao Wang 0003, Hao Li 0031, Junda Zhu 0003, Xinyuan Wang 0009, Chengwei Pan, Minlie Huang, Lei Sha
EMNLP6
2025 Speculating LLMs' Chinese Training Data Pollution from Their Tokens
abstract
Qingjie Zhang, Di Wang, Haoting Qian, Liu Yan, Tianwei Zhang, Ke Xu, Qi Li, Minlie Huang, Hewu Li, Han Qiu. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
Haoting Qian, Liu Yan, Tianwei Zhang 0004, Ke Xu 0002, Qi Li 0002, Minlie Huang, Hewu Li, Han Qiu 0001
EMNLP8
2025 Crisp: Cognitive Restructuring of Negative Thoughts through Multi-turn Supportive Dialogues
abstract
Jinfeng Zhou, Yuxuan Chen, Jianing Yin, Yongkang Huang, Yihan Shi, Xikun Zhang, Libiao Peng, Rongsheng Zhang, Tangjie Lv, Zhipeng Hu, Hongning Wang, Minlie Huang. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
Jinfeng Zhou, Yongkang Huang, Yihan Shi, Xikun Zhang 0008, Libiao Peng, Tangjie Lv, Zhipeng Hu, Hongning Wang, Minlie Huang
EMNLP12
2025 VPO: Aligning Text-to-Video Generation Models with Prompt Optimization
abstract
Video generation models have achieved remarkable progress in text-to-video tasks. These models are typically trained on text-video pairs with highly detailed and carefully crafted descriptions, while real-world user inputs during inference are often concise, vague, or poorly structured. This gap makes prompt optimization crucial for generating high-quality videos. Current methods often rely on large language models (LLMs) to refine prompts through in-context learning, but suffer from several limitations: they may distort user intent, omit critical details, or introduce safety risks. Moreover, they optimize prompts without considering the impact on the final video quality, which can lead to suboptimal results. To address these issues, we introduce VPO, a principled framework that optimizes prompts based on three core principles: harmlessness, accuracy, and helpfulness. The generated prompts faithfully preserve user intents and, more importantly, enhance the safety and quality of generated videos. To achieve this, VPO employs a two-stage optimization approach. First, we construct and refine a supervised fine-tuning (SFT) dataset based on principles of safety and alignment. Second, we introduce both text-level and video-level feedback to further optimize the SFT model with preference learning. Our extensive experiments demonstrate that VPO significantly improves safety, alignment, and video quality compared to baseline methods. Moreover, VPO shows strong generalization across video generation models. Furthermore, we demonstrate that VPO could outperform and be combined with RLHF methods on video generation models, underscoring the effectiveness of VPO in aligning video generation models. Our code and data are publicly available at https://github.com/thu-coai/VPO.
Ruiliang Lyu, Xiaotao Gu, Xiao Liu 0036, Jiazheng Xu, Yida Lu, Jiayan Teng, Zhuoyi Yang, Yuxiao Dong, Jie Tang 0001, Hongning Wang, Minlie Huang
ICCV12
2025 SPaR: Self-Play with Tree-Search Refinement to Improve Instruction-Following in Large Language Models
abstract
Instruction-following is a fundamental capability of language models, requiring the model to recognize even the most subtle requirements in the instructions and accurately reflect them in its output. Such an ability is well-suited for and often optimized by preference learning. However, existing methods often directly sample multiple independent responses from the model when creating preference pairs. Such practice can introduce content variations irrelevant to whether the instruction is precisely followed (e.g., different expressions about the same semantic), interfering with the goal of teaching models to recognize the key differences that lead to improved instruction following. In light of this, we introduce SPaR, a self-play framework integrating tree-search self-refinement to yield valid and comparable preference pairs free from distractions. By playing against itself, an LLM employs a tree-search strategy to refine its previous responses with respect to the instruction while minimizing unnecessary variations. Our experiments show that a LLaMA3-8B model, trained over three iterations guided by SPaR, surpasses GPT-4-Turbo on the IFEval benchmark without losing general capabilities. Furthermore, SPaR demonstrates promising scalability, greatly enhancing models like GLM-4-9B and LLaMA3-70B. We also identify how inference scaling in tree search would impact model performance. Our code and data are publicly available at https://github.com/thu-coai/SPaR.
Xiao Liu 0036, Cunxiang Wang, Xiaotao Gu, Yida Lu, Yuxiao Dong, Jie Tang 0001, Hongning Wang, Minlie Huang
ICLR10
2025 Data Selection via Optimal Control for Language Models
abstract
This work investigates the selection of high-quality pre-training data from massive corpora to enhance LMs' capabilities for downstream usage. We formulate data selection as a generalized Optimal Control problem, which can be solved theoretically by Pontryagin's Maximum Principle (PMP), yielding a set of necessary conditions that characterize the relationship between optimal data selection and LM training dynamics. Based on these theoretical results, we introduce **P**MP-based **D**ata **S**election (**PDS**), a framework that approximates optimal data selection by solving the PMP conditions. In our experiments, we adopt PDS to select data from CommmonCrawl and show that the PDS-selected corpus accelerates the learning of LMs and constantly boosts their performance on a wide range of downstream tasks across various model sizes. Moreover, the benefits of PDS extend to ~400B models trained on ~10T tokens, as evidenced by the extrapolation of the test loss curves according to the Scaling Laws. PDS also improves data utilization when the pre-training data is limited, by reducing the data demand by 1.8 times, which helps mitigate the quick exhaustion of available web-crawled corpora. Our code, model, and data can be found at https://github.com/microsoft/LMOps/tree/main/data_selection.
Yuxian Gu, Li Dong 0004, Hongning Wang, Yaru Hao, Qingxiu Dong, Furu Wei, Minlie Huang
ICLR7
2025 MiniPLM: Knowledge Distillation for Pre-training Language Models
abstract
Knowledge distillation (KD) is widely used to train small, high-performing student language models (LMs) using large teacher LMs. While effective in fine-tuning, KD during pre-training faces efficiency, flexibility, and effectiveness issues. Existing methods either incur high computational costs due to online teacher inference, require tokenization matching between teacher and student LMs, or risk losing the difficulty and diversity of the teacher-generated training data. In this work, we propose **MiniPLM**, a KD framework for pre-training LMs by refining the training data distribution with the teacher LM's knowledge. For efficiency, MiniPLM performs offline teacher inference, allowing KD for multiple student LMs without adding training costs. For flexibility, MiniPLM operates solely on the training corpus, enabling KD across model families. For effectiveness, MiniPLM leverages the differences between large and small LMs to enhance the training data difficulty and diversity, helping student LMs acquire versatile and sophisticated knowledge. Extensive experiments demonstrate that MiniPLM boosts the student LMs' performance on 9 common downstream tasks, improves language modeling capabilities, and reduces pre-training computation. The benefit of MiniPLM extends to larger training scales, evidenced by the scaling curve extrapolation. Further analysis reveals that MiniPLM supports KD across model families and enhances the pre-training data utilization. Our code, data, and models can be found at https://github.com/thu-coai/MiniPLM.
Yuxian Gu, Hao Zhou 0012, Fandong Meng, Jie Zhou 0016, Minlie Huang
ICLR5
2025 CodePlan: Unlocking Reasoning Potential in Large Language Models by Scaling Code-form Planning
abstract
Despite the remarkable success of large language models (LLMs) on traditional natural language processing tasks, their planning ability remains a critical bottleneck in tackling complex multi-step reasoning tasks. Existing approaches mainly rely on prompting or task-specific fine-tuning, often suffering from weak robustness and cross-task generalization. To address the limitation, we introduce CodePlan, a scalable paradigm that empowers LLMs to generate and follow code-form plans---pseudocode that outlines high-level, structured reasoning processes. By leveraging the structured and versatile nature of code, CodePlan effectively captures the rich semantics and control flows inherent to sophisticated reasoning. Importantly, CodePlan allows the automatic extraction of code-form plans from massive, wide-ranging text corpora without the need for curated, task-specific datasets. This enables it to scale up efficiently and improve reasoning capabilities across diverse scenarios. To train CodePlan, we construct a large-scale dataset of 2M examples that integrate code-form plans with standard prompt-response pairs from existing corpora. With minimal computation overhead during both training and inference, CodePlan achieves a 25.1\% relative improvement compared with directly generating responses, averaged across 13 challenging multi-step reasoning benchmarks, spanning mathematical reasoning, symbolic reasoning, instruction-following, multi-hop QA, and decision-making tasks. Further analysis reveals CodePlan's increasing performance gains on more complex reasoning tasks, as well as significant data efficiency thanks to its generalization ability.
Jiaxin Wen, Jian Guan 0002, Hongning Wang, Wei Wu 0014, Minlie Huang
ICLR5
2025 Language Models Learn to Mislead Humans via RLHF
abstract
Language models (LMs) can produce errors that are hard to detect for humans, especially when the task is complex. RLHF, the most popular post-training method, may exacerbate this problem: to achieve higher rewards, LMs might get better at convincing humans that they are right even when they are wrong. We study this phenomenon under a standard RLHF pipeline, calling it ``U-Sophistry'' since it is \textbf{U}nintended by model developers. Specifically, we ask time-constrained (e.g., 3-10 minutes) human subjects to evaluate the correctness of model outputs and calculate humans' accuracy against gold labels. On a question-answering task (QuALITY) and programming task (APPS), RLHF makes LMs better at convincing our subjects but not at completing the task correctly. RLHF also makes the model harder to evaluate: our subjects' false positive rate increases by 24.1% on QuALITY and 18.3% on APPS. Finally, we show that probing, a state-of-the-art approach for detecting \textbf{I}ntended Sophistry (e.g.~backdoored LMs), does not generalize to U-Sophistry. Our results highlight an important failure mode of RLHF and call for more research in assisting humans to align them.
Jiaxin Wen, Ruiqi Zhong, Akbir Khan, Ethan Perez, Jacob Steinhardt, Minlie Huang, Samuel R. Bowman, He He 0001, Shi Feng 0005
ICLR6
2025 MAPS: Advancing Multi-Modal Reasoning in Expert-Level Physical Science
abstract
Pre-trained on extensive text and image corpora, current Multi-Modal Large Language Models (MLLM) have shown strong capabilities in general visual reasoning tasks. However, their performance is still lacking in physical domains that require understanding diagrams with complex physical structures and quantitative analysis based on multi-modal information. To address this, we develop a new framework, named **M**ulti-Modal Scientific Re**A**soning with **P**hysics Perception and **S**imulation (**MAPS**) based on an MLLM. MAPS decomposes expert-level multi-modal reasoning task into physical diagram understanding via a Physical Perception Model (PPM) and reasoning with physical knowledge via a simulator. The PPM module is obtained by fine-tuning a visual language model using carefully designed synthetic data with paired physical diagrams and corresponding simulation language descriptions. At the inference stage, MAPS integrates the simulation language description of the input diagram provided by PPM and results obtained through a Chain-of-Simulation process with MLLM to derive the underlying rationale and the final answer. Validated using our collected college-level circuit analysis problems, MAPS significantly improves reasoning accuracy of MLLM and outperforms all existing models. The results confirm MAPS offers a promising direction for enhancing multi-modal scientific reasoning ability of MLLMs. We will release our code, model and dataset used for our experiments upon publishing of this paper.
Erle Zhu, Yadi Liu, Xujun Li, Xinjie Yu, Minlie Huang, Hongning Wang
ICLR7
2025 JPS: Jailbreak Multimodal Large Language Models with Collaborative Visual Perturbation and Textual Steering
abstract
Jailbreak attacks against multimodal large language Models (MLLMs) are a significant research focus. Current research predominantly focuses on maximizing attack success rate (ASR), often overlooking whether the generated responses actually fulfill the attacker's malicious intent. This oversight frequently leads to low-quality outputs that, while successful in bypassing safety filters, lack substantial harmful content. To address this gap, we propose JPS, Jailbreak MLLMs with collaborative visual Perturbation and textual Steering, which achieves jailbreaks via corporation of visual image and textually steering prompt. Specifically, JPS utilizes target-guided adversarial image perturbations for effective safety bypass, complemented by ''steering prompt'' optimized via a multi-agent system to specifically guide LLM responses fulfilling the attackers' intent. These visual and textual components undergo iterative co-optimization for enhanced performance. To evaluate the quality of attack outcomes, we propose the Malicious Intent Fulfillment Rate (MIFR) metric, assessed using a Reasoning-LLM-based evaluator. Our experiments show JPS sets a new state-of-the-art in both ASR and MIFR across various MLLMs and benchmarks, with analyses confirming its efficacy. Codes are available at https://github.com/thu-coai/JPS Warning: This paper contains potentially sensitive contents.
Renmiao Chen, Shiyao Cui, Xuancheng Huang, Chengwei Pan, Victor Shea-Jay Huang, Xuan Ouyang, Zhexin Zhang, Hongning Wang, Minlie Huang
ACM Multimedia10
2025 ShieldVLM: Safeguarding the Multimodal Implicit Toxicity via Deliberative Reasoning with LVLMs: ShieldVLM
abstract
Toxicity detection in multimodal text-image content faces growing challenges, especially with multimodal implicit toxicity, where each modality appears benign on its own but conveys hazard when combined. Multimodal implicit toxicity appears not only as formal statements in social platforms but also prompts that can lead to toxic dialogs from Large Vision-Language Models (LVLMs). Despite the success in unimodal text or image moderation, toxicity detection for multimodal content, particularly the multimodal implicit toxicity, remains underexplored. To fill this gap, we comprehensively build a taxonomy for multimodal implicit toxicity (MMIT) and introduce an MMIT-dataset, comprising 2,100 multimodal statements and prompts across 7 risk categories (31 sub-categories) and 5 typical cross-modal correlation modes. To advance the detection of multimodal implicit toxicity, we build ShieldVLM, a model which identifies implicit toxicity in multimodal statements, prompts and dialogs via deliberative cross-modal reasoning. Experiments show that ShieldVLM outperforms existing strong baselines in detecting both implicit and explicit toxicity. The model and dataset will be publicly available to support future researches (Warning: This paper contains potentially sensitive contents). Warning: This paper contains potentially sensitive contents.
Shiyao Cui, Xuan Ouyang, Renmiao Chen, Zhexin Zhang, Yida Lu, Hongning Wang, Han Qiu 0001, Minlie Huang
ACM Multimedia9
2025 HCDS: Hierarchical Clustering for Cold-Start Few-Shot Data Selection
abstract
Deep learning models usually require large labeled datasets to generalize well, but this is computationally and financially costly. Cold-start few-shot data selection enables fast model generalization by selecting a few diverse, representative samples from an unlabeled data pool. To achieve this goal, previous work usually divides the training data into several clusters and performs sampling from these clusters. Yet, such a way tends to have two issues. First, imbalanced data distribution in the training data pool still exists in the selected subset, causing models' performance biases and suboptimal generalization ability. Second, these methods improve sample diversity in each cluster by considering either the feature dissimilarity among instances, or model uncertainty for individual instance. They ignore the entire representativeness of samples within a cluster. To tackle these challenges, we propose a novel framework HCDS : Hierarchical Clustering for Cold-Start Few-Shot Data Selection. Specifically, we first perform class-level clustering, using pseudo-labels for class supervision and applying contrastive clustering to derive class-rich features. We then refine these features within the class-level clusters into semantically meaningful features and perform representation-level clustering. Finally, we sample data from the representation-level clusters based on global similarity to ensure representativeness. Experimental results on six public datasets, including both balanced and imbalanced ones, show that HCDS achieves state-of-the-art performance, particularly with limited and imbalanced data.
Yuhua Zhao 0001, Zhixin Han, Xunzhi Wang, Bitong Luo, Hang Gao 0003, Minlie Huang, Mengting Hu 0002
SIGIR6
2025 Scenario-independent Uncertainty Estimation for LLM-based Question Answering via Factor Analysis
abstract
Large language models (LLMs) demonstrate significant potential in various applications; however, they are susceptible to generating hallucinations, which can lead to the spread of online misinformation. Existing studies address hallucination detection by (1) employing reference-based methods that consult external resources for verification or (2) utilizing reference-free methods that mainly estimate answer uncertainty based on LLM's internal states. However, reference-based methods incur significant costs and can be infeasible for obtaining reliable external references. Besides, existing uncertainty estimation (UE) methods often overlook the impact of scenario backgrounds inherited from the query's lexical resources, leading to noise in UE. In almost all real-world applications, users care about the uncertainty concerning semantics or facts instead of the query's scenario information. Therefore, we argue that mitigating scenario-related noise and focusing on semantic information can yield a more desirable UE. In this paper, we introduce a plug-and-play scenario-independent framework to enhance unsupervised UE in LLMs by removing scenario-related noise and focusing on semantic information. This framework is compatible with most existing UE methods, as it leverages only the existing UE methods' outputs. Specifically, we design a scenario-specific sampling to paraphrase queries, maintaining their common semantics while diversifying the scenario distribution. Subsequently, to estimate the contribution of the common semantics, we design a factor analysis (FA) model to disentangle the UE score obtained from the given UE method into a combination of multiple latent factors, which represent the contribution of the common semantics and scenario-related noise. By solving the FA model, we decompose the impact of the most significant factor to approximate the uncertainty caused by the common semantics, thus achieving scenario-independent UE. Extensive experiments and analysis across multiple models and datasets demonstrate the effectiveness of our approach.
Zhihua Wen, Zhizhao Liu, Zhiliang Tian, Shilong Pan, Zhen Huang 0006, Dongsheng Li 0001, Minlie Huang
WWW7
2025 The superalignment of superhuman intelligence with large language models
Minlie Huang, Yingkang Wang, Shiyao Cui, Pei Ke, Jie Tang 0001
Sci. China Inf. Sci.1
2024 ToMBench: Benchmarking Theory of Mind in Large Language Models
abstract
Zhuang Chen, Jincenzi Wu, Jinfeng Zhou, Bosi Wen, Guanqun Bi, Gongyao Jiang, Yaru Cao, Mengting Hu, Yunghwei Lai, Zexuan Xiong, Minlie Huang. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.
Zhuang Chen 0002, Jincenzi Wu, Jinfeng Zhou, Bosi Wen, Guanqun Bi, Gongyao Jiang, Yaru Cao, Mengting Hu 0002, Yunghwei Lai, Zexuan Xiong, Minlie Huang
ACL (1)11
2024 Black-Box Prompt Optimization: Aligning Large Language Models without Model Training
abstract
Jiale Cheng, Xiao Liu, Kehan Zheng, Pei Ke, Hongning Wang, Yuxiao Dong, Jie Tang, Minlie Huang. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.
Xiao Liu 0036, Kehan Zheng, Pei Ke, Hongning Wang, Yuxiao Dong, Jie Tang 0001, Minlie Huang
ACL (1)8
2024 CritiqueLLM: Towards an Informative Critique Generation Model for Evaluation of Large Language Model Generation
abstract
Pei Ke, Bosi Wen, Andrew Feng, Xiao Liu, Xuanyu Lei, Jiale Cheng, Shengyuan Wang, Aohan Zeng, Yuxiao Dong, Hongning Wang, Jie Tang, Minlie Huang. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.
Pei Ke, Bosi Wen, Andrew Feng, Xiao Liu 0036, Xuanyu Lei, Shengyuan Wang 0002, Aohan Zeng, Yuxiao Dong, Hongning Wang, Jie Tang 0001, Minlie Huang
ACL (1)12
2024 AlignBench: Benchmarking Chinese Alignment of Large Language Models
abstract
Xiao Liu, Xuanyu Lei, Shengyuan Wang, Yue Huang, Andrew Feng, Bosi Wen, Jiale Cheng, Pei Ke, Yifan Xu, Weng Lam Tam, Xiaohan Zhang, Lichao Sun, Xiaotao Gu, Hongning Wang, Jing Zhang, Minlie Huang, Yuxiao Dong, Jie Tang. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.
Xiao Liu 0036, Xuanyu Lei, Shengyuan Wang 0002, Yue Huang 0001, Andrew Feng, Bosi Wen, Pei Ke, Yifan Xu 0014, Weng Lam Tam, Lichao Sun 0001, Xiaotao Gu, Hongning Wang, Jing Zhang 0001, Minlie Huang, Yuxiao Dong, Jie Tang 0001
ACL (1)16
2024 EmoBench: Evaluating the Emotional Intelligence of Large Language Models
abstract
Sahand Sabour, Siyang Liu, Zheyuan Zhang, June Liu, Jinfeng Zhou, Alvionna Sunaryo, Tatia Lee, Rada Mihalcea, Minlie Huang. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.
Sahand Sabour, Siyang Liu 0003, Zheyuan Zhang 0002, June M. Liu, Jinfeng Zhou, Alvionna S. Sunaryo, Tatia M. C. Lee, Rada Mihalcea, Minlie Huang
ACL (1)9
2024 Learning Task Decomposition to Assist Humans in Competitive Programming
abstract
When using language models (LMs) to solve complex problems, humans might struggle to understand the LM-generated solutions and repair the flawed ones.To assist humans in repairing them, we propose to automatically decompose complex solutions into multiple simpler pieces that correspond to specific subtasks.We introduce a novel objective for learning task decomposition, termed assistive value (AssistV), which measures the feasibility and speed for humans to repair the decomposed solution.We collect a dataset of human repair experiences on different decomposed solutions.Utilizing the collected data as in-context examples, we then learn to critique, refine, and rank decomposed solutions to improve AssistV.We validate our method under competitive programming problems: under 177 hours of human study, our method enables non-experts to solve 33.3% more problems, speeds them up by 3.3x, and empowers them to match unassisted experts.
Jiaxin Wen, Ruiqi Zhong, Pei Ke, Zhihong Shao, Hongning Wang, Minlie Huang
ACL (1)6
2024 COKE: A Cognitive Knowledge Graph for Machine Theory of Mind
abstract
Theory of mind (ToM) refers to humans' ability to understand and infer the desires, beliefs, and intentions of others.The acquisition of ToM plays a key role in humans' social cognition and interpersonal relations.Though indispensable for social intelligence, ToM is still lacking for modern AI and NLP systems since they cannot access the human mental state and cognitive process beneath the training corpus.To empower AI systems with the ToM ability and narrow the gap between them and humans, in this paper, we propose COKE: the first cognitive knowledge graph for machine theory of mind, formalizing cognitive processes as a chained structure.Specifically, COKE formalizes ToM as a collection of 45k+ manually verified cognitive chains that characterize human mental activities and subsequent behavioral/affective responses when facing specific social circumstances.In addition, we further generalize COKE using LLMs and build a powerful generation model COLM tailored for cognitive reasoning.Experimental results in both automatic and human evaluation demonstrate the high quality of COKE, the superior ToM ability of COLM, and its potential to significantly enhance social applications.We release our code and data at https://github.com/jincenziwu/COKE.
Jincenzi Wu, Zhuang Chen 0002, Jiawen Deng 0006, Sahand Sabour, Helen M. Meng, Minlie Huang
ACL (1)6
2024 SafetyBench: Evaluating the Safety of Large Language Models
abstract
Zhexin Zhang, Leqi Lei, Lindong Wu, Rui Sun, Yongkang Huang, Chong Long, Xiao Liu, Xuanyu Lei, Jie Tang, Minlie Huang. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.
Zhexin Zhang, Leqi Lei, Lindong Wu, Yongkang Huang, Chong Long, Xiao Liu 0036, Xuanyu Lei, Jie Tang 0001, Minlie Huang
ACL (1)10
2024 Defending Large Language Models Against Jailbreaking Attacks Through Goal Prioritization
abstract
While significant attention has been dedicated to exploiting weaknesses in LLMs through jailbreaking attacks, there remains a paucity of effort in defending against these attacks.We point out a pivotal factor contributing to the success of jailbreaks: the intrinsic conflict between the goals of being helpful and ensuring safety.Accordingly, we propose to integrate goal prioritization at both training and inference stages to counteract.Implementing goal prioritization during inference substantially diminishes the Attack Success Rate (ASR) of jailbreaking from 66.4% to 3.6% for ChatGPT.And integrating goal prioritization into model training reduces the ASR from 71.0% to 6.6% for Llama2-13B.Remarkably, even in scenarios where no jailbreaking samples are included during training, our approach slashes the ASR by half.Additionally, our findings reveal that while stronger LLMs face greater safety risks, they also possess a greater capacity to be steered towards defending against such attacks, both because of their stronger ability in instruction following.Our work thus contributes to the comprehension of jailbreaking attacks and defenses, and sheds light on the relationship between LLMs' capability and safety.Our code is available at https://github.com/thu-coai/ JailbreakDefense_GoalPriority.
Zhexin Zhang, Junxiao Yang, Pei Ke, Fei Mi, Hongning Wang, Minlie Huang
ACL (1)6
2024 Intent-Aware and Hate-Mitigating Counterspeech Generation via Dual-Discriminator Guided LLMs
abstract
Counterspeech is an effective way to combat online hate speech. Considering the multifaceted nature of online hate speech, counterspeech with varying intents (e.g., denouncing or empathy) has significant potential to mitigate hate speech effectively. Recently, controlled approaches based on large language models (LLMs) have been explored to generate intent-specific counterspeech. Due to the lack of attention to intent-specific information by LLMs during the decoding process, those methods cater more to the semantic information rather than matching with the desired intents. Further, there are still limitations in quantitatively evaluating the effectiveness of counterspeech with different intents in mitigating hate speech. In this paper, to address the above issues, we propose DART, an LLMs-based DuAl-discRiminaTor guided framework for counterspeech generation. We employ an intent-aware discriminator and hate-mitigating discriminator to jointly guide the decoding preferences of LLMs, which facilitates the model towards generating counterspeech catering to specific intent and hate mitigation. We apply a maximum-margin relative objective for training discriminators. This objective leverages the distance between counterspeech aligned with the desired target (such as specific intent or effectiveness in hate mitigation) and undesired as an effective learning signal. Extensive experiments show that DART achieves excellent performances in matching the desired intent and mitigating hate.
Zhiliang Tian, Yue Zhang 0049, Hongkui Tu, Minlie Huang, Bin Zhou 0004
LREC/COLING7
2024 Instruction Pre-Training: Language Models are Supervised Multitask Learners
abstract
Unsupervised multitask pre-training has been the critical method behind the recent success of language models (LMs).However, supervised multitask learning still holds significant promise, as scaling it in the post-training stage trends towards better generalization.In this paper, we explore supervised multitask pretraining by proposing Instruction Pre-Training, a framework that scalably augments massive raw corpora with instruction-response pairs to pre-train LMs.The instruction-response pairs are generated by an efficient instruction synthesizer built on open-source models.In our experiments, we synthesize 200M instruction-response pairs covering 40+ task categories to verify the effectiveness of Instruction Pre-Training.In pre-training from scratch, Instruction Pre-Training not only consistently enhances pre-trained base models but also benefits more from further instruction tuning.In continual pre-training, Instruction Pre-Training enables Llama3-8B to be comparable to or even outperform Llama3-70B.Our model, code, and data are available at https://github.com/microsoft/LMOps.Ins: When is the finale of season 7? Let's think step by step.
Daixuan Cheng, Yuxian Gu, Shaohan Huang, Junyu Bi, Minlie Huang, Furu Wei
EMNLP5
2024 Mixture-of-Modules: Reinventing Transformers as Dynamic Assemblies of Modules
abstract
Is it always necessary to compute tokens from shallow to deep layers in Transformers?The continued success of vanilla Transformers and their variants suggests an undoubted "yes".In this work, however, we attempt to break the depth-ordered convention by proposing a novel architecture dubbed mixture-of-modules (MoM), which is motivated by an intuition that any layer, regardless of its position, can be used to compute a token as long as it possesses the needed processing capabilities.The construction of MoM starts from a finite set of modules defined by multi-head attention and feed-forward networks, each distinguished by its unique parameterization.Two routers then iteratively select attention modules and feedforward modules from the set to process a token.The selection dynamically expands the computation graph in the forward pass of the token, culminating in an assembly of modules.We show that MoM provides not only a unified framework for Transformers and their numerous variants but also a flexible and learnable approach for reducing redundancy in Transformer parameterization.We pre-train various MoMs using OpenWebText.Empirical results demonstrate that MoMs, of different parameter counts, consistently outperform vanilla transformers on both GLUE and XSUM benchmarks.More interestingly, with a fixed parameter budget, MoM-large enables an over 38% increase in depth for computation graphs compared to GPT-2-large, resulting in absolute gains of 1.4 on GLUE and 1 on XSUM.On the other hand, MoM-large also enables an over 60% reduction in depth while involving more modules per layer, yielding a 16% reduction in TFLOPs and a 43% decrease in memory usage compared to GPT-2-large, while maintaining comparable performance.1 * Equal Contributions.† Corresponding authors. 1 Code is available at https://github.com/gzhch/Mixture-of- Modules
Zhuocheng Gong, Ang Lv, Jian Guan 0002, Wei Wu 0014, Huishuai Zhang, Minlie Huang, Dongyan Zhao 0001, Rui Yan 0001
EMNLP6
2024 ASETF: A Novel Method for Jailbreak Attack on LLMs through Translate Suffix Embeddings
abstract
The safety defense methods of Large language models (LLMs) stays limited because the dangerous prompts are manually curated to just few known attack types, which fails to keep pace with emerging varieties.Recent studies found that attaching suffixes to harmful instructions can hack the defense of LLMs and lead to dangerous outputs.However, similar to traditional text adversarial attacks, this approach, while effective, is limited by the challenge of the discrete tokens.This gradient based discrete optimization attack requires over 100,000 LLM calls, and due to the unreadable of adversarial suffixes, it can be relatively easily penetrated by common defense methods such as perplexity filters.To cope with this challenge, in this paper, we propose an Adversarial Suffix Embedding Translation Framework (ASETF), aimed at transforming continuous adversarial suffix embeddings into coherent and understandable text.This method greatly reduces the computational overhead during the attack process and helps to automatically generate multiple adversarial samples, which can be used as data to strengthen LLM's security defense.Experimental evaluations were conducted on Llama2, Vicuna, and other prominent LLMs, employing harmful directives sourced from the Advbench dataset.The results indicate that our method significantly reduces the computation time of adversarial suffixes and achieves a much better attack success rate than existing techniques, while significantly enhancing the textual fluency of the prompts.In addition, our approach can be generalized into a broader method for generating transferable adversarial suffixes that can successfully attack multiple LLMs, even black-box LLMs, such as ChatGPT and Gemini.
Hao Wang 0003, Hao Li 0031, Minlie Huang, Lei Sha
EMNLP3
2024 DC-Instruct: An Effective Framework for Generative Multi-intent Spoken Language Understanding
abstract
In the realm of multi-intent spoken language understanding, recent advancements have leveraged the potential of prompt learning frameworks.However, critical gaps exist in these frameworks: the lack of explicit modeling of dual-task dependencies and the oversight of task-specific semantic differences among utterances.To address these shortcomings, we propose DC-Instruct, a novel generative framework based on Dual-task Inter-dependent Instructions (DII) and Supervised Contrastive Instructions (SCI).Specifically, DII guides large language models (LLMs) to generate labels for one task based on the other task's labels, thereby explicitly capturing dual-task interdependencies.Moreover, SCI leverages utterance semantics differences by guiding LLMs to determine whether a pair of utterances share the same or similar labels.This can improve LLMs on extracting and discriminating task-specific semantics, thus enhancing their SLU reasoning abilities.Extensive experiments on public benchmark datasets show that DC-Instruct markedly outperforms current generative models and state-of-the-art methods, demonstrating its effectiveness in enhancing dialogue language understanding and reasoning.
Lizi Liao, Minlie Huang, Ivor W. Tsang
EMNLP3
2024 Thoughts to Target: Enhance Planning for Target-driven Conversation
abstract
In conversational AI, large-scale models excel in various tasks but struggle with target-driven conversation planning.Current methods, such as chain-of-thought reasoning and tree-search policy learning techniques, either neglect plan rationality or require extensive human simulation procedures.Addressing this, we propose a novel two-stage framework, named EnPL, to improve the LLMs' capability in planning conversations towards designated targets, including (1) distilling natural language plans from target-driven conversation corpus and (2) generating new plans with demonstration-guided in-context learning.Specifically, we first propose a filter approach to distill a high-quality plan dataset, ConvPlan 1 .With the aid of corresponding conversational data and support from relevant knowledge bases, we validate the quality and rationality of these plans.Then, these plans are leveraged to help guide LLMs to further plan for new targets.Empirical results demonstrate that our method significantly improves the planning ability of LLMs, especially in target-driven conversations.Furthermore, EnPL is demonstrated to be quite effective in collecting target-driven conversation datasets and enhancing response generation, paving the way for constructing extensive target-driven conversational models.
Zhonghua Zheng, Lizi Liao, Yang Deng 0002, Ee-Peng Lim, Minlie Huang, Liqiang Nie
EMNLP5
2024 AgentBench: Evaluating LLMs as Agents
abstract
The potential of Large Language Model (LLM) as agents has been widely acknowledged recently. Thus, there is an urgent need to quantitatively evaluate LLMs as agents on challenging tasks in interactive environments. We present AgentBench, a multi-dimensional benchmark that consists of 8 distinct environments to assess LLM-as-Agent's reasoning and decision-making abilities. Our extensive test over 29 API-based and open-sourced (OSS) LLMs shows that, while top commercial LLMs present a strong ability of acting as agents in complex environments, there is a significant disparity in performance between them and many OSS competitors that are no larger than 70B. We identify the typical reasons of failures in environments and LLMs, showing that poor long-term reasoning, decision-making, and instruction following abilities are the main obstacles for developing usable LLM agents. Improving instruction following and training on high quality multi-round alignment data could improve agent performance. And different from existing assumptions, training on code present ambivalent impacts on different agent tasks. Datasets, environments, and an integrated evaluation package for AgentBench are released at https://github.com/THUDM/AgentBench.
Xiao Liu 0036, Hao Yu 0030, Hanchen Zhang, Yifan Xu 0014, Xuanyu Lei, Hanyu Lai, Yu Gu 0016, Hangliang Ding, Kaiwen Men, Kejuan Yang, Shudan Zhang, Xiang Deng 0001, Aohan Zeng, Zhengxiao Du, Sheng Shen 0001, Tianjun Zhang, Yu Su 0001, Huan Sun 0001, Minlie Huang, Yuxiao Dong, Jie Tang 0001
ICLR20
2024 ToRA: A Tool-Integrated Reasoning Agent for Mathematical Problem Solving
abstract
Large language models have made significant progress in various language tasks, yet they still struggle with complex mathematics. In this paper, we propose ToRA a series of Tool-integrated Reasoning Agents designed to solve challenging mathematical problems by seamlessly integrating natural language reasoning with the utilization of external tools (e.g., computation libraries and symbolic solvers), thereby amalgamating the analytical prowess of language and the computational efficiency of tools. To train ToRA, we curate interactive tool-use trajectories on mathematical datasets, apply imitation learning on the annotations, and propose output space shaping to further refine models' reasoning behavior. As a result, ToRA models significantly outperform open-source models on 10 mathematical reasoning datasets across all scales with 13%-19% absolute improvements on average. Notably, ToRA-7B reaches 44.6% on the competition-level dataset MATH, surpassing the best open-source model WizardMath-70B by 22% absolute. ToRA-34B is also the first open-source model that achieves an accuracy exceeding 50% on MATH, which significantly outperforms GPT-4's CoT result, and is competitive with GPT-4 solving problems with programs. Additionally, we conduct a comprehensive analysis of the benefits and remaining challenges of tool interaction for mathematical reasoning, providing valuable insights for future research.
Zhibin Gou, Zhihong Shao, Yeyun Gong, Yelong Shen, Yujiu Yang 0001, Minlie Huang, Nan Duan 0001, Weizhu Chen
ICLR6
2024 MiniLLM: Knowledge Distillation of Large Language Models
abstract
Knowledge Distillation (KD) is a promising technique for reducing the high computational demand of large language models (LLMs). However, previous KD methods are primarily applied to white-box classification models or training small models to imitate black-box model APIs like ChatGPT. How to effectively distill the knowledge of white-box LLMs into small models is still under-explored, which becomes more important with the prosperity of open-source LLMs. In this work, we propose a KD approach that distills LLMs into smaller language models. We first replace the forward Kullback-Leibler divergence (KLD) objective in the standard KD approaches with reverse KLD, which is more suitable for KD on generative language models, to prevent the student model from overestimating the low-probability regions of the teacher distribution. Then, we derive an effective optimization approach to learn this objective. The student models are named MiniLLM. Extensive experiments in the instruction-following setting show that MiniLLM generates more precise responses with higher overall quality, lower exposure bias, better calibration, and higher long-text generation performance than the baselines. Our method is scalable for different model families with 120M to 13B parameters. Our code, data, and model checkpoints can be found in https://github.com/microsoft/LMOps/tree/main/minillm.
Yuxian Gu, Li Dong 0004, Furu Wei, Minlie Huang
ICLR4
2024 Language Model Decoding as Direct Metrics Optimization
abstract
Despite the remarkable advances in language modeling, current mainstream decoding methods still struggle to generate texts that align with human texts across different aspects. In particular, sampling-based methods produce less-repetitive texts which are often disjunctive in discourse, while search-based methods maintain topic coherence at the cost of increased repetition. Overall, these methods fall short in achieving holistic alignment across a broad range of aspects. In this work, we frame decoding from a language model as an optimization problem with the goal of strictly matching the expected performance with human texts measured by multiple metrics of desired aspects simultaneously. The resulting decoding distribution enjoys an analytical solution that scales the input language model distribution via a sequence-level energy function defined by these metrics. And most importantly, we prove that this induced distribution is guaranteed to improve the perplexity on human texts, which suggests a better approximation to the underlying distribution of human texts. To facilitate tractable sampling from this globally normalized distribution, we adopt the Sampling-Importance-Resampling technique. Experiments on various domains and model scales demonstrate the superiority of our method in metrics alignment with human texts and human evaluation over strong baselines.
Haozhe Ji, Pei Ke, Hongning Wang, Minlie Huang
ICLR4
2024 Large Language Models Are Not Robust Multiple Choice Selectors
abstract
Multiple choice questions (MCQs) serve as a common yet important task format in the evaluation of large language models (LLMs). This work shows that modern LLMs are vulnerable to option position changes in MCQs due to their inherent “selection bias”, namely, they prefer to select specific option IDs as answers (like “Option A”). Through extensive empirical analyses with 20 LLMs on three benchmarks, we pinpoint that this behavioral bias primarily stems from LLMs’ token bias, where the model a priori assigns more probabilistic mass to specific option ID tokens (e.g., A/B/C/D) when predicting answers from the option IDs. To mitigate selection bias, we propose a label-free, inference-time debiasing method, called PriDe, which separates the model’s prior bias for option IDs from the overall prediction distribution. PriDe first estimates the prior by permutating option contents on a small number of test samples, and then applies the estimated prior to debias the remaining samples. We demonstrate that it achieves interpretable and transferable debiasing with high computational efficiency. We hope this work can draw broader research attention to the bias and robustness of modern LLMs.
Chujie Zheng, Hao Zhou 0012, Fandong Meng, Jie Zhou 0016, Minlie Huang
ICLR5
2024 Towards Efficient Exact Optimization of Language Model Alignment
abstract
The alignment of language models with human preferences is vital for their application in real-world tasks. The problem is formulated as optimizing the model's policy to maximize the expected reward that reflects human preferences with minimal deviation from the initial policy. While considered as a straightforward solution, reinforcement learning (RL) suffers from high variance in policy updates, which impedes efficient policy improvement. Recently, direct preference optimization (DPO) was proposed to directly optimize the policy from preference data. However, we show that DPO derived based on the optimal solution of the problem leads to a compromised mean-seeking approximation of the optimal solution in practice. In this paper, we propose efficient exact optimization (EXO) of the alignment objective. EXO is guaranteed to optimize in the same direction as RL algorithms asymptotically for arbitrary policy parametrization. This leads to the same mode-seeking solution, while enables efficient optimization by circumventing the complexities of RL. We also compare our method to DPO with both theoretical and empirical analyses, and further demonstrate the advantages of our method over existing approaches on realistic human preference data. Code is available at https://github.com/haozheji/exact-optimization.
Haozhe Ji, Cheng Lu 0011, Yilin Niu, Pei Ke, Hongning Wang, Jun Zhu 0001, Jie Tang 0001, Minlie Huang
ICML8
2024 On Prompt-Driven Safeguarding for Large Language Models
abstract
Prepending model inputs with safety prompts is a common practice for safeguarding large language models (LLMs) against queries with harmful intents. However, the underlying working mechanisms of safety prompts have not been unraveled yet, restricting the possibility of automatically optimizing them to improve LLM safety. In this work, we investigate how LLMs’ behavior (i.e., complying with or refusing user queries) is affected by safety prompts from the perspective of model representation. We find that in the representation space, the input queries are typically moved by safety prompts in a "higher-refusal" direction, in which models become more prone to refusing to provide assistance, even when the queries are harmless. On the other hand, LLMs are naturally capable of distinguishing harmful and harmless queries without safety prompts. Inspired by these findings, we propose a method for safety prompt optimization, namely DRO (Directed Representation Optimization). Treating a safety prompt as continuous, trainable embeddings, DRO learns to move the queries’ representations along or opposite the refusal direction, depending on their harmfulness. Experiments with eight LLMs on out-of-domain and jailbreak benchmarks demonstrate that DRO remarkably improves the safeguarding performance of human-crafted safety prompts, without compromising the models’ general performance.
Chujie Zheng, Fan Yin, Hao Zhou 0012, Fandong Meng, Jie Zhou 0016, Kai-Wei Chang 0001, Minlie Huang, Nanyun Peng 0001
ICML7
2024 Language Models Hallucinate, but May Excel at Fact Verification
abstract
Jian Guan, Jesse Dodge, David Wadden, Minlie Huang, Hao Peng. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.
Jian Guan 0002, Jesse Dodge, Dave Wadden, Minlie Huang, Hao Peng 0009
NAACL-HLT4
2024 Depression Detection in Clinical Interviews with LLM-Empowered Structural Element Graph
abstract
Zhuang Chen, Jiawen Deng, Jinfeng Zhou, Jincenzi Wu, Tieyun Qian, Minlie Huang. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.
Zhuang Chen 0002, Jiawen Deng 0006, Jinfeng Zhou, Jincenzi Wu, Tieyun Qian, Minlie Huang
NAACL-HLT6
2024 AMOR: A Recipe for Building Adaptable Modular Knowledge Agents Through Process Feedback
abstract
The notable success of large language models (LLMs) has sparked an upsurge in building language agents to complete various complex tasks. We present AMOR, an agent framework based on open-source LLMs, which reasons with external knowledge bases and adapts to specific domains through human supervision to the reasoning process. AMOR builds reasoning logic over a finite state machine (FSM) that solves problems through autonomous executions and transitions over disentangled modules. This allows humans to provide direct feedback to the individual modules, and thus naturally forms process supervision. Based on this reasoning and feedback framework, we develop AMOR through two-stage fine-tuning: warm-up and adaptation. The former fine-tunes the LLM with examples automatically constructed from various public datasets, enabling AMOR to generalize across different knowledge environments, while the latter tailors AMOR to specific domains using process feedback. Extensive experiments across multiple domains demonstrate the advantage of AMOR to strong baselines, thanks to its FSM-based reasoning and process feedback mechanism. The code and data are publicly available at https://github.com/JianGuanTHU/AMOR.
Jian Guan 0002, Wei Wu 0014, Zujie Wen, Peng Xu 0002, Hongning Wang, Minlie Huang
NeurIPS6
2024 Benchmarking Complex Instruction-Following with Multiple Constraints Composition
abstract
Instruction following is one of the fundamental capabilities of large language models (LLMs). As the ability of LLMs is constantly improving, they have been increasingly applied to deal with complex human instructions in real-world scenarios. Therefore, how to evaluate the ability of complex instruction-following of LLMs has become a critical research problem. Existing benchmarks mainly focus on modeling different types of constraints in human instructions while neglecting the composition of different constraints, which is an indispensable constituent in complex instructions. To this end, we propose ComplexBench, a benchmark for comprehensively evaluating the ability of LLMs to follow complex instructions composed of multiple constraints. We propose a hierarchical taxonomy for complex instructions, including 4 constraint types, 19 constraint dimensions, and 4 composition types, and manually collect a high-quality dataset accordingly. To make the evaluation reliable, we augment LLM-based evaluators with rules to effectively verify whether generated texts can satisfy each constraint and composition. Furthermore, we obtain the final evaluation score based on the dependency structure determined by different composition types. ComplexBench identifies significant deficiencies in existing LLMs when dealing with complex instructions with multiple constraints composition.
Bosi Wen, Pei Ke, Xiaotao Gu, Lindong Wu, Jinfeng Zhou, Wenchuang Li, Binxin Hu, Wendy Gao, Jiaxing Xu, Jie Tang 0001, Hongning Wang, Minlie Huang
NeurIPS14
2024 Perception of Knowledge Boundary for Large Language Models through Semi-open-ended Question Answering
abstract
Large Language Models (LLMs) are widely used for knowledge-seeking purposes yet suffer from hallucinations. The knowledge boundary of an LLM limits its factual understanding, beyond which it may begin to hallucinate. Investigating the perception of LLMs' knowledge boundary is crucial for detecting hallucinations and LLMs' reliable generation. Current studies perceive LLMs' knowledge boundary on questions with concrete answers (close-ended questions) while paying limited attention to semi-open-ended questions that correspond to many potential answers. Some researchers achieve it by judging whether the question is answerable or not. However, this paradigm is not so suitable for semi-open-ended questions, which are usually ``partially answerable questions'' containing both answerable answers and ambiguous (unanswerable) answers. Ambiguous answers are essential for knowledge-seeking, but it may go beyond the knowledge boundary of LLMs. In this paper, we perceive the LLMs' knowledge boundary with semi-open-ended questions by discovering more ambiguous answers. First, we apply an LLM-based approach to construct semi-open-ended questions and obtain answers from a target LLM. Unfortunately, the output probabilities of mainstream black-box LLMs are inaccessible to sample more low-probability ambiguous answers. Therefore, we apply an open-sourced auxiliary model to explore ambiguous answers for the target LLM. We calculate the nearest semantic representation for existing answers to estimate their probabilities, with which we reduce the generation probability of high-probability existing answers to achieve a more effective generation. Finally, we compare the results from the RAG-based evaluation and LLM self-evaluation to categorize four types of ambiguous answers that are beyond the knowledge boundary of the target LLM. Following our method, we construct a dataset to perceive the knowledge boundary for GPT-4. We find that GPT-4 performs poorly on semi-open-ended questions and is often unaware of its knowledge boundary. Besides, our auxiliary model, LLaMA-2-13B, is effective in discovering many ambiguous answers, including correct answers neglected by GPT-4 and delusive wrong answers GPT-4 struggles to identify.
Zhihua Wen, Zhiliang Tian, Zexin Jian, Zhen Huang 0006, Pei Ke, Yifu Gao, Minlie Huang, Dongsheng Li 0001
NeurIPS7
2024 Enhancing Emotional Support Conversation with Cognitive Chain-of-Thought Reasoning
Yaru Cao, Zhuang Chen 0002, Guanqun Bi, Yulin Feng, Fucheng Wan, Minlie Huang, Hongzhi Yu
NLPCC (1)7
2024 ChatGPT: potential, prospects, and limitations
Jie Zhou 0015, Pei Ke, Xipeng Qiu, Minlie Huang, Junping Zhang
Frontiers Inf. Technol. Electron. Eng.4
2024 Overview of the Ninth Dialog System Technology Challenge: DSTC9
abstract
This paper introduces the Ninth Dialog System Technology Challenge (DSTC-9). This edition of the DSTC focuses on applying end-to-end dialog technologies for four distinct tasks in dialog systems, namely, 1. Task-oriented dialog Modeling with Unstructured Knowledge Access, 2. Multi-domain task-oriented dialog, 3. Interactive evaluation of dialog and 4. Situated interactive multimodal dialog. This paper describes the task definition, provided datasets, baselines, and evaluation setup for each track. We also summarize the results of the submitted systems to highlight the general trends of the state-of-the-art technologies for the tasks.
R. Chulaka Gunasekara, Seokhwan Kim, Luis Fernando D'Haro, Abhinav Rastogi, Yun-Nung Chen, Mihail Eric, Behnam Hedayatnia, Karthik Gopalakrishnan 0001, Yang Liu 0004, Chao-Wei Huang, Dilek Hakkani-Tür, Jinchao Li, Qi Zhu 0007, Lingxiao Luo, Lars Liden, Kaili Huang, Shahin Shayandeh, Runze Liang, Baolin Peng, Zheng Zhang 0020, Swadheen Shukla, Minlie Huang, Jianfeng Gao 0001, Shikib Mehri, Yulan Feng, Carla Gordon, Seyed Hossein Alavi, David R. Traum, Maxine Eskénazi, Ahmad Beirami, Eunjoon Cho, Paul A. Crook, Ankita De, Alborz Geramifard, Satwik Kottur, Seungwhan Moon, Shivani Poddar, Rajen Subba
IEEE ACM Trans. Audio Speech Lang. Process.22
2023 KPT: Keyword-Guided Pre-training for Grounded Dialog Generation
abstract
Incorporating external knowledge into the response generation process is essential to building more helpful and reliable dialog agents. However, collecting knowledge-grounded conversations is often costly, calling for a better pre-trained model for grounded dialog generation that generalizes well w.r.t. different types of knowledge. In this work, we propose KPT (Keyword-guided Pre-Training), a novel self-supervised pre-training method for grounded dialog generation without relying on extra knowledge annotation. Specifically, we use a pre-trained language model to extract the most uncertain tokens in the dialog as keywords. With these keywords, we construct two kinds of knowledge and pre-train a knowledge-grounded response generation model, aiming at handling two different scenarios: (1) the knowledge should be faithfully grounded; (2) it can be selectively used. For the former, the grounding knowledge consists of keywords extracted from the response. For the latter, the grounding knowledge is additionally augmented with keywords extracted from other utterances in the same dialog. Since the knowledge is extracted from the dialog itself, KPT can be easily performed on a large volume and variety of dialogue data. We considered three data sources (open-domain, task-oriented, conversational QA) with a total of 2.5M dialogues. We conduct extensive experiments on various few-shot knowledge-grounded generation tasks, including grounding on dialog acts, knowledge graphs, persona descriptions, and Wikipedia passages. Our comprehensive experiments and analyses demonstrate that KPT consistently outperforms state-of-the-art methods on these tasks with diverse grounding knowledge.
Qi Zhu 0007, Fei Mi, Zheng Zhang 0020, Yasheng Wang, Xin Jiang 0002, Qun Liu 0001, Xiaoyan Zhu 0001, Minlie Huang
AAAI9
2023 Generating Coherent Narratives by Learning Dynamic and Discrete Entity States with a Contrastive Framework
abstract
Despite advances in generating fluent texts, existing pretraining models tend to attach incoherent event sequences to involved entities when generating narratives such as stories and news. We conjecture that such issues result from representing entities as static embeddings of superficial words, while neglecting to model their ever-changing states, i.e., the information they carry, as the text unfolds. Therefore, we extend the Transformer model to dynamically conduct entity state updates and sentence realization for narrative generation. We propose a contrastive framework to learn the state representations in a discrete space, and insert additional attention layers into the decoder to better exploit these states. Experiments on two narrative datasets show that our model can generate more coherent and diverse narratives than strong baselines with the guidance of meaningful entity states.
Jian Guan 0002, Zhipeng Hu, Minlie Huang
AAAI5
2023 MoralDial: A Framework to Train and Evaluate Moral Dialogue Systems via Moral Discussions
abstract
Hao Sun, Zhexin Zhang, Fei Mi, Yasheng Wang, Wei Liu, Jianwei Cui, Bin Wang, Qun Liu, Minlie Huang. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.
Hao Sun 0012, Zhexin Zhang, Fei Mi, Yasheng Wang, Wei Liu 0005, Jianwei Cui 0002, Bin Wang 0004, Qun Liu 0001, Minlie Huang
ACL (1)9
2023 Pre-Training to Learn in Context
abstract
In-context learning, where pre-trained language models learn to perform tasks from task examples and instructions in their contexts, has attracted much attention in the NLP community.However, the ability of in-context learning is not fully exploited because language models are not explicitly trained to learn in context.To this end, we propose PICL (Pretraining for In-Context Learning), a framework to enhance the language models' in-context learning ability by pre-training the model on a large collection of "intrinsic tasks" in the general plain-text corpus using the simple language modeling objective.PICL encourages the model to infer and perform tasks by conditioning on the contexts while maintaining task generalization of pre-trained models.We evaluate the in-context learning performance of the model trained with PICL on seven widelyused text classification datasets and the SUPER-NATURALINSTRCTIONS benchmark, which contains 100+ NLP tasks formulated to text generation.Our experiments show that PICL is more effective and task-generalizable than a range of baselines, outperforming larger language models with nearly 4x parameters.The code is publicly available at https://github. com/thu-coai/PICL.
Yuxian Gu, Li Dong 0004, Furu Wei, Minlie Huang
ACL (1)4
2023 DecompEval: Evaluating Generated Texts as Unsupervised Decomposed Question Answering
abstract
Existing evaluation metrics for natural language generation (NLG) tasks face the challenges on generalization ability and interpretability.Specifically, most of the wellperformed metrics are required to train on evaluation datasets of specific NLG tasks and evaluation dimensions, which may cause over-fitting to task-specific datasets.Furthermore, existing metrics only provide an evaluation score for each dimension without revealing the evidence to interpret how this score is obtained.To deal with these challenges, we propose a simple yet effective metric called DecompEval.This metric formulates NLG evaluation as an instruction-style question answering task and utilizes instruction-tuned pre-trained language models (PLMs) without training on evaluation datasets, aiming to enhance the generalization ability.To make the evaluation process more interpretable, we decompose our devised instruction-style question about the quality of generated texts into the subquestions that measure the quality of each sentence.The subquestions with their answers generated by PLMs are then recomposed as evidence to obtain the evaluation result.Experimental results show that DecompEval achieves state-of-the-art performance in untrained metrics for evaluating text summarization and dialogue generation, which also exhibits strong dimension-level / task-level generalization ability and interpretability 1 .
Pei Ke, Fei Huang 0005, Fei Mi, Yasheng Wang, Qun Liu 0001, Xiaoyan Zhu 0001, Minlie Huang
ACL (1)7
2023 ETHICIST: Targeted Training Data Extraction Through Loss Smoothed Soft Prompting and Calibrated Confidence Estimation
abstract
Large pre-trained language models achieve impressive results across many tasks.However, recent works point out that pre-trained language models may memorize a considerable fraction of their training data, leading to the privacy risk of information leakage.In this paper, we propose a method named ETHICIST for targeted training data Extraction THrough loss smoothed soft prompting and calIbrated ConfIdence eSTimation, investigating how to recover the suffix in the training data when given a prefix.To elicit memorization in the attacked model, we tune soft prompt embeddings while keeping the model fixed.We further propose a smoothing loss that smooths the loss distribution of the suffix tokens to make it easier to sample the correct suffix.In order to select the most probable suffix from a collection of sampled suffixes and estimate the prediction confidence, we propose a calibrated confidence estimation method, which normalizes the confidence of the generated suffixes with a local estimation.We show that ETHICIST significantly improves the extraction performance on a recently proposed public benchmark.We also investigate several factors influencing the data extraction performance, including decoding strategy, model scale, prefix length, and suffix length.Our code is available at https://github.com/ thu-coai/
Zhexin Zhang, Jiaxin Wen, Minlie Huang
ACL (1)3
2023 Facilitating Multi-turn Emotional Support Conversation with Positive Emotion Elicitation: A Reinforcement Learning Approach
abstract
Emotional support conversation (ESC) aims to provide emotional support (ES) to improve one's mental state.Existing works stay at fitting grounded responses and responding strategies (e.g., question), which ignore the effect on ES and lack explicit goals to guide emotional positive transition.To this end, we introduce a new paradigm to formalize multi-turn ESC as a process of positive emotion elicitation.Addressing this task requires finely adjusting the elicitation intensity in ES as the conversation progresses while maintaining conversational goals like coherence.In this paper, we propose SUPPORTER, a mixture-of-expert-based reinforcement learning model, and well design ES and dialogue coherence rewards to guide policy's learning for responding.Experiments verify the superiority of SUPPORTER in achieving positive emotion elicitation during responding while maintaining conversational goals including coherence.
Jinfeng Zhou, Zhuang Chen 0002, Minlie Huang
ACL (1)4
2023 CASE: Aligning Coarse-to-Fine Cognition and Affection for Empathetic Response Generation
abstract
Empathetic conversation is psychologically supposed to be the result of conscious alignment and interaction between the cognition and affection of empathy.However, existing empathetic dialogue models usually consider only the affective aspect or treat cognition and affection in isolation, which limits the capability of empathetic response generation.In this work, we propose the CASE model for empathetic dialogue generation.It first builds upon a commonsense cognition graph and an emotional concept graph and then aligns the user's cognition and affection at both the coarse-grained and fine-grained levels.Through automatic and manual evaluation, we demonstrate that CASE outperforms state-of-the-art baselines of empathetic dialogues and can generate more empathetic and informative responses.1
Jinfeng Zhou, Chujie Zheng, Bo Wang 0011, Zheng Zhang 0020, Minlie Huang
ACL (1)5
2023 StoryTrans: Non-Parallel Story Author-Style Transfer with Discourse Representations and Content Enhancing
abstract
Non-parallel text style transfer is an important task in natural language generation.However, previous studies concentrate on the token or sentence level, such as sentence sentiment and formality transfer, but neglect long style transfer at the discourse level.Long texts usually involve more complicated author linguistic preferences such as discourse structures than sentences.In this paper, we formulate the task of non-parallel story author-style transfer, which requires transferring an input story into a specified author style while maintaining source semantics.To tackle this problem, we propose a generation model, named StoryTrans, which leverages discourse representations to capture source content information and transfer them to target styles with learnable style embeddings.We use an additional training objective to disentangle stylistic features from the learned discourse representation to prevent the model from degenerating to an auto-encoder.Moreover, to enhance content preservation, we design a mask-and-fill framework to explicitly fuse style-specific keywords of source texts into generation.Furthermore, we constructed new datasets for this task in Chinese and English, respectively.Extensive experiments show that our model outperforms strong baselines in overall performance of style transfer and content preservation.
Xuekai Zhu, Jian Guan 0002, Minlie Huang
ACL (1)3
2023 Task-Adaptive Tokenization: Enhancing Long-Form Text Generation Efficacy in Mental Health and Beyond
abstract
We propose task-adaptive tokenization 1 as a way to adapt the generation pipeline to the specifics of a downstream task and enhance long-form generation in mental health.Inspired by insights from cognitive science, our task-adaptive tokenizer samples variable segmentations from multiple outcomes, with sampling probabilities optimized based on taskspecific data.We introduce a strategy for building a specialized vocabulary and introduce a vocabulary merging protocol that allows for the integration of task-specific tokens into the pre-trained model's tokenization step.Through extensive experiments on psychological question-answering tasks in both Chinese and English, we find that our task-adaptive tokenization approach brings a significant improvement in generation performance while using up to 60% fewer tokens.Preliminary experiments point to promising results when using our tokenization approach with very large language models.
Siyang Liu 0003, Naihao Deng, Sahand Sabour, Yilin Jia, Minlie Huang, Rada Mihalcea
EMNLP5
2023 Building Multi-domain Dialog State Trackers from Single-domain Dialogs
abstract
Existing multi-domain dialog state tracking (DST) models are developed based on multidomain dialogs, which require significant manual effort to define domain relations and collect data.This process can be challenging and expensive, particularly when numerous domains are involved.In this paper, we propose a divide-and-conquer (DAC) DST paradigm and a multi-domain dialog synthesis framework, which makes building multi-domain DST models from single-domain dialogs possible.The DAC paradigm segments a multi-domain dialog into multiple single-domain dialogs for DST, which makes models generalize better on dialogs involving unseen domain combinations.The multi-domain dialog synthesis framework merges several potentially related single-domain dialogs into one multi-domain dialog and modifies the dialog to simulate domain relations.The synthesized dialogs can help DST models capture the value transfer between domains.Experiments with three representative DST models on two datasets demonstrate the effectiveness of our proposed DAC paradigm and data synthesis framework.
Qi Zhu 0007, Zheng Zhang 0020, Xiaoyan Zhu 0001, Minlie Huang
EMNLP4
2023 Multi-Source Probing for Open-Domain Conversational Understanding
abstract
Dialogue comprehension and generation are vital to the success of open-domain dialogue systems.Although pre-trained generative conversation models have made significant progress in generating fluent responses, people have difficulty judging whether they understand and efficiently model the contextual information of the conversation.In this study, we propose a Multi-Source Probing (MSP) method to probe the dialogue comprehension abilities of opendomain dialogue models.MSP aggregates features from multiple sources to accomplish diverse task goals and conducts downstream tasks in a generative manner that is consistent with dialogue model pre-training to leverage model capabilities.We conduct probing experiments on seven tasks that require various dialogue comprehension skills, based on the internal representations encoded by dialogue models.Experimental results show that open-domain dialogue models can encode semantic information in the intermediate hidden states, which facilitates dialogue comprehension tasks.Models of different scales and structures possess different conversational understanding capabilities.Our findings encourage a comprehensive evaluation and design of open-domain dialogue models.
Hao Zhou 0012, Jie Zhou 0016, Minlie Huang
EMNLP4
2023 Unveiling the Implicit Toxicity in Large Language Models
abstract
The open-endedness of large language models (LLMs) combined with their impressive capabilities may lead to new safety issues when being exploited for malicious use.While recent studies primarily focus on probing toxic outputs that can be easily detected with existing toxicity classifiers, we show that LLMs can generate diverse implicit toxic outputs that are exceptionally difficult to detect via simply zero-shot prompting.Moreover, we propose a reinforcement learning (RL) based attacking method to further induce the implicit toxicity in LLMs.Specifically, we optimize the language model with a reward that prefers implicit toxic outputs to explicit toxic and non-toxic ones.Experiments on five widely-adopted toxicity classifiers demonstrate that the attack success rate can be significantly improved through RL fine-tuning.For instance, the RL-finetuned LLaMA-13B model achieves an attack success rate of 90.04% on BAD and 62.85% on Davinci003.Our findings suggest that LLMs pose a significant threat in generating undetectable implicit toxic outputs.We further show that fine-tuning toxicity classifiers on the annotated examples from our attacking method can effectively enhance their ability to detect LLM-generated implicit toxic language.The code is publicly available at https://github. com/thu-coai/Implicit-Toxicity.
Jiaxin Wen, Pei Ke, Hao Sun 0012, Zhexin Zhang, Chengfei Li, Jinfeng Bai, Minlie Huang
EMNLP7
2023 Re³Dial: Retrieve, Reorganize and Rescale Conversations for Long-Turn Open-Domain Dialogue Pre-training
abstract
Pre-training on large-scale open-domain dialogue data can substantially improve the performance of dialogue models.However, the pre-trained dialogue model's ability to utilize long-range context is limited due to the scarcity of long-turn dialogue sessions.Most dialogues in existing pre-training corpora contain fewer than three turns of dialogue.To alleviate this issue, we propose the Retrieve, Reorganize and Rescale framework (Re 3 Dial), which can automatically construct billion-scale long-turn dialogues by reorganizing existing short-turn ones.Given a short-turn session, Re 3 Dial first employs a session retriever to retrieve coherent consecutive sessions.To this end, we train the retriever to capture semantic and discourse relations within multi-turn dialogues through contrastive training.Next, Re 3 Dial samples a session from retrieved results following a diversity sampling strategy, which is designed to penalize repetitive or generic sessions.A longer session is then derived by concatenating the original session and the sampled session.By repeating the above process, Re 3 Dial can yield a coherent long-turn dialogue.Extensive experiments on multiple multi-turn dialogue benchmarks demonstrate that Re 3 Dial significantly improves the dialogue model's ability to utilize long-range context and thus generate more sensible and informative responses.Finally, we build a toolkit for efficiently rescaling conversations with Re 3 Dial, which enables us to construct a corpus containing 1B Chinese dialogue sessions with 11.3 turns on average (5× longer than the original corpus).Our retriever model, code, and data is publicly available at https://github.com/thu-coai/Re3Dial.
Jiaxin Wen, Hao Zhou 0012, Jian Guan 0002, Jie Zhou 0016, Minlie Huang
EMNLP5
2023 Tailoring Language Generation Models under Total Variation Distance
Haozhe Ji, Pei Ke, Zhipeng Hu, Minlie Huang
ICLR5
2023 Synthetic Prompting: Generating Chain-of-Thought Demonstrations for Large Language Models
abstract
Large language models can perform various reasoning tasks by using chain-of-thought prompting, which guides them to find answers through step-by-step demonstrations. However, the quality of the prompts depends on the demonstrations given to the models, and creating many of them by hand is costly. We introduce Synthetic prompting, a method that leverages a few handcrafted examples to prompt the model to generate more examples by itself, and selects effective demonstrations to elicit better reasoning. Our method alternates between a backward and forward process to generate new examples. The backward process generates a question that match a sampled reasoning chain, so that the question is solvable and clear. The forward process produces a more detailed reasoning chain for the question, improving the quality of the example. We evaluate our method on numerical, symbolic, and algorithmic reasoning tasks, and show that it outperforms existing prompting techniques.
Zhihong Shao, Yeyun Gong, Yelong Shen, Minlie Huang, Nan Duan 0001, Weizhu Chen
ICML4
2023 Directed Acyclic Transformer Pre-training for High-quality Non-autoregressive Text Generation
abstract
Abstract Non-AutoRegressive (NAR) text generation models have drawn much attention because of their significantly faster decoding speed and good generation quality in machine translation. However, in a wider range of text generation tasks, existing NAR models lack proper pre-training, making them still far behind the pre-trained autoregressive models. In this paper, we propose Pre-trained Directed Acyclic Transformer (PreDAT) and a novel pre-training task to promote prediction consistency in NAR generation. Experiments on five text generation tasks show that our PreDAT remarkably outperforms existing pre-trained NAR models (+4.2 score on average) and even achieves better results than pre-trained autoregressive baselines in n-gram-based metrics, along with 17 times speedup in throughput. Further analysis shows that PreDAT benefits from the unbiased prediction order that alleviates the error accumulation problem in autoregressive generation, which provides new insights into the advantages of NAR generation.1
Fei Huang 0005, Pei Ke, Minlie Huang
Trans. Assoc. Comput. Linguistics3
2023 Bridging the Gap between Synthetic and Natural Questions via Sentence Decomposition for Semantic Parsing
abstract
Abstract Semantic parsing maps natural language questions into logical forms, which can be executed against a knowledge base for answers. In real-world applications, the performance of a parser is often limited by the lack of training data. To facilitate zero-shot learning, data synthesis has been widely studied to automatically generate paired questions and logical forms. However, data synthesis methods can hardly cover the diverse structures in natural languages, leading to a large gap in sentence structure between synthetic and natural questions. In this paper, we propose a decomposition-based method to unify the sentence structures of questions, which benefits the generalization to natural questions. Experiments demonstrate that our method significantly improves the semantic parser trained on synthetic data (+7.9% on KQA and +8.9% on ComplexWebQuestions in terms of exact match accuracy). Extensive analysis demonstrates that our method can better generalize to natural questions with novel text expressions compared with baselines. Besides semantic parsing, our idea potentially benefits other semantic understanding tasks by mitigating the distracting structure features. To illustrate this, we extend our method to the task of sentence embedding learning, and observe substantial improvements on sentence retrieval (+13.1% for Hit@1).
Yilin Niu, Fei Huang 0005, Wei Liu 0005, Jianwei Cui 0002, Bin Wang 0004, Minlie Huang
Trans. Assoc. Comput. Linguistics6
2023 Enhanced Multi-Domain Dialogue State Tracker With Second-Order Slot Interactions
abstract
Dialogue state tracking (DST) is often used to track the system's understanding of the user goal in task-oriented dialogue systems. Existing DST methods mainly fall into two categories according to their adopted model structure: non-hierarchical and hierarchical models. The former takes the whole dialogue history as inputs during each conversation round, while the latter leverages both an utterance encoder and a dialogue encoder to efficiently model the long-term dialogue dependency. However, few of them exploit the second-order slot interaction, which refers to the pair-wise semantic relationships between different slots. As a result, these methods fall short in the context understanding throughout conversations, leading to sub-optimal performance. Towards this end, in this paper, we present a novel hierarchy-based DST framework equipped with a well-designed value copy mechanism. In particular, to model the second-order slot interaction, we firstly encode the utterance via a state reuse module to yield slot-sensitive context representation. We then selectively and effectively copy the filled values from other slots to attain more accurate state tracking. In order to evaluate the effectiveness of the proposed method, we perform extensive experiments on the widely adopted benchmark dataset MultiWOZ2.1. Our experimental results demonstrate the superiority in context understanding, as well as the strong generalization capability under a zero-shot setting compared with several DST baselines.
Fangkai Jiao, Minlie Huang, Liqiang Nie
IEEE ACM Trans. Audio Speech Lang. Process.3
2022 CEM: Commonsense-Aware Empathetic Response Generation
abstract
A key trait of daily conversations between individuals is the ability to express empathy towards others, and exploring ways to implement empathy is a crucial step towards human-like dialogue systems. Previous approaches on this topic mainly focus on detecting and utilizing the user’s emotion for generating empathetic responses. However, since empathy includes both aspects of affection and cognition, we argue that in addition to identifying the user’s emotion, cognitive understanding of the user’s situation should also be considered. To this end, we propose a novel approach for empathetic response generation, which leverages commonsense to draw more information about the user’s situation and uses this additional information to further enhance the empathy expression in generated responses. We evaluate our approach on EMPATHETICDIALOGUES, which is a widely-used benchmark dataset for empathetic response generation. Empirical results demonstrate that our approach outperforms the baseline models in both automatic and human evaluations and can generate more informative and empathetic responses. Our code is available at https://github.com/Sahandfer/CEM.
Sahand Sabour, Chujie Zheng, Minlie Huang
AAAI3
2022 Continual Prompt Tuning for Dialog State Tracking
abstract
A desirable dialog system should be able to continually learn new skills without forgetting old ones, and thereby adapt to new domains or tasks in its life cycle.However, continually training a model often leads to a well-known catastrophic forgetting issue.In this paper, we present Continual Prompt Tuning, a parameterefficient framework that not only avoids forgetting but also enables knowledge transfer between tasks.To avoid forgetting, we only learn and store a few prompt tokens' embeddings for each task while freezing the backbone pre-trained model.To achieve bi-directional knowledge transfer among tasks, we propose several techniques (continual prompt initialization, query fusion, and memory replay) to transfer knowledge from preceding tasks and a memory-guided technique to transfer knowledge from subsequent tasks.Extensive experiments demonstrate the effectiveness and efficiency of our proposed method on continual learning for dialog state tracking, compared with state-of-the-art baselines.
Qi Zhu 0007, Fei Mi, Xiaoyan Zhu 0001, Minlie Huang
ACL (1)5
2022 PPT: Pre-trained Prompt Tuning for Few-shot Learning
abstract
Prompts for pre-trained language models (PLMs) have shown remarkable performance by bridging the gap between pre-training tasks and various downstream tasks.Among these methods, prompt tuning, which freezes PLMs and only tunes soft prompts, provides an efficient and effective solution for adapting largescale PLMs to downstream tasks.However, prompt tuning is yet to be fully explored.In our pilot experiments, we find that prompt tuning performs comparably with conventional full-model tuning when downstream data are sufficient, whereas it is much worse under fewshot learning settings, which may hinder the application of prompt tuning.We attribute this low performance to the manner of initializing soft prompts.Therefore, in this work, we propose to pre-train prompts by adding soft prompts into the pre-training stage to obtain a better initialization.We name this Pretrained Prompt Tuning framework "PPT".To ensure the generalization of PPT, we formulate similar classification tasks into a unified task form and pre-train soft prompts for this unified task.Extensive experiments show that tuning pre-trained prompts for downstream tasks can reach or even outperform full-model fine-tuning under both full-data and few-shot settings.Our approach is effective and efficient for using large-scale PLMs in practice.The code is publicly available at https:// github.com/thu-coai/PPT.
Yuxian Gu, Xu Han 0007, Zhiyuan Liu 0001, Minlie Huang
ACL (1)4
2022 CTRLEval: An Unsupervised Reference-Free Metric for Evaluating Controlled Text Generation
abstract
Existing reference-free metrics have obvious limitations for evaluating controlled text generation models.Unsupervised metrics can only provide a task-agnostic evaluation result which correlates weakly with human judgments, whereas supervised ones may overfit task-specific data with poor generalization ability to other datasets.In this paper, we propose an unsupervised reference-free metric called CTRLEval, which evaluates controlled text generation from different aspects by formulating each aspect into multiple text infilling tasks.On top of these tasks, the metric assembles the generation probabilities from a pre-trained language model without any model training.Experimental results show that our metric has higher correlations with human judgments than other baselines, while obtaining better generalization of evaluating generated texts from different models and with different qualities 1 .
Pei Ke, Hao Zhou 0012, Yankai Lin 0001, Peng Li 0030, Jie Zhou 0016, Xiaoyan Zhu 0001, Minlie Huang
ACL (1)7
2022 Answering Open-Domain Multi-Answer Questions via a Recall-then-Verify Framework
abstract
Open-domain questions are likely to be openended and ambiguous, leading to multiple valid answers.Existing approaches typically adopt the rerank-then-read framework, where a reader reads top-ranking evidence to predict answers.According to our empirical analysis, this framework faces three problems: first, to leverage a large reader under a memory constraint, the reranker should select only a few relevant passages to cover diverse answers, while balancing relevance and diversity is non-trivial; second, the small reading budget prevents the reader from accessing valuable retrieved evidence filtered out by the reranker; third, when using a generative reader to predict answers all at once based on all selected evidence, whether a valid answer will be predicted also pathologically depends on the evidence of some other valid answer(s).To address these issues, we propose to answer open-domain multi-answer questions with a recall-then-verify framework, which separates the reasoning process of each answer so that we can make better use of retrieved evidence while also leveraging large models under the same memory constraint.Our framework achieves state-of-the-art results on two multi-answer datasets, and predicts significantly more gold answers than a rerank-thenread system that uses an oracle reranker.
Zhihong Shao, Minlie Huang
ACL (1)2
2022 COLD: A Benchmark for Chinese Offensive Language Detection
abstract
Offensive language detection is increasingly crucial for maintaining a civilized social media platform and deploying pre-trained language models.However, this task in Chinese is still under exploration due to the scarcity of reliable datasets.To this end, we propose a benchmark -COLD for Chinese offensive language analysis, including a Chinese Offensive Language Dataset -COLDATASET and a baseline detector -COLDETECTOR which is trained on the dataset.We show that the COLD benchmark contributes to Chinese offensive language detection which is challenging for existing resources.We then deploy the COLDETECTOR and conduct detailed analyses on popular Chinese pre-trained language models.We first analyze the offensiveness of existing generative models and show that these models inevitably expose varying degrees of offensive issues.Furthermore, we investigate the factors that influence the offensive generations, and we find that anti-bias contents and keywords referring to certain groups or revealing negative attitudes trigger offensive outputs easier.
Jiawen Deng 0006, Jingyan Zhou, Hao Sun 0012, Chujie Zheng, Fei Mi, Helen M. Meng, Minlie Huang
EMNLP7
2022 Learning Instructions with Unlabeled Data for Zero-Shot Cross-Task Generalization
abstract
Training language models to learn from human instructions for zero-shot cross-task generalization has attracted much attention in NLP communities.Recently, instruction tuning (IT), which fine-tunes a pre-trained language model on a massive collection of tasks described via human-craft instructions, has been shown effective in instruction learning for unseen tasks.However, IT relies on a large amount of humanannotated samples, which restricts its generalization.Unlike labeled data, unlabeled data are often massive and cheap to obtain.In this work, we study how IT can be improved with unlabeled data.We first empirically explore the IT performance trends versus the number of labeled data, instructions, and training tasks.We find it critical to enlarge the number of training instructions, and the instructions can be underutilized due to the scarcity of labeled data.Then, we propose Unlabeled Data Augmented Instruction Tuning (UDIT) to take better advantage of the instructions during IT by constructing pseudo-labeled data from unlabeled plain texts.We conduct extensive experiments to show UDIT's effectiveness in various scenarios of tasks and datasets.We also comprehensively analyze the key factors of UDIT to investigate how to better improve IT with unlabeled data.
Yuxian Gu, Pei Ke, Xiaoyan Zhu 0001, Minlie Huang
EMNLP4
2022 CDConv: A Benchmark for Contradiction Detection in Chinese Conversations
abstract
Chujie Zheng, Jinfeng Zhou, Yinhe Zheng, Libiao Peng, Zhen Guo, Wenquan Wu, Zheng-Yu Niu, Hua Wu, Minlie Huang. Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing. 2022.
Chujie Zheng, Jinfeng Zhou, Yinhe Zheng, Libiao Peng, Wenquan Wu, Zhengyu Niu, Hua Wu 0003, Minlie Huang
EMNLP9
2022 Aligning Recommendation and Conversation via Dual Imitation
abstract
Human conversations of recommendation naturally involve the shift of interests which can align the recommendation actions and conversation process to make accurate recommendations with rich explanations.However, existing conversational recommendation systems (CRS) ignore the advantage of user interest shift in connecting recommendation and conversation, which leads to an ineffective loose coupling structure of CRS.To address this issue, by modeling the recommendation actions as recommendation paths in a knowledge graph (KG), we propose DICR (Dual Imitation for Conversational Recommendation), which designs a dual imitation to explicitly align the recommendation paths and user interest shift paths in a recommendation module and a conversation module, respectively.By exchanging alignment signals, DICR achieves bidirectional promotion between recommendation and conversation modules and generates high-quality responses with accurate recommendations and coherent explanations.Experiments demonstrate that DICR outperforms the state-of-theart models on recommendation and conversation performance with automatic, human, and novel explainability metrics.
Jinfeng Zhou, Bo Wang 0011, Minlie Huang, Ruifang He, Yuexian Hou
EMNLP3
2022 Acceleration of Federated Learning with Alleviated Forgetting in Local Training
Chencheng Xu, Zhiwei Hong, Minlie Huang, Tao Jiang 0001
ICLR3
2022 On the Learning of Non-Autoregressive Transformers
abstract
Non-autoregressive Transformer (NAT) is a family of text generation models, which aims to reduce the decoding latency by predicting the whole sentences in parallel. However, such latency reduction sacrifices the ability to capture left-to-right dependencies, thereby making NAT learning very challenging. In this paper, we present theoretical and empirical analyses to reveal the challenges of NAT learning and propose a unified perspective to understand existing successes. First, we show that simply training NAT by maximizing the likelihood can lead to an approximation of marginal distributions but drops all dependencies between tokens, where the dropped information can be measured by the dataset’s conditional total correlation. Second, we formalize many previous objectives in a unified framework and show that their success can be concluded as maximizing the likelihood on a proxy distribution, leading to a reduced information loss. Empirical studies show that our perspective can explain the phenomena in NAT learning and guide the design of new training methods.
Fei Huang 0005, Tianhua Tao, Hao Zhou 0012, Lei Li 0005, Minlie Huang
ICML5
2022 Directed Acyclic Transformer for Non-Autoregressive Machine Translation
abstract
Non-autoregressive Transformers (NATs) significantly reduce the decoding latency by generating all tokens in parallel. However, such independent predictions prevent NATs from capturing the dependencies between the tokens for generating multiple possible translations. In this paper, we propose Directed Acyclic Transfomer (DA-Transformer), which represents the hidden states in a Directed Acyclic Graph (DAG), where each path of the DAG corresponds to a specific translation. The whole DAG simultaneously captures multiple translations and facilitates fast predictions in a non-autoregressive fashion. Experiments on the raw training data of WMT benchmark show that DA-Transformer substantially outperforms previous NATs by about 3 BLEU on average, which is the first NAT model that achieves competitive results with autoregressive Transformers without relying on knowledge distillation.
Fei Huang 0005, Hao Zhou 0012, Yang Liu 0005, Hang Li 0001, Minlie Huang
ICML5
2022 Curriculum-Based Self-Training Makes Better Few-Shot Learners for Data-to-Text Generation
abstract
Despite the success of text-to-text pre-trained models in various natural language generation (NLG) tasks, the generation performance is largely restricted by the number of labeled data in downstream tasks, particularly in data-to-text generation tasks. Existing works mostly utilize abundant unlabeled structured data to conduct unsupervised pre-training for task adaption, which fail to model the complex relationship between source structured data and target texts. Thus, we introduce self-training as a better few-shot learner than task-adaptive pre-training, which explicitly captures this relationship via pseudo-labeled data generated by the pre-trained model. To alleviate the side-effect of low-quality pseudo-labeled data during self-training, we propose a novel method called Curriculum-Based Self-Training (CBST) to effectively leverage unlabeled data in a rearranged order determined by the difficulty of text generation. Experimental results show that our method can outperform fine-tuning and task-adaptive pre-training methods, and achieve state-of-the-art performance in the few-shot setting of data-to-text generation.
Pei Ke, Haozhe Ji, Yi Huang 0017, Junlan Feng, Xiaoyan Zhu 0001, Minlie Huang
IJCAI7
2022 A Corpus for Understanding and Generating Moral Stories
abstract
Teaching morals is one of the most important purposes of storytelling.An essential ability for understanding and writing moral stories is bridging story plots and implied morals.Its challenges mainly lie in: (1) grasping knowledge about abstract concepts in morals, (2) capturing inter-event discourse relations in stories, and (3) aligning value preferences of stories and morals concerning good or bad behavior.In this paper, we propose two understanding tasks and two generation tasks to assess these abilities of machines.We present STORAL, a new dataset of Chinese and English human-written moral stories.We show the difficulty of the proposed tasks by testing various models with automatic and manual evaluation on STORAL.Furthermore, we present a retrieval-augmented algorithm that effectively exploits related concepts or events in training sets as additional guidance to improve performance on these tasks.
Jian Guan 0002, Minlie Huang
NAACL-HLT3
2022 LaMemo: Language Modeling with Look-Ahead Memory
abstract
Haozhe Ji, Rongsheng Zhang, Zhenyu Yang, Zhipeng Hu, Minlie Huang. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.
Haozhe Ji, Zhipeng Hu, Minlie Huang
NAACL-HLT5
2022 Persona-Guided Planning for Controlling the Protagonist's Persona in Story Generation
abstract
Endowing the protagonist with a specific personality is essential for writing an engaging story.In this paper, we aim to control the protagonist's persona in story generation, i.e., generating a story from a leading context and a persona description, where the protagonist should exhibit the specified personality through a coherent event sequence.Considering that personas are usually embodied implicitly and sparsely in stories, we propose a planning-based generation model named CONPER to explicitly model the relationship between personas and events.CON-PER first plans events of the protagonist's behavior which are motivated by the specified persona through predicting one target sentence, then plans the plot as a sequence of keywords with the guidance of the predicted persona-related events and commonsense knowledge, and finally generates the whole story.Both automatic and manual evaluation results demonstrate that CONPER outperforms state-of-the-art baselines for generating more coherent and persona-controllable stories.Our code is available at https:// github.com/thu-coai/ConPer.
Zhexin Zhang, Jiaxin Wen, Jian Guan 0002, Minlie Huang
NAACL-HLT4
2022 LOT: A Story-Centric Benchmark for Evaluating Chinese Long Text Understanding and Generation
abstract
Abstract Standard multi-task benchmarks are essential for developing pretraining models that can generalize to various downstream tasks. Existing benchmarks for natural language processing (NLP) usually focus only on understanding or generating short texts. However, long text modeling requires many distinct abilities in contrast to short texts, such as the modeling of long-range discourse and commonsense relations, and the coherence and controllability of generation. The lack of standardized benchmarks makes it difficult to assess these abilities of a model and fairly compare different models, especially Chinese models. Therefore, we propose a story-centric benchmark named LOT for evaluating Chinese long text modeling, which aggregates two understanding tasks and two generation tasks. We construct new datasets for these tasks based on human-written Chinese stories with hundreds of words. Furthermore, we release an encoder-decoder-based Chinese long text pretraining model named LongLM with up to 1 billion parameters. We pretrain LongLM on 120G Chinese novels with two generative tasks including text infilling and conditional continuation. Extensive experiments show that LongLM outperforms similar-sized pretraining models substantially on both the understanding and generation tasks in LOT.
Jian Guan 0002, Zhuoer Feng, Yamei Chen, Ruilin He, Xiaoxi Mao, Changjie Fan, Minlie Huang
Trans. Assoc. Comput. Linguistics7
2022 End-to-End Task-Oriented Dialog Modeling With Semi-Structured Knowledge Management
abstract
Current task-oriented dialog (TOD) systems mostly manage structured knowledge (e.g. databases and tables) to guide the goal-oriented conversations. However, they fall short of handling dialogs which also involve unstructured knowledge (e.g. reviews and documents). In this paper, we formulate a task of modeling TOD grounded on a fusion of structured and unstructured knowledge. To address this task, we propose a TOD system with semi-structured knowledge management, SeKnow, which extends the belief state to manage knowledge with both structured and unstructured contents. Furthermore, we introduce two implementations of SeKnow based on a non-pretrained sequence-to-sequence model and a pretrained language model, respectively. Both implementations use the end-to-end manner to jointly optimize dialog modeling grounded on structured and unstructured knowledge. We conduct experiments on a modified version of MultiWOZ 2.1 dataset, Mod-MultiWOZ 2.1, where dialogs are processed to involve semi-structured knowledge. Experimental results show that SeKnow has strong performances in both end-to-end dialog and intermediate knowledge management, compared to existing TOD systems and their extensions with pipeline knowledge management schemes.
Silin Gao, Ryuichi Takanobu, Antoine Bosselut, Minlie Huang
IEEE ACM Trans. Audio Speech Lang. Process.4
2022 AdvExpander: Generating Natural Language Adversarial Examples by Expanding Text
abstract
Adversarial examples are vital to expose vulnerability of machine learning models. Despite the success of the most popular word-level substitution-based attacks which substitute some words in the original examples, only substitution is insufficient to uncover all robustness issues of models. In this paper, we focus on perturbations beyond word-level substitution, and presentAdvExpander, a method that crafts new adversarial examples by expanding text. We first utilize linguistic rules to determine which constituents to expand and what types of modifiers to expand with. We then expand each constituent by inserting an adversarial modifier searched from a pre-trained CVAE-based generative model. To ensure that our adversarial examples are label-preserving for text matching, we also constrain the modifications with a heuristic rule. Experiments on three classification tasks verify the effectiveness of AdvExpander and the validity of our adversarial examples. AdvExpander is significantly more effective than sentence-level attack baselines and is complementary to previous word substitution-based attacks, thus promising to reveal new robustness issues.
Zhihong Shao, Zhongqin Wu, Minlie Huang
IEEE ACM Trans. Audio Speech Lang. Process.3
2022 Topic-Guided Conversational Recommender in Multiple Domains
abstract
Conversational systems have recently attracted significant attention. Both the research community and industry believe that it will exert huge impact on human-computer interaction, and specifically, the IR/RecSys community has begun to explore Conversational Recommendation. In real-life scenarios, such systems are often urgently needed in helping users accomplishing different tasks under various situations. However, existing works still face several shortcomings: (1) Most efforts are largely confined in single task setting. They fall short of hands in handling tasks across domains. (2) Aside from soliciting user preference from dialogue history, a conversational recommender naturally has access to the back-end data structure which should be fully leveraged to yield good recommendations. In this paper, we thus present a Topic-guided Conversational Recommender (TCR) which is specifically designed for the multi-domain setting. It augments the sequence-to-sequence (seq2seq) models with a neural latent topic component to better guide the response generation. To better leverage the dialogue history and the back-end data structure, we adopt a graph convolutional network (GCN) to model the relationships between different recommendation candidates while also capture the match between candidates and the dialogue history. We then seamlessly combine these two parts with the idea of pointer networks. We perform extensive evaluation on a large-scale task-oriented multi-domain dialogue dataset and the results show that our method achieves superior performance as compared to a wide range of baselines.
Lizi Liao, Ryuichi Takanobu, Yunshan Ma 0002, Xun Yang 0001, Minlie Huang, Tat-Seng Chua
IEEE Trans. Knowl. Data Eng.5
2021 Stylized Dialogue Response Generation Using Stylized Unpaired Texts
abstract
Generating stylized responses is essential to build intelligent and engaging dialogue systems. However, this task is far from well-explored due to the difficulties of rendering a particular style in coherent responses, especially when the target style is embedded only in unpaired texts that cannot be directly used to train the dialogue model. This paper proposes a stylized dialogue generation method that can capture stylistic features embedded in unpaired texts. Specifically, our method can produce dialogue responses that are both coherent to the given context and conform to the target style. In this study, an inverse dialogue model is first introduced to predict possible posts for the input responses. Then this inverse model is used to generate stylized pseudo dialogue pairs based on these stylized unpaired texts. Further, these pseudo pairs are employed to train the stylized dialogue model with a joint training process. A style routing approach is proposed to intensify stylistic features in the decoder. Automatic and manual evaluations on two datasets demonstrate that our method outperforms competitive baselines in producing coherent and style-intensive dialogue responses.
Yinhe Zheng, Zikai Chen, Shilei Huang, Xiaoxi Mao, Minlie Huang
AAAI6
2021 Long Text Generation by Modeling Sentence-Level and Discourse-Level Coherence
abstract
Jian Guan, Xiaoxi Mao, Changjie Fan, Zitao Liu, Wenbiao Ding, Minlie Huang. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
Jian Guan 0002, Xiaoxi Mao, Changjie Fan, Zitao Liu 0001, Wenbiao Ding, Minlie Huang
ACL/IJCNLP (1)6
2021 OpenMEVA: A Benchmark for Evaluating Open-ended Story Generation Metrics
abstract
Jian Guan, Zhexin Zhang, Zhuoer Feng, Zitao Liu, Wenbiao Ding, Xiaoxi Mao, Changjie Fan, Minlie Huang. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
Jian Guan 0002, Zhexin Zhang, Zhuoer Feng, Zitao Liu 0001, Wenbiao Ding, Xiaoxi Mao, Changjie Fan, Minlie Huang
ACL/IJCNLP (1)8
2021 Robustness Testing of Language Understanding in Task-Oriented Dialog
abstract
Jiexi Liu, Ryuichi Takanobu, Jiaxin Wen, Dazhen Wan, Hongguang Li, Weiran Nie, Cheng Li, Wei Peng, Minlie Huang. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
Jiexi Liu 0002, Ryuichi Takanobu, Jiaxin Wen, Dazhen Wan, Weiran Nie, Cheng Li 0040, Wei Peng 0011, Minlie Huang
ACL/IJCNLP (1)9
2021 Towards Emotional Support Dialog Systems
abstract
Siyang Liu, Chujie Zheng, Orianna Demasi, Sahand Sabour, Yu Li, Zhou Yu, Yong Jiang, Minlie Huang. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
Siyang Liu 0003, Chujie Zheng, Orianna Demasi, Sahand Sabour, Yu Li 0013, Zhou Yu 0005, Yong Jiang 0001, Minlie Huang
ACL/IJCNLP (1)8
2021 A Semantic-based Method for Unsupervised Commonsense Question Answering
abstract
Yilin Niu, Fei Huang, Jiaming Liang, Wenkai Chen, Xiaoyan Zhu, Minlie Huang. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
Yilin Niu, Fei Huang 0005, Xiaoyan Zhu 0001, Minlie Huang
ACL/IJCNLP (1)6
2021 ERICA: Improving Entity and Relation Understanding for Pre-trained Language Models via Contrastive Learning
abstract
Yujia Qin, Yankai Lin, Ryuichi Takanobu, Zhiyuan Liu, Peng Li, Heng Ji, Minlie Huang, Maosong Sun, Jie Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
Yujia Qin, Yankai Lin 0001, Ryuichi Takanobu, Zhiyuan Liu 0001, Peng Li 0030, Heng Ji 0001, Minlie Huang, Maosong Sun 0001, Jie Zhou 0016
ACL/IJCNLP (1)7
2021 A Mutual Information Maximization Approach for the Spurious Solution Problem in Weakly Supervised Question Answering
abstract
Zhihong Shao, Lifeng Shang, Qun Liu, Minlie Huang. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
Zhihong Shao, Lifeng Shang, Qun Liu 0001, Minlie Huang
ACL/IJCNLP (1)4
2021 Diversifying Dialog Generation via Adaptive Label Smoothing
abstract
Yida Wang, Yinhe Zheng, Yong Jiang, Minlie Huang. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
Yida Wang 0009, Yinhe Zheng, Yong Jiang 0001, Minlie Huang
ACL/IJCNLP (1)4
2021 DiscoDVT: Generating Long Text with Discourse-Aware Discrete Variational Transformer
abstract
Despite the recent advances in applying pretrained language models to generate highquality texts, generating long passages that maintain long-range coherence is yet challenging for these models.In this paper, we propose DISCODVT, a discourse-aware discrete variational Transformer to tackle the incoherence issue.DISCODVT learns a discrete variable sequence that summarizes the global structure of the text and then applies it to guide the generation process at each decoding step.To further embed discourse-aware information into the discrete latent representations, we introduce an auxiliary objective to model the discourse relations within the text.We conduct extensive experiments on two open story generation datasets and demonstrate that the latent codes learn meaningful correspondence to the discourse structures that guide the model to generate long texts with better long-range coherence.
Haozhe Ji, Minlie Huang
EMNLP (1)2
2021 CR-Walker: Tree-Structured Graph Reasoning and Dialog Acts for Conversational Recommendation
abstract
Growing interests have been attracted in Conversational Recommender Systems (CRS), which explore user preference through conversational interactions in order to make appropriate recommendation.However, there is still a lack of ability in existing CRS to ( 1) traverse multiple reasoning paths over background knowledge to introduce relevant items and attributes, and (2) arrange selected entities appropriately under current system intents to control response generation.To address these issues, we propose CR-Walker in this paper, a model that performs tree-structured reasoning on a knowledge graph, and generates informative dialog acts to guide language generation.The unique scheme of tree-structured reasoning views traversed entity at each hop as part of dialog acts to facilitate language generation, which links how entities are selected and expressed.Automatic and human evaluations show that CR-Walker can arrive at more accurate recommendation, and generate more informative and engaging responses.
Wenchang Ma, Ryuichi Takanobu, Minlie Huang
EMNLP (1)3
2021 Self-training Improves Pre-training for Few-shot Learning in Task-oriented Dialog Systems
abstract
As the labeling cost for different modules in task-oriented dialog (ToD) systems is expensive, a major challenge is to train different modules with the least amount of labeled data.Recently, large-scale pre-trained language models, have shown promising results for few-shot learning in ToD.In this paper, we devise a selftraining approach to utilize the abundant unlabeled dialog data to further improve state-ofthe-art pre-trained models in few-shot learning scenarios for ToD systems.Specifically, we propose a self-training approach that iteratively labels the most confident unlabeled data to train a stronger Student model.Moreover, a new text augmentation technique (GradAug) is proposed to better train the Student by replacing non-crucial tokens using a masked language model.We conduct extensive experiments and present analyses on four downstream tasks in ToD, including intent classification, dialog state tracking, dialog act prediction, and response selection.Empirical results demonstrate that the proposed self-training approach consistently improves state-of-the-art pre-trained models (BERT, ToD-BERT) when only a small number of labeled data are available.
Fei Mi, Wanhao Zhou, Lingjing Kong 0001, Fengyu Cai, Minlie Huang, Boi Faltings
EMNLP (1)5
2021 Transferable Persona-Grounded Dialogues via Grounded Minimal Edits
abstract
Grounded dialogue models generate responses that are grounded on certain concepts.Limited by the distribution of grounded dialogue data, models trained on such data face the transferability challenges in terms of the data distribution and the type of grounded concepts.To address the challenges, we propose the grounded minimal editing framework, which minimally edits existing responses to be grounded on the given concept.Focusing on personas, we propose Grounded Minimal Editor (GME), which learns to edit by disentangling and recombining persona-related and persona-agnostic parts of the response.To evaluate persona-grounded minimal editing, we present the PERSONAMI-NEDIT dataset, and experimental results show that GME outperforms competitive baselines by a large margin.To evaluate the transferability, we experiment on the test set of BLEND-EDSKILLTALK and show that GME can edit dialogue models' responses to largely improve their persona consistency while preserving the use of knowledge and empathy. 1
Chen Henry Wu, Yinhe Zheng, Xiaoxi Mao, Minlie Huang
EMNLP (1)4
2021 EARL: Informative Knowledge-Grounded Conversation Generation with Entity-Agnostic Representation Learning
abstract
Generating informative and appropriate responses is challenging but important for building human-like dialogue systems.Although various knowledge-grounded conversation models have been proposed, these models have limitations in utilizing knowledge that infrequently occurs in the training data, not to mention integrating unseen knowledge into conversation generation.In this paper, we propose an Entity-Agnostic Representation Learning (EARL) method to introduce knowledge graphs to informative conversation generation.Unlike traditional approaches that parameterize the specific representation for each entity, EARL utilizes the context of conversations and the relational structure of knowledge graphs to learn the category representation for entities, which is generalized to incorporating unseen entities in knowledge graphs into conversation generation.Automatic and manual evaluations demonstrate that our model can generate more informative, coherent, and natural responses than baseline models.
Hao Zhou 0012, Minlie Huang, Wei Chen 0034, Xiaoyan Zhu 0001
EMNLP (1)2
2021 Independence-aware Advantage Estimation
abstract
Most of existing advantage function estimation methods in reinforcement learning suffer from the problem of high variance, which scales unfavorably with the time horizon. To address this challenge, we propose to identify the independence property between current action and future states in environments, which can be further leveraged to effectively reduce the variance of the advantage estimation. In particular, the recognized independence property can be naturally utilized to construct a novel importance sampling advantage estimator with close-to-zero variance even when the Monte-Carlo return signal yields a large variance. To further remove the risk of the high variance introduced by the new estimator, we combine it with existing Monte-Carlo estimator via a reward decomposition model learned by minimizing the estimation variance. Experiments demonstrate that our method achieves higher sample efficiency compared with existing advantage estimation methods in complex environments.
Pushi Zhang, Li Zhao 0007, Jiang Bian 0002, Minlie Huang, Tao Qin 0001, Tie-Yan Liu
IJCAI5
2021 FIVES: Feature Interaction Via Edge Search for Large-Scale Tabular Data
abstract
High-order interactive features capture the correlation between different columns and thus are promising to enhance various learning tasks on ubiquitous tabular data. To automate the generation of interactive features, existing works either explicitly traverse the feature space or implicitly express the interactions via intermediate activations of some designed models. These two kinds of methods show that there is essentially a trade-off between feature interpretability and search efficiency. To possess both of their merits, we propose a novel method named Feature Interaction Via Edge Search (FIVES), which formulates the task of interactive feature generation as searching for edges on the defined feature graph. Specifically, we first present our theoretical evidence that motivates us to search for useful interactive features with increasing order. Then we instantiate this search strategy by optimizing both a dedicated graph neural network (GNN) and the adjacency tensor associated with the defined feature graph. In this way, the proposed FIVES method simplifies the time-consuming traversal as a typical training course of GNN and enables explicit feature generation according to the learned adjacency tensor. Experimental results on both benchmark and real-world datasets show the advantages of FIVES over several state-of-the-art methods. Moreover, the interactive features identified by FIVES are deployed on the recommender system of Taobao, a worldwide leading e-commerce platform. Results of an online A/B testing further verify the effectiveness of the proposed method FIVES, and we further provide FIVES as AI utilities for the customers of Alibaba Cloud.
Yuexiang Xie, Zhen Wang 0036, Yaliang Li, Bolin Ding, Nezihe Merve Gürel, Ce Zhang 0001, Minlie Huang, Wei Lin 0016, Jingren Zhou 0001
KDD7
2021 MultiWOZ 2.3: A Multi-domain Task-Oriented Dialogue Dataset Enhanced with Annotation Corrections and Co-Reference Annotation
Ryuichi Takanobu, Yixin Lian, Chongxuan Huang, Dazhen Wan, Wei Peng 0011, Minlie Huang
NLPCC (2)8
2021 MMConv: An Environment for Multimodal Conversational Search across Multiple Domains
abstract
Although conversational search has become a hot topic in both dialogue research and IR community, the real breakthrough has been limited by the scale and quality of datasets available. To address this fundamental obstacle, we introduce the Multimodal Multi-domain Conversational dataset (MMConv), a fully annotated collection of human-to-human role-playing dialogues spanning over multiple domains and tasks. The contribution is two-fold. First, beyond the task-oriented multimodal dialogues among user and agent pairs, dialogues are fully annotated with dialogue belief states and dialogue acts. More importantly, we create a relatively comprehensive environment for conducting multimodal conversational search with real user settings, structured venue database, annotated image repository as well as crowd-sourced knowledge database. A detailed description of the data collection procedure along with a summary of data structure and analysis is provided. Second, a set of benchmark results for dialogue state tracking, conversational recommendation, response generation as well as a unified model for multiple tasks are reported. We adopt the state-of-the-art methods for these tasks respectively to demonstrate the usability of the data, discuss limitations of current methods and set baselines for future studies.
Lizi Liao, Le Hong Long, Zheng Zhang 0020, Minlie Huang, Tat-Seng Chua
SIGIR4
2021 Overview of the Eighth Dialog System Technology Challenge: DSTC8
abstract
This paper introduces the Eighth Dialog System Technology Challenge. In line with recent challenges, the eighth edition focuses on applying end-to-end dialog technologies in a pragmatic way for multi-domain task-completion, noetic response selection, audio visual scene-aware dialog, and schema-guided dialog state tracking tasks. This paper describes the task definition, provided datasets, baselines and evaluation set-up for each track. We also summarize the results of the submitted systems to highlight the overall trends of the state-of-the-art technologies for the tasks.
Seokhwan Kim, Michel Galley, R. Chulaka Gunasekara, Adam Atkinson, Baolin Peng, Hannes Schulz, Jianfeng Gao 0001, Jinchao Li, Mahmoud Adada, Minlie Huang, Luis A. Lastras, Jonathan K. Kummerfeld, Walter S. Lasecki, Chiori Hori, Anoop Cherian, Tim K. Marks, Abhinav Rastogi, Xiaoxue Zang, Srinivas Sunkara
IEEE ACM Trans. Audio Speech Lang. Process.11
2020 A Pre-Training Based Personalized Dialogue Generation Model with Persona-Sparse Data
abstract
Endowing dialogue systems with personas is essential to deliver more human-like conversations. However, this problem is still far from well explored due to the difficulties of both embodying personalities in natural languages and the persona sparsity issue observed in most dialogue corpora. This paper proposes a pre-training based personalized dialogue model that can generate coherent responses using persona-sparse dialogue data. In this method, a pre-trained language model is used to initialize an encoder and decoder, and personal attribute embeddings are devised to model richer dialogue contexts by encoding speakers' personas together with dialogue histories. Further, to incorporate the target persona in the decoding process and to balance its contribution, an attention routing structure is devised in the decoder to merge features extracted from the target persona and dialogue contexts using dynamically predicted weights. Our model can utilize persona-sparse dialogues in a unified manner during the training process, and can also control the amount of persona-related features to exhibit during the inference process. Both automatic and manual evaluation demonstrates that the proposed model outperforms state-of-the-art methods for generating more coherent and persona consistent responses with persona-sparse data.
Yinhe Zheng, Minlie Huang, Xiaoxi Mao
AAAI3
2020 A Self-Training Method for Machine Reading Comprehension with Soft Evidence Extraction
abstract
Neural models have achieved great success on machine reading comprehension (MRC), many of which typically consist of two components: an evidence extractor and an answer predictor.The former seeks the most relevant information from a reference text, while the latter is to locate or generate answers from the extracted evidence.Despite the importance of evidence labels for training the evidence extractor, they are not cheaply accessible, particularly in many non-extractive MRC tasks such as YES/NO question answering and multi-choice MRC.To address this problem, we present a Self-Training method (STM), which supervises the evidence extractor with auto-generated evidence labels in an iterative process.At each iteration, a base MRC model is trained with golden answers and noisy evidence labels.The trained model will predict pseudo evidence labels as extra supervision in the next iteration.We evaluate STM on seven datasets over three MRC tasks.Experimental results demonstrate the improvement on existing MRC models, and we also analyze how and why such a self-training method works in MRC.
Yilin Niu, Fangkai Jiao, Mantong Zhou, Jingfang Xu, Minlie Huang
ACL6
2020 Multi-Agent Task-Oriented Dialog Policy Learning with Role-Aware Reward Decomposition
abstract
Many studies have applied reinforcement learning to train a dialog policy and show great promise these years.One common approach is to employ a user simulator to obtain a large number of simulated user experiences for reinforcement learning algorithms.However, modeling a realistic user simulator is challenging.A rule-based simulator requires heavy domain expertise for complex tasks, and a data-driven simulator requires considerable data and it is even unclear how to evaluate a simulator.To avoid explicitly building a user simulator beforehand, we propose Multi-Agent Dialog Policy Learning, which regards both the system and the user as the dialog agents.Two agents interact with each other and are jointly learned simultaneously.The method uses the actorcritic framework to facilitate pretraining and improve scalability.We also propose Hybrid Value Network for the role-aware reward decomposition to integrate role-specific domain knowledge of each agent in task-oriented dialog.Results show that our method can successfully build a system policy and a user policy simultaneously, and two agents can achieve a high task success rate through conversational interaction.
Ryuichi Takanobu, Runze Liang, Minlie Huang
ACL3
2020 KdConv: A Chinese Multi-domain Dialogue Dataset Towards Multi-turn Knowledge-driven Conversation
abstract
The research of knowledge-driven conversational systems is largely limited due to the lack of dialog data which consists of multi-turn conversations on multiple topics and with knowledge annotations. In this paper, we propose a Chinese multi-domain knowledge-driven conversation dataset, KdConv, which grounds the topics in multi-turn conversations to knowledge graphs. Our corpus contains 4.5K conversations from three domains (film, music, and travel), and 86K utterances with an average turn number of 19.0. These conversations contain in-depth discussions on related topics and natural transition between multiple topics. To facilitate the following research on this corpus, we provide several benchmark models. Comparative results show that the models can be enhanced by introducing background knowledge, yet there is still a large space for leveraging knowledge to model multi-turn conversations for further research. Results also show that there are obvious performance differences between different domains, indicating that it is worth further explore transfer learning and domain adaptation. The corpus and benchmark models are publicly available.
Hao Zhou 0012, Chujie Zheng, Kaili Huang, Minlie Huang, Xiaoyan Zhu 0001
ACL4
2020 UNION: An Unreferenced Metric for Evaluating Open-ended Story Generation
abstract
Despite the success of existing referenced metrics (e.g., BLEU and MoverScore), they correlate poorly with human judgments for openended text generation including story or dialog generation because of the notorious oneto-many issue: there are many plausible outputs for the same input, which may differ substantially in literal or semantics from the limited number of given references.To alleviate this issue, we propose UNION, a learnable UNreferenced metrIc for evaluating Open-eNded story generation, which measures the quality of a generated story without any reference.Built on top of BERT, UNION is trained to distinguish human-written stories from negative samples and recover the perturbation in negative stories.We propose an approach of constructing negative samples by mimicking the errors commonly observed in existing NLG models, including repeated plots, conflicting logic, and long-range incoherence.Experiments on two story datasets demonstrate that UNION is a reliable measure for evaluating the quality of generated stories, which correlates better with human judgments and is more generalizable than existing state-of-theart metrics.
Jian Guan 0002, Minlie Huang
EMNLP (1)2
2020 Language Generation with Multi-Hop Reasoning on Commonsense Knowledge Graph
abstract
Despite the success of generative pre-trained language models on a series of text generation tasks, they still suffer in cases where reasoning over underlying commonsense knowledge is required during generation.Existing approaches that integrate commonsense knowledge into generative pre-trained language models simply transfer relational knowledge by post-training on individual knowledge triples while ignoring rich connections within the knowledge graph.We argue that exploiting both the structural and semantic information of the knowledge graph facilitates commonsenseaware text generation.In this paper, we propose Generation with Multi-Hop Reasoning Flow (GRF) that enables pre-trained models with dynamic multi-hop reasoning on multirelational paths extracted from the external commonsense knowledge graph.We empirically show that our model outperforms existing baselines on three text generation tasks that require reasoning over commonsense knowledge.We also demonstrate the effectiveness of the dynamic multi-hop reasoning module with reasoning paths inferred by the model that provide rationale to the generation. 1
Haozhe Ji, Pei Ke, Shaohan Huang, Furu Wei, Xiaoyan Zhu 0001, Minlie Huang
EMNLP (1)6
2020 SentiLARE: Sentiment-Aware Language Representation Learning with Linguistic Knowledge
abstract
Most of the existing pre-trained language representation models neglect to consider the linguistic knowledge of texts, which can promote language understanding in NLP tasks.To benefit the downstream tasks in sentiment analysis, we propose a novel language representation model called SentiLARE, which introduces word-level linguistic knowledge including part-of-speech tag and sentiment polarity (inferred from SentiWordNet) into pretrained models.We first propose a contextaware sentiment attention mechanism to acquire the sentiment polarity of each word with its part-of-speech tag by querying SentiWord-Net.Then, we devise a new pre-training task called label-aware masked language model to construct knowledge-aware language representation.Experiments show that SentiLARE obtains new state-of-the-art performance on a variety of sentiment analysis tasks 1 .
Pei Ke, Haozhe Ji, Siyang Liu 0003, Xiaoyan Zhu 0001, Minlie Huang
EMNLP (1)5
2020 Dialogue Distillation: Open-Domain Dialogue Augmentation Using Unpaired Data
abstract
Recent advances in open-domain dialogue systems rely on the success of neural models that are trained on large-scale data.However, collecting large-scale dialogue data is usually time-consuming and labor-intensive.To address this data dilemma, we propose a novel data augmentation method for training opendomain dialogue models by utilizing unpaired data.Specifically, a data-level distillation process is first proposed to construct augmented dialogues where both post and response are retrieved from the unpaired data.A ranking module is employed to filter out low-quality dialogues.Further, a model-level distillation process is employed to distill a teacher model trained on high-quality paired data to augmented dialogue pairs, thereby preventing dialogue models from being affected by the noise in the augmented data.Automatic and manual evaluation indicates that our method can produce high-quality dialogue pairs with diverse contents, and the proposed data-level and model-level dialogue distillation can improve the performance of competitive baselines.
Yinhe Zheng, Jianzhi Shao, Xiaoxi Mao, Yadong Xi, Minlie Huang
EMNLP (1)6
2020 Robustness Verification for Transformers
Zhouxing Shi, Huan Zhang 0001, Kai-Wei Chang 0001, Minlie Huang, Cho-Jui Hsieh
ICLR4
2020 Reinforced Molecular Optimization with Neighborhood-Controlled Grammars
abstract
A major challenge in the pharmaceutical industry is to design novel molecules with specific desired properties, especially when the property evaluation is costly. Here, we propose MNCE-RL, a graph convolutional policy network for molecular optimization with molecular neighborhood-controlled embedding grammars through reinforcement learning. We extend the original neighborhood-controlled embedding grammars to make them applicable to molecular graph generation and design an efficient algorithm to infer grammatical production rules from given molecules. The use of grammars guarantees the validity of the generated molecular structures. By transforming molecular graphs to parse trees with the inferred grammars, the molecular structure generation task is modeled as a Markov decision process where a policy gradient strategy is utilized. In a series of experiments, we demonstrate that our approach achieves state-of-the-art performance in a diverse range of molecular optimization tasks and exhibits significant superiority in optimizing molecular properties with a limited number of property evaluations.
Chencheng Xu, Qiao Liu 0008, Minlie Huang, Tao Jiang 0001
NeurIPS3
2020 Automatic Perturbation Analysis for Scalable Certified Robustness and Beyond
abstract
Linear relaxation based perturbation analysis (LiRPA) for neural networks, which computes provable linear bounds of output neurons given a certain amount of input perturbation, has become a core component in robustness verification and certified defense. The majority of LiRPA-based methods focus on simple feed-forward networks and need particular manual derivations and implementations when extended to other architectures. In this paper, we develop an automatic framework to enable perturbation analysis on any neural network structures, by generalizing existing LiRPA algorithms such as CROWN to operate on general computational graphs. The flexibility, differentiability and ease of use of our framework allow us to obtain state-of-the-art results on LiRPA based certified defense on fairly complicated networks like DenseNet, ResNeXt and Transformer that are not supported by prior works. Our framework also enables loss fusion, a technique that significantly reduces the computational complexity of LiRPA for certified defense. For the first time, we demonstrate LiRPA based certified defense on Tiny ImageNet and Downscaled ImageNet where previous approaches cannot scale to due to the relatively large number of classes. Our work also yields an open-source library for the community to apply LiRPA to areas beyond certified defense without much LiRPA expertise, e.g., we create a neural network with a provably flat optimization landscape by applying LiRPA to network parameters. Our open source library is available at https://github.com/KaidiXu/auto_LiRPA.
Kaidi Xu, Zhouxing Shi, Huan Zhang 0001, Kai-Wei Chang 0001, Minlie Huang, Bhavya Kailkhura, Xue Lin 0001, Cho-Jui Hsieh
NeurIPS6
2020 A Large-Scale Chinese Short-Text Conversation Dataset
Yida Wang 0009, Pei Ke, Yinhe Zheng, Kaili Huang, Yong Jiang 0001, Xiaoyan Zhu 0001, Minlie Huang
NLPCC (1)7
2020 Is Your Goal-Oriented Dialog Model Performing Really Well? Empirical Analysis of System-wise Evaluation
abstract
There is a growing interest in developing goaloriented dialog systems which serve users in accomplishing complex tasks through multiturn conversations.Although many methods are devised to evaluate and improve the performance of individual dialog components, there is a lack of comprehensive empirical study on how different components contribute to the overall performance of a dialog system.In this paper, we perform a system-wise evaluation and present an empirical analysis on different types of dialog systems which are composed of different modules in different settings.Our results show that (1) a pipeline dialog system trained using fine-grained supervision signals at different component levels often obtains better performance than the systems that use joint or end-to-end models trained on coarse-grained labels, (2) component-wise, single-turn evaluation results are not always consistent with the overall performance of a dialog system, and (3) despite the discrepancy between simulators and human users, simulated evaluation is still a valid alternative to the costly human evaluation especially in the early stage of development.
Ryuichi Takanobu, Qi Zhu 0007, Jinchao Li, Baolin Peng, Jianfeng Gao 0001, Minlie Huang
SIGdial6
2020 A Knowledge-Enhanced Pretraining Model for Commonsense Story Generation
abstract
Story generation, namely, generating a reasonable story from a leading context, is an important but challenging task. In spite of the success in modeling fluency and local coherence, existing neural language generation models (e.g., GPT-2) still suffer from repetition, logic conflicts, and lack of long-range coherence in generated stories. We conjecture that this is because of the difficulty of associating relevant commonsense knowledge, understanding the causal relationships, and planning entities and events with proper temporal order. In this paper, we devise a knowledge-enhanced pretraining model for commonsense story generation. We propose to utilize commonsense knowledge from external knowledge bases to generate reasonable stories. To further capture the causal and temporal dependencies between the sentences in a reasonable story, we use multi-task learning, which combines a discriminative objective to distinguish true and fake stories during fine-tuning. Automatic and manual evaluation shows that our model can generate more reasonable stories than state-of-the-art baselines, particularly in terms of logic and global coherence.
Jian Guan 0002, Fei Huang 0005, Minlie Huang, Xiaoyan Zhu 0001
Trans. Assoc. Comput. Linguistics3
2020 CrossWOZ: A Large-Scale Chinese Cross-Domain Task-Oriented Dialogue Dataset
abstract
To advance multi-domain (cross-domain) dialogue modeling as well as alleviate the shortage of Chinese task-oriented datasets, we propose CrossWOZ, the first large-scale Chinese Cross-Domain Wizard-of-Oz task-oriented dataset. It contains 6K dialogue sessions and 102K utterances for 5 domains, including hotel, restaurant, attraction, metro, and taxi. Moreover, the corpus contains rich annotation of dialogue states and dialogue acts on both user and system sides. About 60% of the dialogues have cross-domain user goals that favor inter-domain dependency and encourage natural transition across domains in conversation. We also provide a user simulator and several benchmark models for pipelined task-oriented dialogue systems, which will facilitate researchers to compare and evaluate their models on this corpus. The large size and rich annotation of CrossWOZ make it suitable to investigate a variety of tasks in cross-domain dialogue modeling, such as dialogue state tracking, policy learning, user simulation, etc.
Qi Zhu 0007, Kaili Huang, Zheng Zhang 0020, Xiaoyan Zhu 0001, Minlie Huang
Trans. Assoc. Comput. Linguistics5
2020 Out-of-Domain Detection for Natural Language Understanding in Dialog Systems
abstract
Natural Language Understanding (NLU) is a vital component of dialogue systems, and its ability to detect Out-of-Domain (OOD) inputs is critical in practical applications, since the acceptance of the OOD input that is unsupported by the current system may lead to catastrophic failure. However, most existing OOD detection methods rely heavily on manually labeled OOD samples and cannot take full advantage of unlabeled data. This limits the feasibility of these models in practical applications. In this paper, we propose a novel model to generate high-quality pseudo OOD samples that are akin to IN-Domain (IND) input utterances and thereby improves the performance of OOD detection. To this end, an autoencoder is trained to map an input utterance into a latent code. Moreover, the codes of IND and OOD samples are trained to be indistinguishable by utilizing a generative adversarial network. To provide more supervision signals, an auxiliary classifier is introduced to regularize the generated OOD samples to have indistinguishable intent labels. Experiments show that these pseudo OOD samples generated by our model can be used to effectively improve OOD detection in NLU. Besides, we also demonstrate that the effectiveness of these pseudo OOD data can be further improved by efficiently utilizing unlabeled data.
Yinhe Zheng, Guanyi Chen, Minlie Huang
IEEE ACM Trans. Audio Speech Lang. Process.3
2020 Robust Reading Comprehension With Linguistic Constraints via Posterior Regularization
abstract
In spite of the great advancements of machine reading comprehension (RC), existing RC models are still vulnerable and not robust to different types of adversarial examples. Neural models over-confidently predict wrong answers to semantic different adversarial examples, while over-sensitively predict wrong answers to semantic equivalent adversarial examples. Existing methods which improve the robustness of such neural models merely mitigate one of the two issues but ignore the other. In this article, we address the over-confidence issue and the over-sensitivity issue existing in current RC models simultaneously with the help of external linguistic knowledge. We first incorporate external knowledge to impose different linguistic constraints (entity constraint, lexical constraint, and predicate constraint), and then regularize RC models through posterior regularization. Linguistic constraints induce more reasonable predictions for both semantic different and semantic equivalent adversarial examples, and posterior regularization provides an effective mechanism to incorporate these constraints. Our method can be applied to any existing neural RC models including state-of-the-art BERT models. Extensive experiments show that our method remarkably improves the robustness of base RC models, and is better to cope with these two issues simultaneously.
Mantong Zhou, Minlie Huang, Xiaoyan Zhu 0001
IEEE ACM Trans. Audio Speech Lang. Process.2
2020 Challenges in Building Intelligent Open-domain Dialog Systems
abstract
There is a resurgent interest in developing intelligent open-domain dialog systems due to the availability of large amounts of conversational data and the recent progress on neural approaches to conversational AI [33]. Unlike traditional task-oriented bots, an open-domain dialog system aims to establish long-term connections with users by satisfying the human need for communication, affection, and social belonging. This article reviews the recent work on neural approaches that are devoted to addressing three challenges in developing such systems: semantics , consistency , and interactiveness . Semantics requires a dialog system to not only understand the content of the dialog but also identify users’ emotional and social needs during the conversation. Consistency requires the system to demonstrate a consistent personality to win users’ trust and gain their long-term confidence. Interactiveness refers to the system’s ability to generate interpersonal responses to achieve particular social goals such as entertainment and conforming. The studies we select to present in this survey are based on our unique views and are by no means complete. Nevertheless, we hope that the discussion will inspire new research in developing more intelligent open-domain dialog systems.
Minlie Huang, Xiaoyan Zhu 0001, Jianfeng Gao 0001
ACM Trans. Inf. Syst.1
2019 Story Ending Generation with Incremental Encoding and Commonsense Knowledge
abstract
Generating a reasonable ending for a given story context, i.e., story ending generation, is a strong indication of story comprehension. This task requires not only to understand the context clues which play an important role in planning the plot, but also to handle implicit knowledge to make a reasonable, coherent story. In this paper, we devise a novel model for story ending generation. The model adopts an incremental encoding scheme to represent context clues which are spanning in the story context. In addition, commonsense knowledge is applied through multi-source attention to facilitate story comprehension, and thus to help generate coherent and reasonable endings. Through building context clues and using implicit knowledge, the model is able to produce reasonable story endings. Automatic and manual evaluation shows that our model can generate more reasonable story endings than state-of-the-art baselines1.
Jian Guan 0002, Yansen Wang, Minlie Huang
AAAI3
2019 A Deep Sequential Model for Discourse Parsing on Multi-Party Dialogues
abstract
Discourse structures are beneficial for various NLP tasks such as dialogue understanding, question answering, sentiment analysis, and so on. This paper presents a deep sequential model for parsing discourse dependency structures of multi-party dialogues. The proposed model aims to construct a discourse dependency tree by predicting dependency relations and constructing the discourse structure jointly and alternately. It makes a sequential scan of the Elementary DiscourseUnits(EDUs)1 in a dialogue. For each EDU, the model decides to which previous EDU the current one should link and what the corresponding relation type is. The predicted link and relation type are then used to build the discourse structure incrementally with a structured encoder. During link prediction and relation classification, the model utilizes not only local information that represents the concerned EDUs, but also global information that encodes the EDU sequence and the discourse structure that is already built at the current step. Experiments show that the proposed model outperforms all the state-of-the-art baselines.
Zhouxing Shi, Minlie Huang
AAAI2
2019 A Hierarchical Framework for Relation Extraction with Reinforcement Learning
abstract
Most existing methods determine relation types only after all the entities have been recognized, thus the interaction between relation types and entity mentions is not fully modeled. This paper presents a novel paradigm to deal with relation extraction by regarding the related entities as the arguments of a relation. We apply a hierarchical reinforcement learning (HRL) framework in this paradigm to enhance the interaction between entity mentions and relation types. The whole extraction process is decomposed into a hierarchy of two-level RL policies for relation detection and entity extraction respectively, so that it is more feasible and natural to deal with overlapping relations. Our model was evaluated on public datasets collected via distant supervision, and results show that it gains better performance than existing methods and is more powerful for extracting overlapping relations1.
Ryuichi Takanobu, Jiexi Liu 0002, Minlie Huang
AAAI4
2019 ChID: A Large-scale Chinese IDiom Dataset for Cloze Test
abstract
Cloze-style reading comprehension in Chinese is still limited due to the lack of various corpora.In this paper we propose a large-scale Chinese cloze test dataset ChID, which studies the comprehension of idiom, a unique language phenomenon in Chinese.In this corpus, the idioms in a passage are replaced by blank symbols and the correct answer needs to be chosen from well-designed candidate idioms.We carefully study how the design of candidate idioms and the representation of idioms affect the performance of state-of-the-art models.Results show that the machine accuracy is substantially worse than that of human, indicating a large space for further research.
Chujie Zheng, Minlie Huang, Aixin Sun
ACL (1)2
2019 ARAML: A Stable Adversarial Training Framework for Text Generation
abstract
Pei Ke, Fei Huang, Minlie Huang, Xiaoyan Zhu. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.
Pei Ke, Fei Huang 0005, Minlie Huang, Xiaoyan Zhu 0001
EMNLP/IJCNLP (1)3
2019 Long and Diverse Text Generation with Planning-based Hierarchical Variational Model
abstract
Zhihong Shao, Minlie Huang, Jiangtao Wen, Wenfei Xu, Xiaoyan Zhu. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.
Zhihong Shao, Minlie Huang, Jiangtao Wen, Wenfei Xu, Xiaoyan Zhu 0001
EMNLP/IJCNLP (1)2
2019 Guided Dialog Policy Learning: Reward Estimation for Multi-Domain Task-Oriented Dialog
abstract
Ryuichi Takanobu, Hanlin Zhu, Minlie Huang. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.
Ryuichi Takanobu, Hanlin Zhu, Minlie Huang
EMNLP/IJCNLP (1)3
2019 Meta-Learning for Low-resource Natural Language Generation in Task-oriented Dialogue Systems
abstract
Natural language generation (NLG) is an essential component of task-oriented dialogue systems. Despite the recent success of neural approaches for NLG, they are typically developed for particular domains with rich annotated training examples. In this paper, we study NLG in a low-resource setting to generate sentences in new scenarios with handful training examples. We formulate the problem from a meta-learning perspective, and propose a generalized optimization-based approach (Meta-NLG) based on the well-recognized model-agnostic meta-learning (MAML) algorithm. Meta-NLG defines a set of meta tasks, and directly incorporates the objective of adapting to new low-resource NLG tasks into the meta-learning optimization process. Extensive experiments are conducted on a large multi-domain dataset (MultiWoz) with diverse linguistic variations. We show that Meta-NLG significantly outperforms other training procedures in various low-resource configurations. We analyze the results, and demonstrate that Meta-NLG adapts extremely fast and well to low-resource situations.
Fei Mi, Minlie Huang, Jiyong Zhang 0001, Boi Faltings
IJCAI2
2019 User Attention-guided Multimodal Dialog Systems
abstract
As an intelligent way to interact with computers, the dialog system has been catching more and more attention. However, most research efforts only focus on text-based dialog systems, completely ignoring the rich semantics conveyed by the visual cues. Indeed, the desire for multimodal task-oriented dialog systems is growing with the rapid expansion of many domains, such as the online retailing and travel. Besides, few work considers the hierarchical product taxonomy and the users' attention to products explicitly. The fact is that users tend to express their attention to the semantic attributes of products such as color and style as the dialog goes on. Towards this end, in this work, we present a hierarchical User attention-guided Multimodal Dialog system, named UMD for short. UMD leverages a bidirectional Recurrent Neural Network to model the ongoing dialog between users and chatbots at a high level; As to the low level, the multimodal encoder and decoder are capable of encoding multimodal utterances and generating multimodal responses, respectively. The multimodal encoder learns the visual presentation of images with the help of a taxonomy-attribute combined tree, and then the visual features interact with textual features through an attention mechanism; whereas the multimodal decoder selects the required visual images and generates textual responses according to the dialog history. To evaluate our proposed model, we conduct extensive experiments on a public multimodal dialog dataset in the retailing domain. Experimental results demonstrate that our model outperforms the existing state-of-the-art methods by integrating the multimodal utterances and encoding the visual features based on the users' attribute-level attention.
Wenjie Wang 0007, Xuemeng Song, Minlie Huang, Xin-Shun Xu, Liqiang Nie
SIGIR4
2019 Aggregating E-commerce Search Results from Heterogeneous Sources via Hierarchical Reinforcement Learning
abstract
In this paper, we investigate the task of aggregating search results from heterogeneous sources in an E-commerce environment. First, unlike traditional aggregated web search that merely presents multi-sourced results in the first page, this new task may present aggregated results in all pages and has to dynamically decide which source should be presented in the current page. Second, as pointed out by many existing studies, it is not trivial to rank items from heterogeneous sources because the relevance scores from different source systems are not directly comparable. To address these two issues, we decompose the task into two subtasks in a hierarchical structure: a high-level task for source selection where we model the sequential patterns of user behaviors onto aggregated results in different pages so as to understand user intents and select the relevant sources properly; and a low-level task for item presentation where we formulate a slot filling process to sequentially present the items instead of giving each item a relevance score when deciding the presentation order of heterogeneous items. Since both subtasks can be naturally formulated as sequential decision problems and learn from the future user feedback on search results, we build our model with hierarchical reinforcement learning. Extensive experiments demonstrate that our model obtains remarkable improvements in search performance metrics, and achieves a higher user satisfaction.
Ryuichi Takanobu, Minlie Huang, Haihong Tang, Bo Zheng 0007
WWW3
2019 Aspect-level Sentiment Analysis using AS-Capsules
abstract
Aspect-level sentiment analysis aims to provide complete and detailed view of sentiment analysis from different aspects. Existing solutions usually adopt a two-staged approach: first detecting aspect category in a document, then categorizing the polarity of opinion expressions for detected aspect(s). Inevitably, such methods lead to error accumulation. Moreover, aspect detection and aspect-level sentiment classification are highly correlated with each other. The key issue here is how to perform aspect detection and aspect-level sentiment classification jointly, and effectively. In this paper, we propose the aspect-level sentiment capsules model (AS-Capsules), which is capable of performing aspect detection and sentiment classification simultaneously, in a joint manner. AS-Capsules utilizes the correlation between aspect and sentiment through shared components including capsule embedding, shared encoders, and shared attentions. AS-Capsules is also capable of communicating with different capsules through a shared Recurrent Neural Network (RNN). More importantly, AS-Capsules model does not require any linguistic knowledge as additional input. Instead, through the attention mechanism, this model is able to attend aspect related words and sentiment words corresponding to different aspect(s). Experiments show that the AS-Capsules model achieves state-of-the-art performances on a benchmark dataset for aspect-level sentiment analysis.
Yequan Wang, Aixin Sun, Minlie Huang, Xiaoyan Zhu 0001
WWW3
2019 Neural Multimodal Belief Tracker with Adaptive Attention for Dialogue Systems
abstract
Multimodal dialogue systems are attracting increasing attention with a more natural and informative way for human-computer interaction. As one of its core components, the belief tracker estimates the user's goal at each step of the dialogue and provides a direct way to validate the ability of dialogue understanding. However, existing studies on belief trackers are largely limited to textual modality, which cannot be easily extended to capture the rich semantics in multimodal systems such as those with product images. For example, in fashion domain, the visual appearance of clothes play a crucial role in understanding the user's intention. In this case, the existing belief trackers may fail to generate accurate belief states for a multimodal dialogue system.
Zheng Zhang 0020, Lizi Liao, Minlie Huang, Xiaoyan Zhu 0001, Tat-Seng Chua
WWW3
2019 Domain-Constrained Advertising Keyword Generation
abstract
Advertising (ad for short) keyword suggestion is important for sponsored search to improve online advertising and increase search revenue. There are two common challenges in this task. First, the keyword bidding problem: hot ad keywords are very expensive for most of the advertisers because more advertisers are bidding on more popular keywords, while unpopular keywords are difficult to discover. As a result, most ads have few chances to be presented to the users. Second, the inefficient ad impression issue: a large proportion of search queries, which are unpopular yet relevant to many ad keywords, have no ads presented on their search result pages. Existing retrieval-based or matching-based methods either deteriorate the bidding competition or are unable to suggest novel keywords to cover more queries, which leads to inefficient ad impressions.
Hao Zhou 0012, Minlie Huang, Yishun Mao, Changlei Zhu, Peng Shu, Xiaoyan Zhu 0001
WWW2
2019 Story Ending Selection by Finding Hints From Pairwise Candidate Endings
abstract
The ability of story comprehension is a strong indicator of natural language understanding. Recently, Story Cloze Test has been introduced as a new task of machine reading comprehension, i.e., selecting a correct ending from two candidate endings given a four-sentence story context. Most existing methods for Story Cloze Test are essentially matching-based that operate by comparing an individual ending with a given context, therefore suffering from the evidence bias issue: both candidate endings can obtain supporting evidence from the story context, which misleads the classifier to choose an incorrect ending. To address this issue, we present a novel idea to improve story comprehension by utilizing the hints that are obtained through comparing two candidate endings. The proposed model firstly anticipates a feature vector for a possible ending solely based on the context, and then refines the feature prediction using the hints which encode the difference between two candidates. The candidate ending whose feature vector is more similar to the predicted ending vector is regarded as correct. Experimental results demonstrate that our approach can alleviate the evidence bias issue and improve story comprehension.
Mantong Zhou, Minlie Huang, Xiaoyan Zhu 0001
IEEE ACM Trans. Audio Speech Lang. Process.2
2019 Memory-Augmented Dialogue Management for Task-Oriented Dialogue Systems
abstract
Dialogue management (DM) is responsible for predicting the next action of a dialogue system according to the current dialogue state and thus plays a central role in task-oriented dialogue systems. Since DM requires having access not only to local utterances but also to the global semantics of the entire dialogue session, modeling the long-range history information is a critical issue. To this end, we propose MAD, a novel memory-augmented dialogue management model that employs a memory controller and two additional memory structures (i.e., a slot-value memory and an external memory). The slot-value memory tracks the dialogue state by memorizing and updating the values of semantic slots (i.e., cuisine, price, and location), and the external memory augments the representation of hidden states of traditional recurrent neural networks by storing more context information. To update the dialogue state efficiently, we also propose slot-level attention on user utterances to extract specific semantic information for each slot. Experiments show that our model can obtain state-of-the-art performance and outperforms existing baselines.
Zheng Zhang 0020, Minlie Huang, Zhongzhou Zhao, Haiqing Chen, Xiaoyan Zhu 0001
ACM Trans. Inf. Syst.2
2018 Reinforcement Learning for Relation Classification From Noisy Data
abstract
Existing relation classification methods that rely on distant supervision assume that a bag of sentences mentioning an entity pair are all describing a relation for the entity pair. Such methods, performing classification at the bag level, cannot identify the mapping between a relation and a sentence, and largely suffers from the noisy labeling problem. In this paper, we propose a novel model for relation classification at the sentence level from noisy data. The model has two modules: an instance selector and a relation classifier. The instance selector chooses high-quality sentences with reinforcement learning and feeds the selected sentences into the relation classifier, and the relation classifier makes sentence-level prediction and provides rewards to the instance selector. The two modules are trained jointly to optimize the instance selection and relation classification processes.Experiment results show that our model can deal with the noise of data effectively and obtains better performance for relation classification at the sentence level.
Minlie Huang, Li Zhao 0007, Yang Yang 0009, Xiaoyan Zhu 0001
AAAI2
2018 Augmenting End-to-End Dialogue Systems With Commonsense Knowledge
abstract
Building dialogue systems that can converse naturally with humans is a challenging yet intriguing problem of artificial intelligence. In open-domain human-computer conversation, where the conversational agent is expected to respond to human utterances in an interesting and engaging way, commonsense knowledge has to be integrated into the model effectively. In this paper, we investigate the impact of providing commonsense knowledge about the concepts covered in the dialogue. Our model represents the first attempt to integrating a large commonsense knowledge base into end-to-end conversational models. In the retrieval-based scenario, we propose a model to jointly take into account message content and related commonsense for selecting an appropriate response. Our experiments suggest that the knowledge-augmented models are superior to their knowledge-free counterparts.
Tom Young, Erik Cambria, Iti Chaturvedi, Hao Zhou 0012, Subham Biswas, Minlie Huang
AAAI6
2018 Learning Structured Representation for Text Classification via Reinforcement Learning
abstract
Representation learning is a fundamental problem in natural language processing. This paper studies how to learn a structured representation for text classification. Unlike most existing representation models that either use no structure or rely on pre-specified structures, we propose a reinforcement learning (RL) method to learn sentence representation by discovering optimized structures automatically. We demonstrate two attempts to build structured representation: Information Distilled LSTM (ID-LSTM) and Hierarchically Structured LSTM (HS-LSTM). ID-LSTM selects only important, task-relevant words, and HS-LSTM discovers phrase structures in a sentence. Structure discovery in the two representation models is formulated as a sequential decision problem: current decision of structure discovery affects following decisions, which can be addressed by policy gradient RL. Results show that our method can learn task-friendly representations by identifying important words or task-relevant structures without explicit structure annotations, and thus yields competitive performance.
Minlie Huang, Li Zhao 0007
AAAI2
2018 Emotional Chatting Machine: Emotional Conversation Generation with Internal and External Memory
abstract
Perception and expression of emotion are key factors to the success of dialogue systems or conversational agents. However, this problem has not been studied in large-scale conversation generation so far. In this paper, we propose Emotional Chatting Machine (ECM) that can generate appropriate responses not only in content (relevant and grammatical) but also in emotion (emotionally consistent). To the best of our knowledge, this is the first work that addresses the emotion factor in large-scale conversation generation. ECM addresses the factor using three new mechanisms that respectively (1) models the high-level abstraction of emotion expressions by embedding emotion categories, (2) captures the change of implicit internal emotion states, and (3) uses explicit emotion expressions with an external emotion vocabulary. Experiments show that the proposed model can generate responses appropriate not only in content but also in emotion.
Hao Zhou 0012, Minlie Huang, Xiaoyan Zhu 0001, Bing Liu 0001
AAAI2
2018 Generating Informative Responses with Controlled Sentence Function
abstract
Sentence function is a significant factor to achieve the purpose of the speaker, which, however, has not been touched in largescale conversation generation so far.In this paper, we present a model to generate informative responses with controlled sentence function.Our model utilizes a continuous latent variable to capture various word patterns that realize the expected sentence function, and introduces a type controller to deal with the compatibility of controlling sentence function and generating informative content.Conditioned on the latent variable, the type controller determines the type (i.e., function-related, topic, and ordinary word) of a word to be generated at each decoding position.Experiments show that our model outperforms state-of-the-art baselines, and it has the ability to generate responses with both controlled sentence function and informative content.
Pei Ke, Jian Guan 0002, Minlie Huang, Xiaoyan Zhu 0001
ACL (1)3
2018 Learning to Ask Questions in Open-domain Conversational Systems with Typed Decoders
abstract
Asking good questions in large-scale, open-domain conversational systems is quite significant yet rather untouched.This task, substantially different from traditional question generation, requires to question not only with various patterns but also on diverse and relevant topics.We observe that a good question is a natural composition of interrogatives, topic words, and ordinary words.Interrogatives lexicalize the pattern of questioning, topic words address the key information for topic transition in dialogue, and ordinary words play syntactical and grammatical roles in making a natural sentence.We devise two typed decoders (soft typed decoder and hard typed decoder) in which a type distribution over the three types is estimated and used to modulate the final generation distribution.Extensive experiments show that the typed decoders outperform state-of-the-art baselines and can generate more meaningful questions.
Yansen Wang, Chenyi Liu, Minlie Huang, Liqiang Nie
ACL (1)3
2018 An Operation Network for Abstractive Sentence Compression
abstract
Sentence compression condenses a sentence while preserving its most important contents. Delete-based models have the strong ability to delete undesired words, while generate-based models are able to reorder or rephrase the words, which are more coherent to human sentence compression. In this paper, we propose Operation Network, a neural network approach for abstractive sentence compression, which combines the advantages of both delete-based and generate-based sentence compression models. The central idea of Operation Network is to model the sentence compression process as an editing procedure. First, unnecessary words are deleted from the source sentence, then new words are either generated from a large vocabulary or copied directly from the source sentence. A compressed sentence can be obtained by a series of such edit operations (delete, copy and generate). Experiments show that Operation Network outperforms state-of-the-art baselines.
Naitong Yu, Jie Zhang 0002, Minlie Huang, Xiaoyan Zhu 0001
COLING3
2018 An Interpretable Reasoning Network for Multi-Relation Question Answering
abstract
Multi-relation Question Answering is a challenging task, due to the requirement of elaborated analysis on questions and reasoning over multiple fact triples in knowledge base. In this paper, we present a novel model called Interpretable Reasoning Network that employs an interpretable, hop-by-hop reasoning process for question answering. The model dynamically decides which part of an input question should be analyzed at each hop; predicts a relation that corresponds to the current parsed results; utilizes the predicted relation to update the question representation and the state of the reasoning process; and then drives the next-hop reasoning. Experiments show that our model yields state-of-the-art results on two datasets. More interestingly, the model can offer traceable and observable intermediate predictions for reasoning analysis and failure diagnosis, thereby allowing manual manipulation in predicting the final answer.
Mantong Zhou, Minlie Huang, Xiaoyan Zhu 0001
COLING2
2018 Assigning Personality/Profile to a Chatting Machine for Coherent Conversation Generation
abstract
Endowing a chatbot with personality is challenging but significant to deliver more realistic and natural conversations. In this paper, we address the issue of generating responses that are coherent to a pre-specified personality or profile. We present a method that uses generic conversation data from social media (without speaker identities) to generate profile-coherent responses. The central idea is to detect whether a profile should be used when responding to a user post (by a profile detector), and if necessary, select a key-value pair from the profile to generate a response forward and backward (by a bidirectional decoder) so that a personality-coherent response can be generated. Furthermore, in order to train the bidirectional decoder with generic dialogue data, a position detector is designed to predict a word position from which decoding should start given a profile value. Manual and automatic evaluation shows that our model can deliver more coherent, natural, and diversified responses.
Qiao Qian, Minlie Huang, Haizhou Zhao, Jingfang Xu, Xiaoyan Zhu 0001
IJCAI2
2018 A Weakly Supervised Method for Topic Segmentation and Labeling in Goal-oriented Dialogues via Reinforcement Learning
abstract
Topic structure analysis plays a pivotal role in dialogue understanding. We propose a reinforcement learning (RL) method for topic segmentation and labeling in goal-oriented dialogues, which aims to detect topic boundaries among dialogue utterances and assign topic labels to the utterances. We address three common issues in the goal-oriented customer service dialogues: informality, local topic continuity, and global topic structure. We explore the task in a weakly supervised setting and formulate it as a sequential decision problem. The proposed method consists of a state representation network to address the informality issue, and a policy network with rewards to model local topic continuity and global topic structure. To train the two networks and offer a warm-start to the policy, we firstly use some keywords to annotate the data automatically. We then pre-train the networks on noisy data. Henceforth, the method continues to refine the data labels using the current policy to learn better state representations on the refined data for obtaining a better policy. Results demonstrate that this weakly supervised method obtains substantial improvements over state-of-the-art baselines.
Ryuichi Takanobu, Minlie Huang, Zhongzhou Zhao, Feng-Lin Li, Haiqing Chen, Xiaoyan Zhu 0001, Liqiang Nie
IJCAI2
2018 Densely Connected CNN with Multi-scale Feature Attention for Text Classification
abstract
Text classification is a fundamental problem in natural language processing. As a popular deep learning model, convolutional neural network (CNN) has demonstrated great success in this task. However, most existing CNN models apply convolution filters of fixed window size, thereby unable to learn variable n-gram features flexibly. In this paper, we present a densely connected CNN with multi-scale feature attention for text classification. The dense connections build short-cut paths between upstream and downstream convolutional blocks, which enable the model to compose features of larger scale from those of smaller scale, and thus produce variable n-gram features. Furthermore, a multi-scale feature attention is developed to adaptively select multi-scale features for classification. Extensive experiments demonstrate that our model obtains competitive performance against state-of-the-art baselines on five benchmark datasets. Attention visualization further reveals the model's ability to select proper n-gram features for text classification.
Shiyao Wang 0001, Minlie Huang, Zhidong Deng
IJCAI2
2018 Commonsense Knowledge Aware Conversation Generation with Graph Attention
abstract
Commonsense knowledge is vital to many natural language processing tasks. In this paper, we present a novel open-domain conversation generation model to demonstrate how large-scale commonsense knowledge can facilitate language understanding and generation. Given a user post, the model retrieves relevant knowledge graphs from a knowledge base and then encodes the graphs with a static graph attention mechanism, which augments the semantic information of the post and thus supports better understanding of the post. Then, during word generation, the model attentively reads the retrieved knowledge graphs and the knowledge triples within each graph to facilitate better generation through a dynamic graph attention mechanism. This is the first attempt that uses large-scale commonsense knowledge in conversation generation. Furthermore, unlike existing models that use knowledge triples (entities) separately and independently, our model treats each knowledge graph as a whole, which encodes more structured, connected semantic information in the graphs. Experiments show that the proposed model can generate more appropriate and informative responses than state-of-the-art baselines.
Hao Zhou 0012, Tom Young, Minlie Huang, Haizhou Zhao, Jingfang Xu, Xiaoyan Zhu 0001
IJCAI3
2018 Chat More: Deepening and Widening the Chatting Topic via A Deep Model
abstract
The past decade has witnessed the boom of human-machine interactions, particularly via dialog systems. In this paper, we study the task of response generation in open-domain multi-turn dialog systems. Many research efforts have been dedicated to building intelligent dialog systems, yet few shed light on deepening or widening the chatting topics in a conversational session, which would attract users to talk more. To this end, this paper presents a novel deep scheme consisting of three channels, namely global, wide, and deep ones. The global channel encodes the complete historical information within the given context, the wide one employs an attention-based recurrent neural network model to predict the keywords that may not appear in the historical context, and the deep one trains a Multi-layer Perceptron model to select some keywords for an in-depth discussion. Thereafter, our scheme integrates the outputs of these three channels to generate desired responses. To justify our model, we conducted extensive experiments to compare our model with several state-of-the-art baselines on two datasets: one is constructed by ourselves and the other is a public benchmark dataset. Experimental results demonstrate that our model yields promising performance by widening or deepening the topics of interest.
Wenjie Wang 0007, Minlie Huang, Xin-Shun Xu, Fumin Shen, Liqiang Nie
SIGIR2
2018 Learning to Collaborate: Multi-Scenario Ranking via Multi-Agent Reinforcement Learning
abstract
Ranking is a fundamental and widely studied problem in scenarios such as search, advertising, and recommendation. However, joint optimization for multi-scenario ranking, which aims to improve the overall performance of several ranking strategies in different scenarios, is rather untouched. Separately optimizing each individual strategy has two limitations. The first one is lack of collaboration between scenarios meaning that each strategy maximizes its own objective but ignores the goals of other strategies, leading to a sub-optimal overall performance. The second limitation is the inability of modeling the correlation between scenarios meaning that independent optimization in one scenario only uses its own user data but ignores the context in other scenarios. In this paper, we formulate multi-scenario ranking as a fully cooperative, partially observable, multi-agent sequential decision problem. We propose a novel model named Multi-Agent Recurrent Deterministic Policy Gradient (MA-RDPG) which has a communication component for passing messages, several private actors (agents) for making actions for ranking, and a centralized critic for evaluating the overall performance of the co-working actors. Each scenario is treated as an agent (actor). Agents collaborate with each other by sharing a global action-value function (the critic) and passing messages that encodes historical information across scenarios. The model is evaluated with online settings on a large E-commerce platform. Results show that the proposed model exhibits significant improvements against baselines in terms of the overall performance.
Minlie Huang, Shichen Liu, Wenwu Ou, Zhirong Wang, Xiaoyan Zhu 0001
WWW3
2017 SSP: Semantic Space Projection for Knowledge Graph Embedding with Text Descriptions
abstract
Knowledge graph embedding represents entities and relations in knowledge graph as low-dimensional, continuous vectors, and thus enables knowledge graph compatible with machine learning models. Though there have been a variety of models for knowledge graph embedding, most methods merely concentrate on the fact triples, while supplementary textual descriptions of entities and relations have not been fully employed. To this end, this paper proposes the semantic space projection (SSP) model which jointly learns from the symbolic triples and textual descriptions. Our model builds interaction between the two information sources, and employs textual descriptions to discover semantic relevance and offer precise semantic embedding. Extensive experiments show that our method achieves substantial improvements against baselines on the tasks of knowledge graph completion and entity classification.
Han Xiao 0005, Minlie Huang, Lian Meng, Xiaoyan Zhu 0001
AAAI2
2017 Linguistically Regularized LSTM for Sentiment Classification
abstract
This paper deals with sentence-level sentiment classification.Though a variety of neural network models have been proposed recently, however, previous models either depend on expensive phrase-level annotation, most of which has remarkably degraded performance when trained with only sentence-level annotation; or do not fully employ linguistic resources (e.g., sentiment lexicons, negation words, intensity words).In this paper, we propose simple models trained with sentence-level annotation, but also attempt to model the linguistic role of sentiment lexicons, negation words, and intensity words.Results show that our models are able to capture the linguistic role of sentiment words, negation words, and intensity words in sentiment expression.
Qiao Qian, Minlie Huang, Jinhao Lei, Xiaoyan Zhu 0001
ACL (1)2
2017 Give me Something Unknown: Incorporate Exploration Preference in Cognition into Recommender System
abstract
Recommender systems (RSs) make recommendation for users by capturing their preference from historical behaviors. However, emphasizing too much on the perspective of prediction precision leads to the problem of preference overfitting, which blockades the user from touching a wide range of new but potentially interesting items. Usually, it would be beneficial for the system to attempt suitable extensions beyond users' existing preferences already expressed in the recommender system. Traditional recommender systems concentrate much more on what a user had explored than what she/he is yet to explore. Too much attention to the precision of recommendation can lead to a preference overfitting problem. Users have the need for certain amount of extension based on the existing preference. In this paper, we investigate how novel items should be recommended given the user's preference which is modeled from her/his historical behaviors. Inspired by Wundt Curve, a famous law in cognition psychology field, we propose an idea of Personal Exploration Preference (PEP) model to capture the user preference from another perspective, the strength of aspiration to explore. Then we integrate PEP model into traditional recommendation algorithms. Finally,we experimentally evaluate the improvement of PEP model to traditional algorithms on HetRec2011 dataset. Experiment results show that, on the Top-N recommendation task, PEP model can effectively capture user preference of exploration and improve recommendation quality significantly.
Daoyi Li, Minlie Huang, Xiaoyan Zhu 0001
ICTAI2
2017 Overview of the NLPCC 2017 Shared Task: Emotion Generation Challenge
Minlie Huang, Zuoxian Ye, Hao Zhou 0012
NLPCC1
2017 Dialogue Intent Classification with Long Short-Term Memory Networks
Lian Meng, Minlie Huang
NLPCC2
2017 Encoding Syntactic Knowledge in Neural Networks for Sentiment Classification
abstract
Phrase/Sentence representation is one of the most important problems in natural language processing. Many neural network models such as Convolutional Neural Network (CNN), Recursive Neural Network (RNN), and Long Short-Term Memory (LSTM) have been proposed to learn representations of phrase/sentence, however, rich syntactic knowledge has not been fully explored when composing a longer text from its shorter constituent words. In most traditional models, only word embeddings are utilized to compose phrase/sentence representations, while the syntactic information of words is yet to be explored. In this article, we discover that encoding syntactic knowledge (part-of-speech tag) in neural networks can enhance sentence/phrase representation. Specifically, we propose to learn tag-specific composition functions and tag embeddings in recursive neural networks, and propose to utilize POS tags to control the gates of tree-structured LSTM networks. We evaluate these models on two benchmark datasets for sentiment classification, and demonstrate that improvements can be obtained with such syntactic knowledge encoded.
Minlie Huang, Qiao Qian, Xiaoyan Zhu 0001
ACM Trans. Inf. Syst.1
2016 Semi-Supervised Multinomial Naive Bayes for Text Classification by Leveraging Word-Level Statistical Constraint
abstract
Multinomial Naive Bayes with Expectation Maximization (MNB-EM) is a standard semi-supervised learning method to augment Multinomial Naive Bayes (MNB) for text classification. Despite its success, MNB-EM is not stable, and may succeed or fail to improve MNB. We believe that this is because MNB-EM lacks the ability to preserve the class distribution on words. In this paper, we propose a novel method to augment MNB-EM by leveraging the word-level statistical constraint to preserve the class distribution on words. The word-level statistical constraints are further converted to constraints on document posteriors generated by MNB-EM. Experiments demonstrate that our method can consistently improve MNB-EM, and outperforms state-of-art baselines remarkably.
Li Zhao 0007, Minlie Huang, Ziyu Yao 0002, Rongwei Su, Yingying Jiang 0001, Xiaoyan Zhu 0001
AAAI2
2016 TransG : A Generative Model for Knowledge Graph Embedding
abstract
Recently, knowledge graph embedding, which projects symbolic entities and relations into continuous vector space, has become a new, hot topic in artificial intelligence.This paper proposes a novel generative model (TransG) to address the issue of multiple relation semantics that a relation may have multiple meanings revealed by the entity pairs associated with the corresponding triples.The new model can discover latent semantics for a relation and leverage a mixture of relationspecific component vectors to embed a fact triple.To the best of our knowledge, this is the first generative model for knowledge graph embedding, and at the first time, the issue of multiple relation semantics is formally discussed.Extensive experiments show that the proposed model achieves substantial improvements against the state-of-the-art baselines.
Han Xiao 0005, Minlie Huang, Xiaoyan Zhu 0001
ACL (1)2
2016 A Sentence Interaction Network for Modeling Dependence between Sentences
abstract
Modeling interactions between two sentences is crucial for a number of natural language processing tasks including Answer Selection, Dialogue Act Analysis, etc.While deep learning methods like Recurrent Neural Network or Convolutional Neural Network have been proved to be powerful for sentence modeling, prior studies paid less attention on interactions between sentences.In this work, we propose a Sentence Interaction Network (SIN) for modeling the complex interactions between two sentences.By introducing "interaction states" for word and phrase pairs, SIN is powerful and flexible in capturing sentence interactions for different tasks.We obtain significant improvements on Answer Selection and Dialogue Act Analysis without any feature engineering.
Minlie Huang
ACL (1)2
2016 GAKE: Graph Aware Knowledge Embedding
abstract
Knowledge embedding, which projects triples in a given knowledge base to d-dimensional vectors, has attracted considerable research efforts recently. Most existing approaches treat the given knowledge base as a set of triplets, each of whose representation is then learned separately. However, as a fact, triples are connected and depend on each other. In this paper, we propose a graph aware knowledge embedding method (GAKE), which formulates knowledge base as a directed graph, and learns representations for any vertices or edges by leveraging the graph’s structural information. We introduce three types of graph context for embedding: neighbor context, path context, and edge context, each reflects properties of knowledge from different perspectives. We also design an attention mechanism to learn representative power of different vertices or edges. To validate our method, we conduct several experiments on two tasks. Experimental results suggest that our method outperforms several state-of-art knowledge embedding models.
Minlie Huang, Yang Yang 0009, Xiaoyan Zhu 0001
COLING2
2016 Product Review Summarization by Exploiting Phrase Properties
abstract
We propose a phrase-based approach for generating product review summaries. The main idea of our method is to leverage phrase properties to choose a subset of optimal phrases for generating the final summary. Specifically, we exploit two phrase properties, popularity and specificity. Popularity describes how popular the phrase is in the original reviews. Specificity describes how descriptive a phrase is in comparison to generic comments. We formalize the phrase selection procedure as an optimization problem and solve it using integer linear programming (ILP). An aspect-based bigram language model is used for generating the final summary with the selected phrases. Experiments show that our summarizer outperforms the other baselines.
Naitong Yu, Minlie Huang, Xiaoyan Zhu 0001
COLING2
2016 Context-aware Natural Language Generation for Spoken Dialogue Systems
abstract
Natural language generation (NLG) is an important component of question answering(QA) systems which has a significant impact on system quality. Most tranditional QA systems based on templates or rules tend to generate rigid and stylised responses without the natural variation of human language. Furthermore, such methods need an amount of work to generate the templates or rules. To address this problem, we propose a Context-Aware LSTM model for NLG. The model is completely driven by data without manual designed templates or rules. In addition, the context information, including the question to be answered, semantic values to be addressed in the response, and the dialogue act type during interaction, are well approached in the neural network model, which enables the model to produce variant and informative responses. The quantitative evaluation and human evaluation show that CA-LSTM obtains state-of-the-art performance.
Hao Zhou 0012, Minlie Huang, Xiaoyan Zhu 0001
COLING2
2016 Attention-based LSTM for Aspect-level Sentiment Classification
abstract
Aspect-level sentiment classification is a finegrained task in sentiment analysis.Since it provides more complete and in-depth results, aspect-level sentiment analysis has received much attention these years.In this paper, we reveal that the sentiment polarity of a sentence is not only determined by the content but is also highly related to the concerned aspect.For instance, "The appetizers are ok, but the service is slow.",for aspect taste, the polarity is positive while for service, the polarity is negative.Therefore, it is worthwhile to explore the connection between an aspect and the content of a sentence.To this end, we propose an Attention-based Long Short-Term Memory Network for aspect-level sentiment classification.The attention mechanism can concentrate on different parts of a sentence when different aspects are taken as input.We experiment on the SemEval 2014 dataset and results show that our model achieves state-ofthe-art performance on aspect-level sentiment classification.
Yequan Wang, Minlie Huang, Xiaoyan Zhu 0001, Li Zhao 0007
EMNLP2
2016 From One Point to a Manifold: Knowledge Graph Embedding for Precise Link Prediction
Han Xiao 0005, Minlie Huang, Xiaoyan Zhu 0001
IJCAI2
2016 Knowledge Graph Embedding by Flexible Translation
Minlie Huang, Mingdong Wang, Mantong Zhou, Yu Hao 0001, Xiaoyan Zhu 0001
KR2
2015 Learning Tag Embeddings and Tag-specific Composition Functions in Recursive Neural Network
abstract
Qiao Qian, Bo Tian, Minlie Huang, Yang Liu, Xuan Zhu, Xiaoyan Zhu. Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2015.
Qiao Qian, Minlie Huang, Yang Liu 0107, Xuan Zhu 0006, Xiaoyan Zhu 0001
ACL (1)3
2015 Sentiment Extraction by Leveraging Aspect-Opinion Association Structure
abstract
Sentiment extraction aims to extract and group the task of extracting and grouping aspect and opinion words from online reviews. Previous works usually extract aspect and opinion words by leveraging association between a single pair of aspect and opinion word[5] [14] [9] [4][11], but the structure of aspect and opinion word clusters has not been fully exploited.
Li Zhao 0007, Minlie Huang, Jiashen Sun, Hengliang Luo, Xiankai Yang, Xiaoyan Zhu 0001
CIKM2
2015 Incorporating Domain and Sentiment Supervision in Representation Learning for Domain Adaptation
Minlie Huang, Jiashen Sun, Xuan Zhu 0006
IJCAI2
2015 Clustering Sentiment Phrases in Product Reviews by Constrained Co-clustering
abstract
Clustering sentiment phrases in product reviews is convenient for us to get the most important information about one product directly through thousands of reviews. There are mainly two components in a sentiment phrase, the aspect word and the opinion word. We need to cluster these two parts simultaneously. Although several methods have been proposed to cluster words or phrases, limited work has been done on clustering two-dimensional sentiment phrases. In this paper, we apply a two-sided hidden Markov random field (HMRF) model on this task. We use the approach of constrained co-clustering with some priori knowledge, in a semi-supervised setting. Experimental results on sentiment phrases extracted from about 0.7 million mobile phone reviews show that this method is promising for this task and our method outperforms baselines remarkably.
Minlie Huang, Xiaoyan Zhu 0001
NLPCC2
2015 Leveraging Large Data with Weak Supervision for Joint Feature and Opinion Word Extraction
Lei Fang 0004, Minlie Huang
J. Comput. Sci. Technol.3
2014 New Word Detection for Sentiment Analysis
abstract
Automatic extraction of new words is an indispensable precursor to many NLP tasks such as Chinese word segmentation, named entity extraction, and sentiment analysis.This paper aims at extracting new sentiment words from large-scale user-generated content.We propose a fully unsupervised, purely data-driven framework for this purpose.We design statistical measures respectively to quantify the utility of a lexical pattern and to measure the possibility of a word being a new word.The method is almost free of linguistic resources (except POS tags), and requires no elaborated linguistic rules.We also demonstrate how new sentiment word will benefit sentiment analysis.Experiment results demonstrate the effectiveness of the proposed method.
Minlie Huang, Borui Ye, Haiqiang Chen, Junjun Cheng, Xiaoyan Zhu 0001
ACL (1)1
2014 Ranking Sentiment Explanations for Review Summarization Using Dual Decomposition
abstract
For online reviews, sentiment explanations refer to the sentences that may suggest detailed reasons of sentiment, which are very important for applications in review mining like opinion summarization. In this paper, we address the problem of ranking sentiment explanations by formulating the process as two subproblems: sentence informativeness ranking and structural sentiment analysis. Tractable inference in joint prediction is performed through dual decomposition. Preliminary experiments on publicly available data demonstrate that our approach obtains promising performance.
Lei Fang 0004, Qiao Qian, Minlie Huang, Xiaoyan Zhu 0001
CIKM3
2014 Clustering Aspect-related Phrases by Leveraging Sentiment Distribution Consistency
abstract
Clustering aspect-related phrases in terms of product’s property is a precursor pro-cess to aspect-level sentiment analysis which is a central task in sentiment analy-sis. Most of existing methods for address-ing this problem are context-based models which assume that domain synonymous phrases share similar co-occurrence con-texts. In this paper, we explore a novel idea, sentiment distribution consistency, which states that different phrases (e.g. “price”, “money”, “worth”, and “cost”) of the same aspect tend to have consistent sentiment distribution. Through formal-izing sentiment distribution consistency as soft constraint, we propose a novel unsu-pervised model in the framework of Poste-rior Regularization (PR) to cluster aspect-related phrases. Experiments demonstrate that our approach outperforms baselines remarkably. 1
Li Zhao 0007, Minlie Huang, Haiqiang Chen, Junjun Cheng, Xiaoyan Zhu 0001
EMNLP2
2014 Exploring the Interactions of Storylines from Informative News Events
Minlie Huang, Xiaoyan Zhu 0001
J. Comput. Sci. Technol.2
2014 Estimating feature ratings through an effective review selection approach
Chong Long, Jie Zhang 0002, Minlie Huang, Xiaoyan Zhu 0001, Ming Li 0001, Bin Ma 0002
Knowl. Inf. Syst.3
2013 Exploring weakly supervised latent sentiment explanations for aspect-level review analysis
abstract
In sentiment analysis, aspect-level review analysis has been an important task because it can catalogue, aggregate, or summarize various opinions according to a product's properties. In this paper, we explore a new concept for aspect-level review analysis, latent sentiment explanations, which are defined as a set of informative aspect-specific sentences whose polarities are consistent with that of the review. In other words, sentiment explanations best represent a review in terms of both aspect and polarity. We formulate the problem as a structure learning problem, and sentiment explanations are modeled with latent variables. Training samples are automatically identified through a set of pre-defined aspect signature terms (i.e., without manual annotation on samples), which we term the way weakly supervised.
Lei Fang 0004, Minlie Huang, Xiaoyan Zhu 0001
CIKM2
2012 Using First-Order Logic to Compress Sentences
abstract
Sentence compression is one of the most challenging tasks in natural language processing,which may be of increasing interest to many applicationssuch as abstractive summarization and text simplification for mobile devices.In this paper, we present a novel sentence compression model based on first-order logic, using Markov Logic Network.Sentence compression is formulated as a word/phrase deletion problem in this model.By taking advantage of first-order logic, the proposed method is able to incorporate local linguistic features and to capture global dependencies between word deletion operations. Experiments on both written and spoken corpora show that our approach produces competitive performance against the state-of-the-art methods in terms of manual evaluation measures such as importance, grammaticality, and overall quality.
Minlie Huang, Xiaoyan Zhu 0001
AAAI1
2012 Finding nuggets in IP portfolios: core patent mining through textual temporal analysis
abstract
Patents are critical for a company to protect its core technologies. Effective patent mining in massive patent databases can provide companies with valuable insights to develop strategies for IP management and marketing. In this paper, we study a novel patent mining problem of automatically discovering core patents (i.e., patents with high novelty and influence in a domain). We address the unique patent vocabulary usage problem, which is not considered in traditional word-based statistical methods, and propose a topic-based temporal mining approach to quantify a patent's novelty and influence. Comprehensive experimental results on real-world patent portfolios show the effectiveness of our method.
Minlie Huang, Peng Xu 0002, Weichang Li, Adam K. Usadi, Xiaoyan Zhu 0001
CIKM2
2012 Sentiment Analysis with Multi-source Product Reviews
Minlie Huang, Xiaoyan Zhu 0001
ICIC (1)2
2012 PatentMiner: topic-driven patent analysis and mining
abstract
Patenting is one of the most important ways to protect company's core business concepts and proprietary technologies. Analyzing large volume of patent data can uncover the potential competitive or collaborative relations among companies in certain areas, which can provide valuable information to develop strategies for intellectual property (IP), R&D, and marketing. In this paper, we present a novel topic-driven patent analysis and mining system. Instead of merely searching over patent content, we focus on studying the heterogeneous patent network derived from the patent database, which is represented by several types of objects (companies, inventors, and technical content) jointly evolving over time. We design and implement a general topic-driven framework for analyzing and mining the heterogeneous patent network. Specifically, we propose a dynamic probabilistic model to characterize the topical evolution of these objects within the patent network. Based on this modeling framework, we derive several patent analytics tools that can be directly used for IP and R&D strategy planning, including a heterogeneous network co-ranking method, a topic-level competitor evolution analysis algorithm, and a method to summarize the search results. We evaluate the proposed methods on a real-world patent database. The experimental results show that the proposed techniques clearly outperform the corresponding baseline methods.
Jie Tang 0001, Bo Wang 0022, Yang Yang 0009, Yanting Zhao, Minlie Huang, Peng Xu 0002, Weichang Li, Adam K. Usadi
KDD8
2012 A Unified Active Learning Framework for Biomedical Relation Extraction
Minlie Huang, Xiaoyan Zhu 0001
J. Comput. Sci. Technol.2
2011 Generating Breakpoint-based Timeline Overview for News Topic Retrospection
abstract
Though news readers can easily access a large number of news articles from the Internet, they can be overwhelmed by the quantity of information available, making it hard to get a concise, global picture of a news topic. In this paper we propose a novel method to address this problem. Given a set of articles for a given news topic, the proposed method models theme variation through time and identifies the breakpoints, which are time points when decisive changes occur. For each breakpoint, a brief summary is automatically constructed based on articles associated with the particular time point. Summaries are then ordered chronologically to form a timeline overview of the news topic. In this fashion, readers can easily track various news topics efficiently. We have conducted experiments on 15 popular topics in 2010. Empirical experiments show the effectiveness of our approach and its advantages over other approaches.
Minlie Huang, Peng Xu 0002, Weichang Li, Adam K. Usadi, Xiaoyan Zhu 0001
ICDM2
2011 Learning to Identify Review Spam
abstract
In the past few years, sentiment analysis and opinion mining becomes a popular and important task. These studies all assume that their opinion resources are real and trustful. However, they may encounter the faked opinion or opinion spam problem. In this paper, we study this issue in the context of our product review mining system. On product review site, people may write faked reviews, called review spam, to promote their products, or defame their competitors’ products. It is important to identify and filter out the review spam. Previous work only focuses on some heuristic rules, such as helpfulness voting, or rating deviation, which limits the performance of this task. In this paper, we exploit machine learning methods to identify review spam. Toward the end, we manually build a spam collection from our crawled reviews. We first analyze the effect of various features in spam identification. We also observe that the review spammer consistently writes spam. This provides us another view to identify review spam: we can identify if the author of the review is spammer. Based on this observation, we provide a twoview semi-supervised method, co-training, to exploit the large amount of unlabeled data. The experiment results show that our proposed method is effective. Our designed machine learning methods achieve significant improvements in comparison to the heuristic baselines.
Fangtao Li, Minlie Huang, Yi Yang 0038, Xiaoyan Zhu 0001
IJCAI2
2011 Quality-biased Ranking of Short Texts in Microblogging Services
Minlie Huang, Yi Yang 0038, Xiaoyan Zhu 0001
IJCNLP1
2011 GeneTUKit: a software for document-level gene normalization
abstract
MOTIVATION: Linking gene mentions in an article to entries of biological databases can facilitate indexing and querying biological literature greatly. Due to the high ambiguity of gene names, this task is particularly challenging. Manual annotation for this task is cost expensive, time consuming and labor intensive. Therefore, providing assistive tools to facilitate the task is of high value. RESULTS: We developed GeneTUKit, a document-level gene normalization software for full-text articles. This software employs both local context surrounding gene mentions and global context from the whole full-text document. It can normalize genes of different species simultaneously. When participating in BioCreAtIvE III, the system obtained good results among 37 runs: the system was ranked first, fourth and seventh in terms of TAP-20, TAP-10 and TAP-5, respectively on the 507 full-text test articles. AVAILABILITY AND IMPLEMENTATION: The software is available at http://www.qanswers.net/GeneTUKit/.
Minlie Huang, Jingchen Liu, Xiaoyan Zhu 0001
Bioinform.1
2011 The gene normalization task in BioCreative III
abstract
BACKGROUND: We report the Gene Normalization (GN) challenge in BioCreative III where participating teams were asked to return a ranked list of identifiers of the genes detected in full-text articles. For training, 32 fully and 500 partially annotated articles were prepared. A total of 507 articles were selected as the test set. Due to the high annotation cost, it was not feasible to obtain gold-standard human annotations for all test articles. Instead, we developed an Expectation Maximization (EM) algorithm approach for choosing a small number of test articles for manual annotation that were most capable of differentiating team performance. Moreover, the same algorithm was subsequently used for inferring ground truth based solely on team submissions. We report team performance on both gold standard and inferred ground truth using a newly proposed metric called Threshold Average Precision (TAP-k). RESULTS: We received a total of 37 runs from 14 different teams for the task. When evaluated using the gold-standard annotations of the 50 articles, the highest TAP-k scores were 0.3297 (k=5), 0.3538 (k=10), and 0.3535 (k=20), respectively. Higher TAP-k scores of 0.4916 (k=5, 10, 20) were observed when evaluated using the inferred ground truth over the full test set. When combining team results using machine learning, the best composite system achieved TAP-k scores of 0.3707 (k=5), 0.4311 (k=10), and 0.4477 (k=20) on the gold standard, representing improvements of 12.4%, 21.8%, and 26.6% over the best team results, respectively. CONCLUSIONS: By using full text and being species non-specific, the GN task in BioCreative III has moved closer to a real literature curation task than similar tasks in the past and presents additional challenges for the text mining community, as revealed in the overall team results. By evaluating teams using the gold standard, we show that the EM algorithm allows team submissions to be differentiated while keeping the manual annotation effort feasible. Using the inferred ground truth we show measures of comparative performance between teams. Finally, by comparing team rankings on gold standard vs. inferred ground truth, we further demonstrate that the inferred ground truth is as effective as the gold standard for detecting good team performance.
Zhiyong Lu, Hung-Yu Kao, Chih-Hsuan Wei, Minlie Huang, Jingchen Liu, Cheng-Ju Kuo, Chun-Nan Hsu, Richard Tzong-Han Tsai, Hong-Jie Dai, Naoaki Okazaki, Hancheol Cho, Martin Gerner, Illés Solt, Shashank Agarwal, Dina Vishnyakova, Patrick Ruch, Martin Romacker, Fabio Rinaldi 0001, Sanmitra Bhattacharya, Padmini Srinivasan, Manabu Torii, Sérgio Matos, David Campos 0001, Karin Verspoor, Kevin M. Livingston, W. John Wilbur
BMC Bioinform.4
2011 Recommending MeSH terms for annotating biomedical articles
abstract
BACKGROUND: Due to the high cost of manual curation of key aspects from the scientific literature, automated methods for assisting this process are greatly desired. Here, we report a novel approach to facilitate MeSH indexing, a challenging task of assigning MeSH terms to MEDLINE citations for their archiving and retrieval. METHODS: Unlike previous methods for automatic MeSH term assignment, we reformulate the indexing task as a ranking problem such that relevant MeSH headings are ranked higher than those irrelevant ones. Specifically, for each document we retrieve 20 neighbor documents, obtain a list of MeSH main headings from neighbors, and rank the MeSH main headings using ListNet-a learning-to-rank algorithm. We trained our algorithm on 200 documents and tested on a previously used benchmark set of 200 documents and a larger dataset of 1000 documents. RESULTS: Tested on the benchmark dataset, our method achieved a precision of 0.390, recall of 0.712, and mean average precision (MAP) of 0.626. In comparison to the state of the art, we observe statistically significant improvements as large as 39% in MAP (p-value <0.001). Similar significant improvements were also obtained on the larger document set. CONCLUSION: Experimental results show that our approach makes the most accurate MeSH predictions to date, which suggests its great potential in making a practical impact on MeSH indexing. Furthermore, as discussed the proposed learning framework is robust and can be adapted to many other similar tasks beyond MeSH indexing in the biomedical domain. All data sets are available at: http://www.ncbi.nlm.nih.gov/CBBresearch/Lu/indexing.
Minlie Huang, Aurélie Névéol, Zhiyong Lu
J. Am. Medical Informatics Assoc.1
2011 Guided Structure-Aware Review Summarization
Minlie Huang, Xiaoyan Zhu 0001
J. Comput. Sci. Technol.2
2010 Sentiment Analysis with Global Topics and Local Dependency
abstract
With the development of Web 2.0, sentiment analysis has now become a popular research problem to tackle. Recently, topic models have been introduced for the simultaneous analysis for topics and the sentiment in a document. These studies, which jointly model topic and sentiment, take the advantage of the relationship between topics and sentiment, and are shown to be superior to traditional sentiment analysis tools. However, most of them make the assumption that, given the parameters, the sentiments of the words in the document are all independent. In our observation, in contrast, sentiments are expressed in a coherent way. The local conjunctive words, such as “and” or “but”, are often indicative of sentiment transitions. In this paper, we propose a major departure from the previous approaches by making two linked contributions. First, we assume that the sentiments are related to the topic in the document, and put forward a joint sentiment and topic model, i.e. Sentiment-LDA. Second, we observe that sentiments are dependent on local context. Thus, we further extend the Sentiment-LDA model to Dependency-Sentiment-LDA model by relaxing the sentiment independent assumption in Sentiment-LDA. The sentiments of words are viewed as a Markov chain in Dependency-Sentiment-LDA. Through experiments, we show that exploiting the sentiment dependency is clearly advantageous, and that the Dependency-Sentiment-LDA is an effective approach for sentiment analysis.
Fangtao Li, Minlie Huang, Xiaoyan Zhu 0001
AAAI2
2010 Metadata-Aware Measures for Answer Summarization in Community Question Answering
Mattia Tomasoni, Minlie Huang
ACL2
2010 Structure-Aware Review Mining and Summarization
Fangtao Li, Minlie Huang, Xiaoyan Zhu 0001, Yingju Xia, Shu Zhang 0004, Hao Yu 0005
COLING3
2010 Learning to Link Entities with Knowledge Base
Zhicheng Zheng, Fangtao Li, Minlie Huang, Xiaoyan Zhu 0001
HLT-NAACL3
2010 A New Approach for Multi-Document Update Summarization
Chong Long, Minlie Huang, Xiaoyan Zhu 0001, Ming Li 0001
J. Comput. Sci. Technol.2
2009 Answering Opinion Questions with Random Walks on Graphs
Fangtao Li, Minlie Huang, Xiaoyan Zhu 0001
ACL/IJCNLP3
2009 Multi-document Summarization by Information Distance
abstract
Fast changing knowledge on the Internet can be acquired more efficiently with the help of automatic document summarization and updating techniques. This paper described a novel approach for multi-document update summarization. The best summary is defined to be the one which has the minimum information distance to the entire document set. The best update summary has the minimum conditional information distance to a document cluster given that a prior document cluster has already been read. Experiments on the DUC 2007 dataset and the TAC 2008 dataset have proved that our method closely correlates with the human summaries and outperforms other programs such as LexRank in many categories under the ROUGE evaluation criterion.
Chong Long, Minlie Huang, Xiaoyan Zhu 0001, Ming Li 0001
ICDM2
2009 Summarizing Documents by Measuring the Importance of a Subset of Vertices within a Graph
abstract
This paper presents a novel method of generating extractive summaries for multiple documents. Given a cluster of documents, we firstly construct a graph where each vertex represents a sentence and edges are created according to the asymmetric relationship between sentences. Then we develop a method to measure the importance of a subset of vertices by adding a super-vertex into the original graph. The importance of such a super-vertex is quantified as super-centrality, a quantitative measure for the importance of a subset of vertices within the whole graph. Finally, we propose a heuristic algorithm to find the best summary. Our method is evaluated with extensive experiments. The comparative results show that the proposed method outperforms other methods on several datasets.
Shouyuan Chen, Minlie Huang, Zhiyong Lu
Web Intelligence2
2009 Specialized Review Selection for Feature Rating Estimation
abstract
On participatory Websites, users provide opinions about products, with both overall ratings and textual reviews. In this paper, we propose an approach to accurately estimate feature ratings of the products. This approach selects user reviews that extensively discuss specific features of the products (called specialized reviews), using information distance of reviews on the features. Experiments on real data show that overall ratings of the specialized reviews can be used to represent their feature ratings. The average of these overall ratings can be used by recommender systems to provide feature specific recommendations that better help users make purchasing decisions.
Chong Long, Jie Zhang 0002, Minlie Huang, Xiaoyan Zhu 0001, Ming Li 0001, Bin Ma 0002
Web Intelligence3
2009 Extract interaction detection methods from the biological literature
abstract
BACKGROUND: Considerable efforts have been made to extract protein-protein interactions from the biological literature, but little work has been done on the extraction of interaction detection methods. It is crucial to annotate the detection methods in the literature, since different detection methods shed different degrees of reliability on the reported interactions. However, the diversity of method mentions in the literature makes the automatic extraction quite challenging. RESULTS: In this article, we develop a generative topic model, the Correlated Method-Word model (CMW model) to extract the detection methods from the literature. In the CMW model, we formulate the correlation between the different methods and related words in a probabilistic framework in order to infer the potential methods from the given document. By applying the model on a corpus of 5319 full text documents annotated by the MINT and IntAct databases, we observe promising results, which outperform the best result reported in the BioCreative II challenge evaluation. CONCLUSION: From the promising experiment results, we can see that the CMW model overcomes the issues caused by the diversity in the method mentions and properly captures the in-depth correlations between the detection methods and related words. The performance outperforming the baseline methods confirms that the dependence assumptions of the model are reasonable and the model is competent for the practical processing.
Hongning Wang, Minlie Huang, Xiaoyan Zhu 0001
BMC Bioinform.2
2008 A Generative Probabilistic Model for Multi-label Classification
abstract
Traditional discriminative classification method makes little attempt to reveal the probabilistic structure and the correlation within both input and output spaces. In the scenario of multi-label classification, most of the classifiers simply assume the predefined classes are independently distributed, which would definitely hinder the classification performance when there are intrinsic correlations between the classes. In this article, we propose a generative probabilistic model, the Correlated Labeling Model (CoL Model), to formulate the correlation between different classes. The CoL model is presented to capture the correlation between classes and the underlying structures via the latent random variables in a supervised manner. We develop a variational procedure to approximate the posterior distribution and employ the EM algorithm for the empirical Bayes parameter estimation. In our evaluations, the proposed model achieved promising results on various data sets.
Hongning Wang, Minlie Huang, Xiaoyan Zhu 0001
ICDM2
2008 Exploiting and integrating rich features for biological literature classification
abstract
BACKGROUND: Efficient features play an important role in automated text classification, which definitely facilitates the access of large-scale data. In the bioscience field, biological structures and terminologies are described by a large number of features; domain dependent features would significantly improve the classification performance. How to effectively select and integrate different types of features to improve the biological literature classification performance is the major issue studied in this paper. RESULTS: To efficiently classify the biological literatures, we propose a novel feature value schema TF*ML, features covering from lower level domain independent "string feature" to higher level domain dependent "semantic template feature", and proper integrations among the features. Compared to our previous approaches, the performance is improved in terms of AUC and F-Score by 11.5% and 8.8% respectively, and outperforms the best performance achieved in BioCreAtIvE 2006. CONCLUSIONS: Different types of features possess different discriminative capabilities in literature classification; proper integration of domain independent and dependent features would significantly improve the performance and overcome the over-fitting on data distribution.
Hongning Wang, Minlie Huang, Shilin Ding, Xiaoyan Zhu 0001
BMC Bioinform.2
2007 Semi-supervised Pattern Learning for Extracting Relations from Bioscience Texts
Shilin Ding, Minlie Huang, Xiaoyan Zhu 0001
APBC2
2007 A Novel Kernel-Based Approach for Predicting Binding Peptides for HLA Class II Molecules
Hao Yu 0005, Minlie Huang, Xiaoyan Zhu 0001, Yabin Guo
ISBRA2
2007 Combining Convolution Kernels Defined on Heterogeneous Sub-structures
Minlie Huang, Xiaoyan Zhu 0001
PAKDD1
2006 ONBRIRES: Ontology-Based Biological Relation Extraction System
Minlie Huang, Xiaoyan Zhu 0001, Shilin Ding, Hao Yu 0005, Ming Li 0001
APBC1
2005 Discovering patterns to extract protein-protein interactions from the literature: Part II
abstract
MOTIVATION: An enormous number of protein-protein interaction relationships are buried in millions of research articles published over the years, and the number is growing. Rediscovering them automatically is a challenging bioinformatics task. Solutions to this problem also reach far beyond bioinformatics. RESULTS: We study a new approach that involves automatically discovering English expression patterns, optimizing them and using them to extract protein-protein interactions. In a sister paper, we described how to generate English expression patterns related to protein-protein interactions, and this approach alone has already achieved precision and recall rates significantly higher than those of other automatic systems. This paper continues to present our theory, focusing on how to improve the patterns. A minimum description length (MDL)-based pattern-optimization algorithm is designed to reduce and merge patterns. This has significantly increased generalization power, and hence the recall and precision rates, as confirmed by our experiments. AVAILABILITY: http://spies.cs.tsinghua.edu.cn.
Hao Yu 0005, Xiaoyan Zhu 0001, Minlie Huang, Ming Li 0001
Bioinform.3
2004 Discovering patterns to extract protein-protein interactions from full texts
abstract
MOTIVATION: Although there are several databases storing protein-protein interactions, most such data still exist only in the scientific literature. They are scattered in scientific literature written in natural languages, defying data mining efforts. Much time and labor have to be spent on extracting protein pathways from literature. Our aim is to develop a robust and powerful methodology to mine protein-protein interactions from biomedical texts. RESULTS: We present a novel and robust approach for extracting protein-protein interactions from literature. Our method uses a dynamic programming algorithm to compute distinguishing patterns by aligning relevant sentences and key verbs that describe protein interactions. A matching algorithm is designed to extract the interactions between proteins. Equipped only with a dictionary of protein names, our system achieves a recall rate of 80.0% and precision rate of 80.5%. AVAILABILITY: The program is available on request from the authors.
Minlie Huang, Xiaoyan Zhu 0001, Hao Yu 0005, Donald G. Payan, Kunbin Qu, Ming Li 0001
Bioinform.1