Outcome-based reinforcement learning has made notable advances in training language models (LMs) for reasoning. However, without explicit incentives and controls, this paradigm has limitations and instability in eliciting high-quality reasoning trajectories with diverse actions—particularly for models whose pretraining lacked extensive reasoning-related data. To this end, we introduce MetaAct-RL, a new RL framework that frames LMs’ thinking as sequential decision making over meta-actions. In this framework, the model chooses and executes a high-level action at each step—such as forward reasoning, critique, or refinement—to gradually reach the correct answer. To encourage deeper exploration, richer action diversity, and to improve sampling efficiency in the RL optimization process, MetaAct-RL incorporates appropriate length-based reward and regularization, and a key-state restart mechanism. Extensive experiments across six benchmarks show that MetaAct-RL improves reasoning performance by 7.99 on Llama3.2-1B and 7.17 on Llama3.1-8B relative to vanilla RL method. Moreover, on the challenging AIME-2024, our method outperforms the vanilla RL by 7.5 with Qwen2.5-1.5B.

Large language models (LLMs) have traditionally been aligned through one-size-fits-all approaches that assume uniform human preferences, fundamentally overlooking the diversity in user values and needs.This paper introduces a comprehensive framework for scalable personalized alignment of LLMs.We establish a systematic preference space characterizing psychological and behavioral dimensions, alongside diverse persona representations for robust preference inference in real-world scenarios.Building upon this foundation, we introduce ALIGNX, a large-scale dataset of over 1.3 million personalized preference examples, and develop two complementary alignment approaches: in-context alignment directly conditioning on persona representations and preference-bridged alignment modeling intermediate preference distributions.Extensive experiments demonstrate substantial improvements over existing methods, with an average 17.06% accuracy gain across four benchmarks while exhibiting a strong adaptation capability to novel preferences, robustness to limited user data, and precise preference controllability.These results validate our approach toward user-adaptive AI systems.

Despite rapid development, large language models (LLMs) still encounter challenges in multi-turn decision-making tasks (i.e., agent tasks) like web shopping and browser navigation, which require making a sequence of intelligent decisions based on environmental feedback. Previous work for LLM agents typically relies on elaborate prompt engineering or fine-tuning with expert trajectories to improve performance. In this work, we take a different perspective: we explore constructing process reward models (PRMs) to evaluate each decision and guide the agent's decision-making process. Unlike LLM reasoning, where each step is scored based on correctness, actions in agent tasks do not have a clear-cut correctness. Instead, they should be evaluated based on their proximity to the goal and the progress they have made. Building on this insight, we propose a re-defined PRM for agent tasks, named AgentPRM, to capture both the interdependence between sequential decisions and their contribution to the final goal. This enables better progress tracking and exploration-exploitation balance. To scalably obtain labeled data for training AgentPRM, we employ a Temporal Difference-based (TD-based) estimation method combined with Generalized Advantage Estimation (GAE), which proves more sample-efficient than prior methods. Extensive experiments across different agentic tasks show that AgentPRM is over 8× more compute-efficient than baselines, and it demonstrates robust improvement when scaling up test-time compute. Moreover, we perform detailed analyses to show how our method works and offer more insights, e.g., applying AgentPRM to the reinforcement learning of LLM agents.

While Direct Preference Optimization (DPO) eliminates complex reward modeling in aligning large language models (LLMs) with human preferences, its online variant faces significant efficiency bottlenecks due to costly real-time preference sampling and the reward model annotation.We propose a novel framework that bridges offline-to-online alignment by systematically transforming static datasets into dynamically adaptive equivalents, without the need for an explicit reward model.Our approach employs paraphrasing techniques to preserve response correctness while aligning data distributions with model-generated outputs, circumventing the need for resource-intensive online interactions.Experiments on mathematical reasoning and conversational tasks demonstrate that our method matches or exceeds the performance of a fully online DPO.This work establishes a computationally sustainable paradigm for LLM alignment, particularly benefiting scenarios requiring iterative preference updates and domain adaptation.

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.

Long-form video processing fundamentally challenges vision-language models (VLMs) due to the high computational costs of handling extended temporal sequences. Existing token pruning and feature merging methods often sacrifice critical temporal dependencies or dilute semantic information. We introduce differential distillation, a principled approach that systematically preserves task-relevant information while suppressing redundancy. Based on this principle, we develop ViLAMP, a hierarchical video-language model that processes hour-long videos at "mixed precision" through two key mechanisms: (1) differential keyframe selection that maximizes query relevance while maintaining temporal distinctiveness at the frame level and (2) differential feature merging that preserves query-salient features in non-keyframes at the patch level. Hence, ViLAMP retains full information in keyframes while reducing non-keyframes to their most salient features, resembling mixed-precision training. Extensive experiments demonstrate ViLAMP's superior performance across five video understanding benchmarks, particularly on long-form content. Notably, ViLAMP can process ultra-long videos (up to 10K frames) on a single NVIDIA A100 GPU, achieving substantial computational efficiency while maintaining state-of-the-art performance. Code and model are available at https://github.com/steven-ccq/ViLAMP.

Large language models (LLMs) have achieved remarkable success, yet aligning their generations with human preferences remains a critical challenge. Existing approaches to preference modeling often rely on an explicit or implicit reward function, overlooking the intricate and multifaceted nature of human preferences that may encompass conflicting factors across diverse tasks and populations. To address this limitation, we introduce Latent Preference Coding (LPC), a novel framework that models the implicit factors as well as their combinations behind holistic preferences using discrete latent codes. LPC seamlessly integrates with various offline alignment algorithms, automatically inferring the underlying factors and their importance from data without relying on pre-defined reward functions and hand-crafted combination weights. Extensive experiments on multiple benchmarks demonstrate that LPC consistently improves upon three alignment algorithms (DPO, SimPO, and IPO) using three base models (Mistral-7B, Llama3-8B, and Llama3-Instruct-8B). Furthermore, deeper analysis reveals that the learned latent codes effectively capture the differences in the distribution of human preferences and significantly enhance the robustness of alignment algorithms against noise in data. By providing a unified representation for the multifarious preference factors, LPC paves the way towards developing more robust and versatile alignment techniques for responsible deployment of powerful LLMs.

Diffusion language models have emerged as a new approach for text generation. By enabling the parallel sampling of multiple tokens in each diffusion step, they appear to offer a more efficient alternative to auto-regressive models. However, our observations show that current open-sourced diffusion language models require more sampling steps to achieve comparable accuracy on representative tasks--resulting in even higher inference costs than their auto-regressive counterparts. To investigate whether this is an inherent limitation, we conduct a rigorous theoretical analysis of a widely adopted variant: the Masked Diffusion Model (MDM). Surprisingly, our analysis reveals that the conclusion is highly sensitive to the choice of evaluation metric. Under mild conditions, we prove that when the target is near-optimal perplexity, MDMs can achieve this goal in a constant number of sampling steps, independent of sequence length. This result demonstrates that efficiency can, in principle, be attained without compromising generation quality. However, when targeting low sequence error rate--which is important for assessing the ``correctness" of a generated sequence, such as a reasoning chain--we show that in the worst case, the required sampling steps must scale linearly with sequence length, thereby eliminating the efficiency advantage. Our analysis establishes the first theoretical foundation for understanding the comparative strengths and limitations of MDMs, offering practical guidance on when to favor MDMs over the auto-regressive models and vice versa.

As textual reasoning with large language models (LLMs) has advanced significant, there has been growing interest in enhancing the multimodal reasoning capabilities of large vision-language models (LVLMs). However, existing methods primarily approach multimodal reasoning in a straightforward, text-centric manner, where both reasoning and answer derivation are conducted purely through text, with the only difference being the presence of multimodal input. As a result, these methods often encounter fundamental limitations in spatial reasoning tasks that demand precise geometric understanding and continuous spatial tracking\textemdash capabilities that humans achieve through mental visualization and manipulation. To address the limitations, we propose drawing to reason in space, a novel paradigm that enables LVLMs to reason through elementary drawing operations in the visual space. By equipping models with basic drawing operations including annotating bounding boxes and drawing auxiliary lines, we empower them to express and analyze spatial relationships through direct visual manipulation, meanwhile avoiding the performance ceiling imposed by specialized perception tools in previous tool-integrated reasoning approaches. To cultivate this capability, we develop a three-stage training framework: cold-start training with synthetic data to establish basic drawing abilities, reflective rejection sampling to enhance self-reflection behaviors, and reinforcement learning to directly optimize for target rewards. Extensive experiments demonstrate that our model, named \textsc{Spark}, consistently outperforms existing methods across diverse spatial reasoning benchmarks involving maze navigation, static spatial reasoning, video-based reasoning and multi-view-based reasoning tasks, with an average improvement of 11.5\%. Ablation studies reveal the critical role of each training stage, with reflective rejection sampling particularly enhancing the model's self-correction capabilities and reasoning potential.

In modern sequential decision-making systems, the construction of an optimal candidate action space is critical to efficient inference. However, existing approaches either rely on manually defined action spaces that lack scalability or utilize unstructured spaces that render exhaustive search computationally prohibitive. In this paper, we propose a novel framework named \textsc{DynaAct} for automatically constructing a compact action space to enhance sequential reasoning in complex problem-solving scenarios. Our method first estimates a proxy for the complete action space by extracting general sketches observed in a corpus covering diverse complex reasoning problems using large language models. We then formulate a submodular function that jointly evaluates candidate actions based on their utility to the current state and their diversity, and employ a greedy algorithm to select an optimal candidate set. Extensive experiments on six diverse standard benchmarks demonstrate that our approach significantly improves overall performance, while maintaining efficient inference without introducing substantial latency. The implementation is available at \url{https://github.com/zhaoxlpku/DynaAct}.

Retrieval-Augmented Generation (RAG) provides additional contextual knowledge to complement the parametric knowledge in Large Language Models (LLMs). These two knowledge interweave to enhance the accuracy and timeliness of LLM responses. However, the internal mechanisms by which LLMs utilize these knowledge remain unclear. We propose modeling the forward propagation of knowledge as an entity flow, employing this framework to trace LLMs' internal behaviors when processing mixed-source knowledge. Linear probing utilizes a trainable linear classifier to detect specific attributes in hidden layers. However, once trained, a probe cannot adapt to dynamically specified entities. To address this challenge, we construct an entity-aware probe, which introduces special tokens to mark probing targets and employs a small trainable rank-8 lora update to process these special markers. We first verify this approach through an attribution experiment, demonstrating that it can accurately detect information about ad-hoc entities from complex hidden states. Next, we trace entity flows across layers to understand how LLMs reconcile conflicting knowledge internally. Our probing results reveal that contextual and parametric knowledge are routed between tokens through distinct sets of attention heads, supporting attention competition only within knowledge types. While conflicting knowledge maintains a residual presence across layers, aligned knowledge from multiple sources gradually accumulates, with the magnitude of this accumulation directly determining its influence on final outputs.

Tables are ubiquitous across various domains for concisely representing structured information. Empowering large language models (LLMs) to reason over tabular data represents an actively explored direction. However, since typical LLMs only support one-dimensional (1D) inputs, existing methods often flatten the two-dimensional (2D) table structure into a sequence of tokens, which can severely disrupt the spatial relationships and result in an inevitable loss of vital contextual information. In this paper, we first empirically demonstrate the detrimental impact of such flattening operations on the performance of LLMs in capturing the spatial information of tables through two elaborate proxy tasks. Subsequently, we introduce a simple yet effective positional encoding method, termed "2D-TPE" (Two-Dimensional Table Positional Encoding), to address this challenge. 2D-TPE enables each attention head to dynamically select a permutation order of tokens within the context for attending to them, where each permutation represents a distinct traversal mode for the table, such as column-wise or row-wise traversal. 2D-TPE effectively mitigates the risk of losing essential spatial information while preserving computational efficiency, thus better preserving the table structure. Extensive experiments across five benchmarks demonstrate that 2D-TPE outperforms strong baselines, underscoring the importance of preserving the table structure for accurate table comprehension. Comprehensive analysis further reveals the substantially better scalability of 2D-TPE to large tables than baselines.

We explore multi-step reasoning in vision-language models (VLMs). The problem is challenging, as reasoning data consisting of multiple steps of visual and language processing are barely available. To overcome the challenge, we first introduce a least-to-most visual reasoning paradigm, which interleaves steps of decomposing a question into sub-questions and invoking external tools for resolving sub-questions. Based on the paradigm, we further propose a novel data synthesis approach that can automatically create questions and multi-step reasoning paths for an image in a bottom-up manner. Our approach divides the complex synthesis task into a few simple sub-tasks, and (almost entirely) relies on open-sourced models to accomplish the sub-tasks. Therefore, the entire synthesis process is reproducible and cost-efficient, and the synthesized data is quality guaranteed. With the approach, we construct 50k visual reasoning examples. Then, we develop a visual reasoner through supervised fine-tuning, which is capable of generally enhancing the reasoning abilities of a wide range of existing VLMs in a plug-and-play fashion. Extensive experiments indicate that the visual reasoner can consistently and significantly improve four VLMs on four VQA benchmarks.

Personalized dialogue systems have gained significant attention in recent years for their ability to generate responses in alignment with different personas.However, most existing approaches rely on pre-defined personal profiles, which are not only time-consuming and labor-intensive to create but also lack flexibility.We propose In-Dialogue Learning (IDL), a fine-tuning framework that enhances the ability of pre-trained large language models to leverage dialogue history to characterize persona for personalized dialogue generation tasks without pre-defined profiles.Our experiments on three datasets demonstrate that IDL brings substantial improvements, with BLEU and ROUGE scores increasing by up to 200% and 247%, respectively.Additionally, the results of human evaluations further validate the efficacy of our proposed method.

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

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.

Predicting the volume of flow from an origin to a destination is essential to understanding the mobility pattern and improving many transportation services, such as ride-hailing and food delivery. However, making long-term prediction for all possible pairs of origins and destinations is still a challenging problem. Existing works either suffer from cumulative error and high complexity of iterative computation on long sequence, or neglect the complex spatial correlation obscured by elongated time span. In this paper, we present CAMLO, a cross-attentive multi-view network for this task. Our model adopts a multi-view framework that separately models the asymmetric characteristics of origin and destination in origin- and destination-oriented view. In each view, a relational graph aggregation module captures the sparse and multi-relational correlation among origins and destinations. Subsequently, a Transformer-based forecasting module is applied to discover intricate temporal dynamics. The two views internally interact with each other via a cross-attention mechanism and are later fused for the final prediction. Extensive experiments conducted on two real-world datasets demonstrate the effectiveness of our model.

The sharing of pathological data is of utmost importance in various applications such as remote diagnosis, graded diagnosis, illness treatment, and specialist system development. However, ensuring reliable, secure, privacy-preserving, and efficient sharing of pathological data pose significant challenges. This paper presents a novel solution that leverages blockchain technology to ensure reliability in pathological data sharing. Additionally, it employs conditional proxy re-encryption (C-PRE) and public key encryption with equality test technology to control the scope and preserve the privacy of shared data. To assess the practicality of our solution, we have implemented a prototype system using Hyperledger Fabric and conducted evaluations with various metrics. We also compared the solution with relevant schemes. The results demonstrate that the proposed solution effectively meets the requirements for pathological data sharing and is practical in production scenarios.

Blockchain is commonly employed in access control to provide safe medical data exchange because of the characteristics of decentralization, nontamperability, and traceability. Patients share personal health data by granting access rights to users or medical institutions. The major purpose of the existing access control techniques is to identify users who are permitted to access medical data. They hardly ever recognize internal assailants from legitimate entities. Medical data will involve multilayer access within the authorized organizations. Considering the cost of permissions management and the problem of insider malicious node attacks, users hope to implement authorization constraints within the authorized institutions. It can prevent their data from being maliciously disclosed by end‐users from different authorized healthcare domains. For the purpose to achieve the fine‐grained permissions propagation control of medical data in sharing institutions, a trust‐based authorization access control mechanism is suggested in this study. Trust thresholds are assigned to different privileges based on their sensitivity and used to generate zero‐knowledge proof to be broadcasted among blockchain nodes. This method evaluates the trust of each user through the dynamic trust calculation model. And meanwhile, smart contract is employed to verify whether the user’s trust can activate some permissions and ensure the privacy of the user’s trust in the process of authorization verification. In addition, the authorization transaction between users and institutions is recorded on the blockchain for patient traceability and accountability. The feasibility and effectiveness of the scheme are demonstrated through comprehensive comparisons and extensive experiments.

Most previous sequential labeling models are task-specific, while recent years have witnessed the rise of generative models due to the advantage of unifying all named entity recognition (NER) tasks into the encoder-decoder framework. Although achieving promising performance, our pilot studies demonstrate that existing generative models are ineffective at detecting entity boundaries and estimating entity types. In this paper, we propose a multi-task Transformer, which incorporates an entity boundary detection task into the named entity recognition task. More concretely, we achieve entity boundary detection by classifying the relations between tokens within the sentence. To improve the accuracy of entity-type mapping during decoding, we adopt an external knowledge base to calculate the prior entity-type distributions and then incorporate the information into the model via the self- and cross-attention mechanisms. We perform experiments on extensive NER benchmarks, including flat, nested, and discontinuous NER datasets involving long entities. It substantially increases nearly$+0.3 \sim +1.5\;{F_1}$scores across a broad spectrum or performs closely to the best generative NER model. Experimental results show that our approach improves the performance of the generative NER model considerably.

Jiduan Liu, Jiahao Liu, Qifan Wang, Jingang Wang, Wei Wu, Yunsen Xian, Dongyan Zhao, Kai Chen, Rui Yan. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Existing controllable dialogue generation work focuses on the single-attribute control and lacks generalization capability to out-of-distribution multiple attribute combinations.In this paper, we explore the compositional generalization for multi-attribute controllable dialogue generation where a model can learn from seen attribute values and generalize to unseen combinations.We propose a prompt-based disentangled controllable dialogue generation model, DCG.It learns attribute concept composition by generating attribute-oriented prompt vectors and uses a disentanglement loss to disentangle different attributes for better generalization.Besides, we design a unified reference-free evaluation framework for multiple attributes with different levels of granularities.Experiment results on two benchmarks prove the effectiveness of our method and the evaluation metric.* The first two authors contribute equally.Weiran Xu is the corresponding author.

Translating natural language queries into SQLs in a seq2seq manner has attracted much attention recently. However, compared with abstract-syntactic-tree-based SQL generation, seq2seq semantic parsers face much more challenges, including poor quality on schematical information prediction and poor semantic coherence between natural language queries and SQLs. This paper analyses the above difficulties and proposes a seq2seq-oriented decoding strategy called SR, which includes a new intermediate representation SSQL and a reranking method with score re-estimator to solve the above obstacles respectively. Experimental results demonstrate the effectiveness of our proposed techniques and T5-SR-3b achieves new state-of-the-art results on the Spider dataset.

Knowledge graphs (KGs) have received increasing attention due to its wide applications on natural language processing. However, its use case on temporal question answering (QA) has not been well-explored. Most of existing methods are developed based on pre-trained language models, which might not be capable to learn temporal-specific presentations of entities in terms of temporal KGQA task. To alleviate this problem, we propose a novel Time-aware Multiway Adaptive (TMA) fusion network. Inspired by the step-by-step reasoning behavior of humans. For each given question, TMA first extracts the relevant concepts from the KG, and then feeds them into a multiway adaptive module to produce a temporal-specific representation of the question. This representation can be incorporated with the pre-trained KG embedding to generate the final prediction. Empirical results verify that the proposed model achieves better performance than the state-of-the-art models in the benchmark dataset. Notably, the Hits@1 and Hits@10 results of TMA on the CronQuestions dataset’s complex questions are absolutely improved by 24% and 10% compared to the best-performing baseline. Furthermore, we also show that TMA employing an adaptive fusion mechanism can provide interpretability by analyzing the proportion of information in question representations.

The task of Points of Interest (POI) description generation aims to generate an objective and informative description for a given POI based on POI-related information. High-quality descriptions can better guide users and improve the performance of POI-related recommendation systems. A practical POI description generation model should have effective multimodal fusion and information encoding methods suitable for various data forms. However, due to model structure and data utilization limitations, the previous method is challenging to meet the above requirements. We propose a novel Guide-Select multimodal fusion method that combines the guiding and selecting process to fuse various POI-related information efficiently. In addition, we propose a reasonable review encoding method and a category encoding method that has strong generalization ability. We integrate these methods into our Guide-Select Generation Model (GSGM). Experimental results demonstrate that our model significantly outperforms the state-of-the-art model while having a strong generalization ability on category information.

Named entity recognition (NER) is an important research problem in natural language processing. There are three types of NER tasks, including flat, nested and discontinuous entity recognition. Most previous sequential labeling models are task-specific, while recent years have witnessed the rising of generative models due to the advantage of unifying all NER tasks into the seq2seq model framework. Although achieving promising performance, our pilot studies demonstrate that existing generative models are ineffective at detecting entity boundaries and estimating entity types. This paper proposes a multi-task Transformer, which incorporates an entity boundary detection task into the named entity recognition task. More concretely, we achieve entity boundary detection by classifying the relations between tokens within the sentence. To improve the accuracy of entity-type mapping during decoding, we adopt an external knowledge base to calculate the prior entity-type distributions and then incorporate the information into the model via the self and cross-attention mechanisms. We perform experiments on an extensive set of NER benchmarks, including two flat, three nested, and three discontinuous NER datasets. Experimental results show that our approach considerably improves the generative NER model’s performance.

Graph neural network (GNN) based recommender systems have become one of the mainstream trends due to the powerful learning ability from user behavior data. Understanding the user intents from behavior data is the key to recommender systems, which poses two basic requirements for GNN-based recommender systems. One is how to learn complex and diverse intents especially when the user behavior is usually inadequate in reality. The other is different behaviors have different intent distributions, so how to establish their relations for a more explainable recommender system. In this paper, we present the Intent-aware Recommendation via Disentangled Graph Contrastive Learning (IDCL), which simultaneously learns interpretable intents and behavior distributions over those intents. Specifically, we first model the user behavior data as a user-item-concept graph, and design a GNN based behavior disentangling module to learn the different intents. Then we propose the intent-wise contrastive learning to enhance the intent disentangling and meanwhile infer the behavior distributions. Finally, the coding rate reduction regularization is introduced to make the behaviors of different intents orthogonal. Extensive experiments demonstrate the effectiveness of IDCL in terms of substantial improvement and the interpretability.

Cross-domain recommendation (CDR) is an effective way to alleviate the data sparsity problem. Content-based CDR is one of the most promising branches since most kinds of products can be described by a piece of text, especially when cold-start users or items have few interactions. However, two vital issues are still under-explored: (1) From the content modeling perspective, sufficient long-text descriptions are usually scarce in a real recommender system, more often the light-weight textual features, such as a few keywords or tags, are more accessible, which is improperly modeled by existing methods. (2) From the CDR perspective, not all inter-domain interests are helpful to infer intra-domain interests. Caused by domain-specific features, there are part of signals benefiting for recommendation in the source domain but harmful for that in the target domain. Therefore, how to distill useful interests is crucial. To tackle the above two problems, we propose a metapath and multi-interest aggregated graph neural network (M2GNN). Specifically, to model the tag-based contents, we construct a heterogeneous information network to hold the semantic relatedness between users, items, and tags in all domains. The metapath schema is predefined according to domain-specific knowledge, with one metapath for one domain. User representations are learned by GNN with a hierarchical aggregation framework, where the intra-metapath aggregation firstly filters out trivial tags and the inter-metapath aggregation further filters out useless interests. Offline experiments and online A/B tests demonstrate that M2GNN achieves significant improvements over the state-of-the-art methods and current industrial recommender system in Dianping, respectively. Further analysis shows that M2GNN offers an interpretable recommendation.

Recommender system usually suffers from severepopularity bias— the collected interaction data usually exhibits quite imbalanced or even long-tailed distribution over items. Such skewed distribution may result from the users’conformityto the group, which deviates from reflecting users’ true preference. Existing efforts for tackling this issue mainly focus on completely eliminating popularity bias. However, we argue that not all popularity bias is evil. Popularity bias not only results from conformity but alsoitem quality, which is usually ignored by existing methods. Some items exhibit higher popularity as they have intrinsic better property. Blindly removing the popularity bias would lose such important signal, and further deteriorate model performance. To sufficiently exploit such important information for recommendation, it is essential to disentangle the benign popularity bias caused by item quality from the harmful popularity bias caused by conformity. Although important, it is quite challenging as we lack an explicit signal to differentiate the two factors of popularity bias. In this paper, we propose to leverage temporal information as the two factors exhibit quite different patterns along the time: item quality revealing item inherent property is stable and static while conformity that depends on items’ recent clicks is highly time-sensitive. Correspondingly, we further propose a novelTime-awareDisEntangled framework (TIDE), where a click is generated from three components namely the static item quality, the dynamic conformity effect, as well as the user-item matching score returned by any recommendation model. Lastly, we conduct interventional inference so that the recommendation can benefit from the benign popularity bias while circumvent the harmful one. Extensive experiments on four real-world datasets demonstrated the effectiveness of TIDE.

Multi-hop question generation focuses on generating complex questions that require reasoning over multiple pieces of information of the input passage.Current models with state-of-the-art performance have been able to generate the correct questions corresponding to the answers.However, most models can not ensure the complexity of generated questions, so they may generate shallow questions that can be answered without multi-hop reasoning.To address this challenge, we propose the CQG, which is a simple and effective controlled framework.CQG employs a simple method to generate the multi-hop questions that contain key entities in multi-hop reasoning chains, which ensure the complexity and quality of the questions.In addition, we introduce a novel controlled Transformer-based decoder to guarantee that key entities appear in the questions.Experiment results show that our model greatly improves performance, which also outperforms the state-of-the-art model about 25% by 5 BLEU points on HotpotQA 1 .

Shengding Hu, Ning Ding, Huadong Wang, Zhiyuan Liu, Jingang Wang, Juanzi Li, Wei Wu, Maosong Sun. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.

Rui Zheng, Bao Rong, Yuhao Zhou, Di Liang, Sirui Wang, Wei Wu, Tao Gui, Qi Zhang, Xuanjing Huang. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.

Dialogue bots have been widely applied in customer service scenarios to provide timely and user-friendly experience. These bots must classify the appropriate domain of a dialogue, understand the intent of users, and generate proper responses. Existing dialogue pre-training models are designed only for several dialogue tasks and ignore weakly-supervised expert knowledge in customer service dialogues. In this paper, we propose a novel unified knowledge prompt pre-training framework, UFA (Unified Model F or All Tasks), for customer service dialogues. We formulate all the tasks of customer service dialogues as a unified text-to-text generation task and introduce a knowledge-driven prompt strategy to jointly learn from a mixture of distinct dialogue tasks. We pre-train UFA on a large-scale Chinese customer service corpus collected from practical scenarios and get significant improvements on both natural language understanding (NLU) and natural language generation (NLG) benchmarks.

Candidate generation task requires that candidates related to user interests need to be extracted in realtime. Previous works usually transform a user's behavior sequence to a unified embedding, which can not reflect the user's multiple interests. Some recent works like Comirec and Octopus use multi-channel structures to capture users' diverse interests. They cluster users' historical behaviors into several groups, claiming that one group represents one interest. However, these methods have some limitations. First, an item may correspond to multiple interests of users, thereby simply allocating it to just one interest group will make the modeling of users' interests coarse-grained and inaccurate. Second, explaining user interests at the level of items is rather vague and not convincing. In this paper, we propose a Knowledge Enhanced Multi-Interest Network: KEMI, which exploits knowledge graphs to help learn users' diverse interest representations via heterogeneous graph neural networks (HGNNs) and a novel dual memory network. Specifically, we use HGNNs to capture the semantic representation of knowledge entities and a novel dual memory network to learn a user's diverse interests from his behavior sequence. Through memory slots of the user memory network and the item memory network, we can learn multiple interests for each user and each item. Meanwhile, by binding the entities to the channels of memory networks, we enable it to be explained from the perspective of the knowledge graph, which enhances the interpretability and understanding of user interests. We conduct extensive experiments on two industrial and publicly available datasets. Experimental results demonstrate that our model achieves significant improvements over state-of-the-art baseline models.

Pre-trained Language Models (PLMs) have achieved remarkable performance gains across numerous downstream tasks in natural language understanding. Various Chinese PLMs have been successively proposed for learning better Chinese language representation. However, most current models use Chinese characters as inputs and are not able to encode semantic information contained in Chinese words. While recent pre-trained models incorporate both words and characters simultaneously, they usually suffer from deficient semantic interactions and fail to capture the semantic relation between words and characters. To address the above issues, we propose a simple yet effective PLM CLOWER, which adopts the Contrastive Learning Over Word and charactER representations. In particular, CLOWER implicitly encodes the coarse-grained information (i.e., words) into the fine-grained representations (i.e., characters) through contrastive learning on multi-grained information. CLOWER is of great value in realistic scenarios since it can be easily incorporated into any existing fine-grained based PLMs without modifying the production pipelines. Extensive experiments conducted on a range of downstream tasks demonstrate the superior performance of CLOWER over several state-of-the-art baselines.

Traditional intent classification models are based on a pre-defined intent set and only recognize limited in-domain (IND) intent classes. But users may input out-of-domain (OOD) queries in a practical dialogue system. Such OOD queries can provide directions for future improvement. In this paper, we define a new task, Generalized Intent Discovery (GID), which aims to extend an IND intent classifier to an open-world intent set including IND and OOD intents. We hope to simultaneously classify a set of labeled IND intent classes while discovering and recognizing new unlabeled OOD types incrementally. We construct three public datasets for different application scenarios and propose two kinds of frameworks, pipeline-based and end-to-end for future work. Further, we conduct exhaustive experiments and qualitative analysis to comprehend key challenges and provide new guidance for future GID research.

Transformer-based pre-trained language models such as BERT have achieved remarkable results in Semantic Sentence Matching. However, existing models still suffer from insufficient ability to capture subtle differences. Minor noise like word addition, deletion, and modification of sentences may cause flipped predictions. To alleviate this problem, we propose a novel Dual Attention Enhanced BERT (DABERT) to enhance the ability of BERT to capture fine-grained differences in sentence pairs. DABERT comprises (1) Dual Attention module, which measures soft word matches by introducing a new dual channel alignment mechanism to model affinity and difference attention. (2) Adaptive Fusion module, this module uses attention to learn the aggregation of difference and affinity features, and generates a vector describing the matching details of sentence pairs. We conduct extensive experiments on well-studied semantic matching and robustness test datasets, and the experimental results show the effectiveness of our proposed method.

Structural bias has recently been exploited for aspect sentiment triplet extraction (ASTE) and led to improved performance. On the other hand, it is recognized that explicitly incorporating structural bias would have a negative impact on efficiency, whereas pretrained language models (PLMs) can already capture implicit structures. Thus, a natural question arises: Is structural bias still a necessity in the context of PLMs? To answer the question, we propose to address the efficiency issues by using an adapter to integrate structural bias in the PLM and using a cheap-to-compute relative position structure in place of the syntactic dependency structure. Benchmarking evaluation is conducted on the SemEval datasets. The results show that our proposed structural adapter is beneficial to PLMs and achieves state-of-the-art performance over a range of strong baselines, yet with a light parameter demand and low latency. Meanwhile, we give rise to the concern that the current evaluation default with data of small scale is under-confident. Consequently, we release a large-scale dataset for ASTE. The results on the new dataset hint that the structural adapter is confidently effective and efficient to a large scale. Overall, we draw the conclusion that structural bias shall still be a necessity even with PLMs.

Adversarial training, which minimizes the loss of adversarially perturbed examples, has received considerable attention. However, these methods require modifying all model parameters and optimizing the model from scratch, which is parameter inefficient and unfriendly to the already deployed models. As an alternative, we propose a pluggable defense module PlugAT, to provide robust predictions by adding a few trainable parameters to the model inputs while keeping the original model frozen. To reduce the potential side effects of using defense modules, we further propose a novel forgetting restricted adversarial training, which filters out bad adversarial examples that impair the performance of original ones. The PlugAT-equipped BERT model substantially improves robustness over several strong baselines on various text classification tasks, whilst training only 9.1% parameters. We observe that defense modules trained under the same model architecture have domain adaptation ability between similar text classification datasets.

Large pre-trained language models (PLMs) have demonstrated superior performance in industrial applications. Recent studies have explored parameter-efficient PLM tuning, which only updates a small amount of task-specific parameters while achieving both high efficiency and comparable performance against standard fine-tuning. However, all these methods ignore the inefficiency problem caused by the task-specific output layers, which is inflexible for us to re-use PLMs and introduces non-negligible parameters. In this work, we focus on the text classification task and propose plugin-tuning, a framework that further improves the efficiency of existing parameter-efficient methods with a unified classifier. Specifically, we re-formulate both token and sentence classification tasks into a unified language modeling task, and map label spaces of different tasks into the same vocabulary space. In this way, we can directly re-use the language modeling heads of PLMs, avoiding introducing extra parameters for different tasks. We conduct experiments on six classification benchmarks. The experimental results show that plugin-tuning can achieve comparable performance against fine-tuned PLMs, while further saving around 50% parameters on top of other parameter-efficient methods.

Text matching is a fundamental research problem in natural language understanding.Interaction-based approaches treat the text pair as a single sequence and encode it through cross encoders, while representation-based models encode the text pair independently with siamese or dual encoders.Interactionbased models require dense computations and thus are impractical in real-world applications.Representation-based models have become the mainstream paradigm for efficient text matching.However, these models suffer from severe performance degradation due to the lack of interactions between the pair of texts.To remedy this, we propose a Virtual InteRacTion mechanism (VIRT) for improving representation-based text matching while maintaining its efficiency.In particular, we introduce an interactive knowledge distillation module that is only applied during training.It enables deep interaction between texts by effectively transferring knowledge from the interaction-based model.A light interaction strategy is designed to fully leverage the learned interactive knowledge.Experimental results on six text matching benchmarks demonstrate the superior performance of our method over several state-of-the-art representationbased models.We further show that VIRT can be integrated into existing methods as plugins to lift their performances.

Prompt tuning learns soft prompts to condition the frozen Pre-trained Language Models (PLMs) for performing downstream tasks in a parameter-efficient manner.While prompt tuning has gradually reached the performance level of fine-tuning as the model scale increases, there is still a large performance gap between prompt tuning and fine-tuning for models of moderate and small scales (typically less than 11B parameters).In this paper, we empirically show that the trained prompt tokens can have a negative impact on a downstream task and thus degrade its performance.To bridge the gap, we propose a novel PROMPT tuning model with an eXtremely small scale (XPROMPT) under the regime of lottery tickets hypothesis.Specifically, XPROMPT eliminates the negative prompt tokens at different granularity levels through a hierarchical structured pruning, yielding a more parameter-efficient prompt yet with a competitive performance.Comprehensive experiments are carried out on the SuperGLUE tasks, and the results indicate that XPROMPT is able to close the performance gap at smaller model scales. 1 Recently, Prompt-Tuning (Lester et al., 2021; Liu et al., 2021b) has been proposed to address this issue by prepending a soft prompt to the input and only updating the parameters of prompt tokens during tuning.Prompt-Tuning provides a parameter-efficient alternative to fine-tuning, since the scale of the soft prompt is tens of thousand smaller.It is also conceptually simpler and more flexible than other parameter-efficient tuning methods (such as Adapters), that require intrusive modifications to transformer layers (Houlsby et al., 2019;Guo et al., 2021).Using fewer tunable parameters, prompt tuning achieves competitive performance to fine-tuning with the increase of the model scale.However, there is still a large performance gap between prompt tuning and fine-tuning for models of smaller scales (as shown in Figure 1).This paper aims to fill the gap, from the perspective of the lottery tickets hypothesis (LTH) (Frankle and Carbin, 2019).We are motivated by an observation that, on a specific task, not all prompt tokens contribute equally to the task performance, while certain prompt tokens may even bring a negative

Discovering out-of-domain (OOD) intent is important for developing new skills in taskoriented dialogue systems.The key challenges lie in how to transfer prior in-domain (IND) knowledge to OOD clustering, as well as jointly learn OOD representations and cluster assignments.Previous methods suffer from indomain overfitting problem, and there is a natural gap between representation learning and clustering objectives.In this paper, we propose a unified K-nearest neighbor contrastive learning framework to discover OOD intents.Specifically, for IND pre-training stage, we propose a KCL objective to learn inter-class discriminative features, while maintaining intraclass diversity, which alleviates the in-domain overfitting problem.For OOD clustering stage, we propose a KCC method to form compact clusters by mining true hard negative samples, which bridges the gap between clustering and representation learning.Extensive experiments on three benchmark datasets show that our method achieves substantial improvements over the state-of-the-art methods. 1

Detecting out-of-domain (OOD) intents from user queries is essential for avoiding wrong operations in task-oriented dialogue systems.The key challenge is how to distinguish indomain (IND) and OOD intents.Previous methods ignore the alignment between representation learning and scoring function, limiting the OOD detection performance.In this paper, we propose a unified neighborhood learning framework (UniNL) to detect OOD intents.Specifically, we design a K-nearest neighbor contrastive learning (KNCL) objective for representation learning and introduce a KNNbased scoring function for OOD detection.We aim to align representation learning with scoring function.Experiments and analysis on two benchmark datasets show the effectiveness of our method.1

Prompt-tuning has shown appealing performance in few-shot classification by virtue of its capability in effectively exploiting pre-trained knowledge.This motivates us to check the hypothesis that prompt-tuning is also a promising choice for long-tailed classification, since the tail classes are intuitively few-shot ones.To achieve this aim, we conduct empirical studies to examine the hypothesis.The results demonstrate that prompt-tuning makes pretrained language models at least good long-tailed learners.For intuitions on why prompt-tuning can achieve good performance in long-tailed classification, we carry out in-depth analyses by progressively bridging the gap between prompttuning and commonly used finetuning.The summary is that the classifier structure and parameterization form the key to making good long-tailed learners, in comparison with the less important input structure.Finally, we verify the applicability of our finding to few-shot classification. 1

A wide range of NLP tasks benefit from the fine-tuning of pretrained language models (PLMs).However, a number of redundant parameters which contribute less to the downstream task are observed in a directly fine-tuned model.We think the gap between pretraining and downstream tasks hinders the training of these redundant parameters, and results in a suboptimal performance of the overall model.In this paper, we present PATS (Perturbation According To Sensitivity), a noisy training mechanism which considers each parameter's importance in the downstream task to help fine-tune PLMs.The main idea of PATS is to add bigger noise to parameters with lower sensitivity and vice versa, in order to activate more parameters' contributions to downstream tasks without affecting the sensitive ones much.Extensive experiments conducted on different tasks of the GLUE benchmark show PATS can consistently empower the fine-tuning of different sizes of PLMs, and the parameters in the well-performing models always have more concentrated distributions of sensitivities, which experimentally proves the effectiveness of our method.

For task-oriented dialogue systems, Natural Language Generation (NLG) is the last and vital step which aims at generating an appropriate response according to the dialogue act (DA). While end-to-end neural networks have achieved promising performances on this task, the existing models still struggle to avoid slot mistakes. To address this challenge, we propose a novel segmented generation approach in this paper. The proposed method operates by progressively generating text for the span between two adjacent keywords (act type and slots) in semantically ordered DA. This procedure is recursively applied from left to right until a response is completed. Besides, a retrieval mechanism is utilized to better match the diversity and fluency in human language. Experimental results on four datasets demonstrate that our model achieves state-of-the-art slot error rate and also gets competitive performance on BLEU score with all strong baselines.

Taxonomies have been widely used in various domains to underpin numerous applications. Specially, product taxonomies serve an essential role in the e-commerce domain for the recommendation, browsing, and query understanding. However, taxonomies need to constantly capture the newly emerged terms or concepts in e-commerce platforms to keep up-to-date, which is expensive and labor-intensive if it relies on manual maintenance and updates. Therefore, we target the taxonomy expansion task to attach new concepts to existing taxonomies automatically. In this paper, we present a self-supervised and user behavior-oriented product taxonomy expansion framework to append new concepts into existing taxonomies. Our framework extracts hyponymy relations that conform to users' intentions and cognition. Specifically, i) to fully exploit user behavioral information, we extract candidate hyponymy relations that match user interests from query-click concepts; ii) to enhance the semantic information of new concepts and better detect hyponymy relations, we model concepts and relations through both user-generated content and structural information in existing taxonomies and user click logs, by leveraging Pre-trained Language Models and Graph Neural Network combined with Contrastive Learning; iii) to reduce the cost of dataset construction and overcome data skews, we construct a high-quality and balanced training dataset from existing taxonomy with no supervision. Extensive experiments on real-world product taxonomies in Meituan Platform, a leading Chinese vertical e-commerce platform to order take-out with more than 70 million daily active users, demonstrate the superiority of our proposed framework over state-of-the-art methods. Notably, our method enlarges the size of real-world product taxonomies from 39,263 to 94,698 relations with 88% precision. Our implementation is available: https://github.com/AdaCheng/Product_Taxonomy_Expansion.

Input distribution shift is one of the vital problems in unsupervised domain adaptation (UDA). The most popular UDA approaches focus on domain-invariant representation learning, trying to align the features from different domains into a similar feature distribution. However, these approaches ignore the direct alignment of input word distributions between domains, which is a vital factor in word-level classification tasks such as cross-domain NER. In this work, we shed new light on cross-domain NER by introducing a subword-level solution, X-Piece, for input word-level distribution shift in NER. Specifically, we re-tokenize the input words of the source domain to approach the target subword distribution, which is formulated and solved as an optimal transport problem. As this approach focuses on the input level, it can also be combined with previous DIRL methods for further improvement. Experimental results show the effectiveness of the proposed method based on BERT-tagger on four benchmark NER datasets. Also, the proposed method is proved to benefit DIRL methods such as DANN.

It is well established that graph neural networks (GNNs) can be interpreted and designed from the perspective of optimization objective. With this clear optimization objective, the deduced GNNs architecture has sound theoretical foundation, which is able to flexibly remedy the weakness of GNNs. However, this optimization objective is only proved for GNNs with single-relational graph. Can we infer a new type of GNNs for multi-relational graphs by extending this optimization objective, so as to simultaneously solve the issues in previous multi-relational GNNs, e.g., over-parameterization? In this paper, we propose a novel ensemble multi-relational GNNs by designing an ensemble multi-relational (EMR) optimization objective. This EMR optimization objective is able to derive an iterative updating rule, which can be formalized as an ensemble message passing (EnMP) layer with multi-relations. We further analyze the nice properties of EnMP layer, e.g., the relationship with multi-relational personalized PageRank. Finally, a new multi-relational GNNs which well alleviate the over-smoothing and over-parameterization issues are proposed. Extensive experiments conducted on four benchmark datasets well demonstrate the effectiveness of the proposed model.

Generating text from abstract meaning repre-sentation (AMR) is a challenging task. Graph-to-sequence (Graph2Seq-based) methods and pre-trained-based methods are proposed for this task. However, both methods have advantages and disadvantages. Graph2Seq-based methods can make use of the structural information of the graph but not the extra knowledge, while pre-trained-based methods have the advantage of the utilization of extra knowledge but may lose the structural information. In addition, both types of methods often suffer from the under- and over-translation problem. To address these prob-lems, we propose a graph structure reconstruction and coverage enhanced model for this task. The graph structure reconstruction uses two auxiliary objectives, relationship prediction and distance prediction of nodes in AMR graphs to enhance the information of graph structure. In addition, we design a coverage mechanism to solve the problem of information under-translation or over-translation in AMR-to-text generation. Experimental results on three datasets show that our proposed method outperforms the existing methods significantly.

The Question Answering over Knowledge Graph (KGQA) task seeks entities (answers) from the Knowledge Graph (KG) in order to answer natural language questions. In practice, KG is often incomplete, with numerous missing links and nodes. With such an incomplete KG, it is tricky to use the semantics inside the KG to get the golden answers, particularly for complex questions. Some current efforts concentrate on using external corpora to overcome KG sparsity; however, identifying and obtaining the corpora is challenging. Other types of work aim to leverage the pre-trained embeddings to resolve the issue but perform slightly worse on complex questions involving numerous triple facts in KG. To address the aforementioned problems, we present a framework CAPKGQA, which transforms Complex KGQA into an n-Ary link Prediction task capable of explicitly modeling complex questions. Furthermore, previous methods also suffer from incomplete KG throughout the candidate answer generation phase. Therefore, we devise an embedding-based retrieval strategy to extract more reliable candidate answers from incomplete KG. Extensive experiments reveal that our approach beats the state-of-the-art models on incomplete and complex KGQA tasks by a significant margin.

Conversational text-to-SQL aims at converting multi-turn natural language queries into their corresponding SQL representations. One of the most intractable problem of conversational text-to-SQL is modeling the semantics of multi-turn queries and gathering proper information required for the current query. This paper shows that explicit modeling the semantic changes by adding each turn and the summarization of the whole context can bring better performance on converting conversational queries into SQLs. In particular, we propose two conversational modeling tasks in both turn grain and conversation grain. These two tasks simply work as auxiliary training tasks to help with multi-turn conversational semantic parsing. We conducted empirical studies and achieve new state-of-the-art results on large-scale open-domain conversational text-to-SQL dataset. The results demonstrate that the proposed mechanism significantly improves the performance of multi-turn semantic parsing.

Knowledge graph (KG) embeddings have been a mainstream approach for reasoning over incomplete KGs. However, limited by their inherently shallow and static architectures, they can hardly deal with the rising focus on complex logical queries, which comprise logical operators, imputed edges, multiple source entities, and unknown intermediate entities. In this work, we present the Knowledge Graph Transformer (kgTransformer) with masked pre-training and fine-tuning strategies. We design a KG triple transformation method to enable Transformer to handle KGs, which is further strengthened by the Mixture-of-Experts (MoE) sparse activation. We then formulate the complex logical queries as masked prediction and introduce a two-stage masked pre-training strategy to improve transferability and generalizability.Extensive experiments on two benchmarks demonstrate that kgTransformer can consistently outperform both KG embedding-based baselines and advanced encoders on nine in-domain and out-of-domain reasoning tasks. Additionally, kgTransformer can reason with explainability via providing the full reasoning paths to interpret given answers.

Yanan Wu, Keqing He, Yuanmeng Yan, QiXiang Gao, Zhiyuan Zeng, Fujia Zheng, Lulu Zhao, Huixing Jiang, Wei Wu, Weiran Xu. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Xueliang Zhao, Tingchen Fu, Chongyang Tao, Wei Wu, Dongyan Zhao, Rui Yan. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Lulu Zhao, Fujia Zheng, Weihao Zeng, Keqing He, Weiran Xu, Huixing Jiang, Wei Wu, Yanan Wu. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Despite the great success of pre-trained language models (PLMs) in a large set of natural language processing (NLP) tasks, there has been a growing concern about their security in real-world applications. Backdoor attack, which poisons a small number of training samples by inserting backdoor triggers, is a typical threat to security. Trained on the poisoned dataset, a victim model would perform normally on benign samples but predict the attacker-chosen label on samples containing pre-defined triggers. The vulnerability of PLMs under backdoor attacks has been proved with increasing evidence in the literature. In this paper, we present several simple yet effective training strategies that could effectively defend against such attacks. To the best of our knowledge, this is the first work to explore the possibility of backdoor-free adaptation for PLMs. Our motivation is based on the observation that, when trained on the poisoned dataset, the PLM's adaptation follows a strict order of two stages: (1) a moderate-fitting stage, where the model mainly learns the major features corresponding to the original task instead of subsidiary features of backdoor triggers, and (2) an overfitting stage, where both features are learned adequately. Therefore, if we could properly restrict the PLM's adaptation to the moderate-fitting stage, the model would neglect the backdoor triggers but still achieve satisfying performance on the original task. To this end, we design three methods to defend against backdoor attacks by reducing the model capacity, training epochs, and learning rate, respectively. Experimental results demonstrate the effectiveness of our methods in defending against several representative NLP backdoor attacks. We also perform visualization-based analysis to attain a deeper understanding of how the model learns different features, and explore the effect of the poisoning ratio. Finally, we explore whether our methods could defend against backdoor attacks for the pre-trained CV model. The codes are publicly available at https://github.com/thunlp/Moderate-fitting.

Cross-domain sentiment classification (CDSC) aims to use the transferable semantics learned from the source domain to predict the sentiment of reviews in the unlabeled target domain. Existing studies in this task attach more attention to the sequence modeling of sentences while largely ignoring the rich domain-invariant semantics embedded in graph structures (i.e., the part-of-speech tags and dependency relations). As an important aspect of exploring characteristics of language comprehension, adaptive graph representations have played an essential role in recent years. To this end, in the paper, we aim to explore the possibility of learning invariant semantic features from graph-like structures in CDSC. Specifically, we present Graph Adaptive Semantic Transfer (GAST) model, an adaptive syntactic graph embedding method that is able to learn domain-invariant semantics from both word sequences and syntactic graphs. More specifically, we first raise a POS-Transformer module to extract sequential semantic features from the word sequences as well as the part-of-speech tags. Then, we design a Hybrid Graph Attention (HGAT) module to generate syntax-based semantic features by considering the transferable dependency relations. Finally, we devise an Integrated aDaptive Strategy (IDS) to guide the joint learning process of both modules. Extensive experiments on four public datasets indicate that GAST achieves comparable effectiveness to a range of state-of-the-art models.

An opinion tag is a sequence of words on a specific aspect of a product or service. Opinion tags reflect key characteristics of product reviews and help users quickly understand their content in e-commerce portals. The task of abstractive opinion tagging has previously been proposed to automatically generate a ranked list of opinion tags for a given review. However, current models for opinion tagging are not personalized, even though personalization is an essential ingredient of engaging user interactions, especially in e-commerce. In this paper, we focus on the task of personalized abstractive opinion tagging. There are two main challenges when developing models for the end-to-end generation of personalized opinion tags: sparseness of reviews and difficulty to integrate multi-type signals, i.e., explicit review signals and implicit behavioral signals. To address these challenges, we propose an end-to-end model, named POT, that consists of three main components: (1) a review-based explicit preference tracker component based on a hierarchical heterogeneous review graph to track user preferences from reviews; (2)a behavior-based implicit preference tracker component using a heterogeneous behavior graph to track the user preferences from implicit behaviors; and (3) a personalized rank-aware tagging component to generate a ranked sequence of personalized opinion tags. In our experiments, we evaluate POT on a real-world dataset collected from e-commerce platforms and the results demonstrate that it significantly outperforms strong baselines.

Traditional dialogue summarization models rely on a large-scale manually-labeled corpus, lacking generalization ability to new domains, and domain adaptation from a labeled source domain to an unlabeled target domain is important in practical summarization scenarios. However, existing domain adaptation works in dialogue summarization generally require large-scale pre-training using extensive external data. To explore the lightweight fine-tuning methods, in this paper, we propose an efficient Adversarial Disentangled Prompt Learning (ADPL) model for domain adaptation in dialogue summarization. We introduce three kinds of prompts including domain-invariant prompt (DIP), domain-specific prompt (DSP), and task-oriented prompt (TOP). DIP aims to disentangle and transfer the shared knowledge from the source domain and target domain in an adversarial way, which improves the accuracy of prediction about domain-invariant information and enhances the ability for generalization to new domains. DSP is designed to guide our model to focus on domain-specific knowledge using domain-related features. TOP is to capture task-oriented knowledge to generate high-quality summaries. Instead of fine-tuning the whole pre-trained language model (PLM), we only update the prompt networks but keep PLM fixed. Experimental results on the zero-shot setting show that the novel design of prompts can yield more coherent, faithful, and relevant summaries than baselines using the prefix-tuning, and perform at par with fine-tuning while being more efficient. Overall, our work introduces a prompt-based perspective to the zero-shot learning for dialogue summarization task and provides valuable findings and insights for future research.

Dialogue systems powered by conversational artificial intelligence (AI) have never been so popular. Interacting with computer through languages reveals a more natural interface to give orders and acquire information---just like human communication. Due to promising potential as virtual assistants and/or social bots, major NLP, AI and even Search & Mining communities are explicitly calling-out for contributions of conversational studies. Learning towards real conversational intelligence is a trip to Mars; perhaps we are yet on Earth. We have achieved substantial progress from recent research outputs. Still we have major obstacles to overcome. In this paper, we present an overview of progress and look forward to future trends so as to shed light on possible directions towards success.

We consider grounding open domain dialogues with images. Existing work assumes that both an image and a textual context are available, but image-grounded dialogues by nature are more difficult to obtain than textual dialogues. Thus, we propose learning a response generation model with both image-grounded dialogues and textual dialogues by assuming that the visual scene information at the time of a conversation can be represented by an image, and trying to recover the latent images of the textual dialogues through text-to-image generation techniques. The likelihood of the two types of dialogues is then formulated by a response generator and an image reconstructor that are learned within a conditional variational auto-encoding framework. Empirical studies are conducted in both image-grounded conversation and text-based conversation. In the first scenario, image-grounded dialogues, especially under a low-resource setting, can be effectively augmented by textual dialogues with latent images; while in the second scenario, latent images can enrich the content of responses and at the same time keep them relevant to contexts.

Hongyin Tang, Xingwu Sun, Beihong Jin, Jingang Wang, Fuzheng Zhang, Wei Wu. 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.

Xi Xiangyu, Wei Ye, Shikun Zhang, Quanxiu Wang, Huixing Jiang, Wei Wu. 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.

Yuanmeng Yan, Rumei Li, Sirui Wang, Fuzheng Zhang, Wei Wu, Weiran Xu. 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.

Knowledge graph completion (KGC) has become a focus of attention across deep learning community owing to its excellent contribution to numerous downstream tasks. Although recently have witnessed a surge of work on KGC, they are still insufficient to accurately capture complex relations, since they adopt the single and static representations. In this work, we propose a novel Disentangled Knowledge Graph Attention Network (DisenKGAT) for KGC, which leverages both micro-disentanglement and macro-disentanglement to exploit representations behind Knowledge graphs (KGs). To achieve micro-disentanglement, we put forward a novel relation-aware aggregation to learn diverse component representation. For macro-disentanglement, we leverage mutual information as a regularization to enhance independence. With the assistance of disentanglement, our model is able to generate adaptive representations in terms of the given scenario. Besides, our work has strong robustness and flexibility to adapt to various score functions. Extensive experiments on public benchmark datasets have been conducted to validate the superiority of DisenKGAT over existing methods in terms of both accuracy and explainability.

The key challenge of question answering over knowledge bases (KBQA) is the inconsistency between the natural language questions and the reasoning paths in the knowledge base (KB).Recent graph-based KBQA methods are good at grasping the topological structure of the graph but often ignore the textual information carried by the nodes and edges.Meanwhile, pre-trained language models learn massive open-world knowledge from the large corpus, but it is in the natural language form and not structured.To bridge the gap between the natural language and the structured KB, we propose three relation learning tasks for BERTbased KBQA, including relation extraction, relation matching, and relation reasoning.By relation-augmented training, the model learns to align the natural language expressions to the relations in the KB as well as reason over the missing connections in the KB.Experiments on WebQSP show that our method consistently outperforms other baselines, especially when the KB is incomplete.

Recent works have shown that powerful pretrained language models (PLM) can be fooled by small perturbations or intentional attacks.To solve this issue, various data augmentation techniques are proposed to improve the robustness of PLMs.However, it is still challenging to augment semantically relevant examples with sufficient diversity.In this work, we present Virtual Data Augmentation (VDA), a general framework for robustly fine-tuning PLMs.Based on the original token embeddings, we construct a multinomial mixture for augmenting virtual data embeddings, where a masked language model guarantees the semantic relevance and the Gaussian noise provides the augmentation diversity.Furthermore, a regularized training strategy is proposed to balance the two aspects.Extensive experiments on six datasets show that our approach is able to improve the robustness of PLMs and alleviate the performance degradation under adversarial attacks.

Event types are hierarchical, yet most existing methods for event detection classify candidate triggers into fine-grained event types directly, without considering the rich semantic correlations in the hierarchy of event types. To fully utilize such information to improve the detection of fine-grained event types, we propose a three-layer label hierarchy and introduce the detection of two coarser-grained types as auxiliary classification tasks. In particular, we leverage the supplementary supervision information from label hierarchy by a novel Logits Mapping (LM) strategy, which generates logits (the intermediate representations fed into classifier) for coarser-grained types by heuristic mapping of logits for fine-grained types. In this way, training signals provided by auxiliary tasks can help the encoder produce more precise logits via back propagation, thus providing a simple (no extra parameter needed) yet effective way to improve the target task. Results of extensive experiments on the ACE 2005 show that LM can not only be easily integrated into the state-of-the-art methods and achieve significant improvement over them, but also can effectively alleviate the data sparseness problem.

Building an intelligent dialogue system capable of naturally and coherently conversing with humans has been a long-standing goal of artificial intelligence. In the past decade, with the development of machine/deep learning technology and the explosive growth of available conversation data in social media, numerous neural models have been developed for context-response matching tasks in retrieval-based dialogue systems, with more fluent and informative responses compared with generative models. This paper presents a comprehensive survey of recent advances in response selection for retrieval-based dialogues. In particular, we first formulate the problem of response selection and review state-of-the-art context-response matching models categorized by their architecture. Then we summarize some recent advances on the research of response selection, including incorporation with extra knowledge and exploration on more effective model learning. Finally, we highlight the challenges which are not yet well addressed in this task and present future research directions.

The paper describes an end-to-end system for optimal Bluetooth beacon deployment in an indoor environment. The system allows accurate indoor positioning in realistic scenarios with signal attenuation from various objects, e.g. walls or furniture. It also satisfies a number of common practical requirements: 1) it computes beacon placement that achieves maximal coverage for a target area within an allocated beacon budget; 2) it allows specifying areas of higher importance whose signal coverage needs to be prioritized, in case of insufficient beacon budget; 3) it allows optimal local-only beacon adjustment in case of layout changes, without the need of re-positioning all previously placed beacons.There are three main components in the proposed system: floorplan representation converter, beacon signal coverage simulator, and beacon placement optimization model. The floorplan representation converter conflates and transforms multi-source inputs into floorplan representations that are used by the signal coverage simulator and the optimization model. The simulator provides realistic radio coverage simulation for the target environment. In the beacon placement optimization model we propose a novel multi-objective loss function, the solution to which is computed through an evolutionary algorithm. We demonstrate the generated optimal placement on synthetic as well as real office building floorplans. The system is compared to random and grid-based beacon placements.

Jiahao Bu, Lei Ren, Shuang Zheng, Yang Yang, Jingang Wang, Fuzheng Zhang, Wei Wu. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2021.

In recent years, with the development of deep learning, text-generation technology has undergone great changes and provided many kinds of services for human beings, such as restaurant reservation and daily communication. The automatically generated text is becoming more and more fluent so researchers begin to consider more anthropomorphic text-generation technology, that is, the conditional text generation, including emotional text generation, personalized text generation, and so on. Conditional Text Generation (CTG) has thus become a research hotspot. As a promising research field, we find that much attention has been paid to exploring it. Therefore, we aim to give a comprehensive review of the new research trends of CTG. We first summarize several key techniques and illustrate the technical evolution route in the field of neural text generation, based on the concept model of CTG. We further make an investigation of existing CTG fields and propose several general learning models for CTG. Finally, we discuss the open issues and promising research directions of CTG.

Building an intelligent dialogue system with the ability to select a proper response according to a multi-turn context is challenging in three aspects: (1) the meaning of a context–response pair is built upon language units from multiple granularities (e.g., words, phrases, and sub-sentences, etc.); (2) local (e.g., a small window around a word) and long-range (e.g., words across the context and the response) dependencies may exist in dialogue data; and (3) the relationship between the context and the response candidate lies in multiple relevant semantic clues or relatively implicit semantic clues in some real cases. However, existing approaches usually encode the dialogue with mono-type representation and the interaction processes between the context and the response candidate are executed in a rather shallow manner, which may lead to an inadequate understanding of dialogue content and hinder the recognition of the semantic relevance between the context and response. To tackle these challenges, we propose a representation [ K ] -interaction [ L ] -matching framework that explores multiple types of deep interactive representations to build context-response matching models for response selection. Particularly, we construct different types of representations for utterance–response pairs and deepen them via alternate encoding and interaction. By this means, the model can handle the relation of neighboring elements, phrasal pattern, and long-range dependencies during the representation and make a more accurate prediction through multiple layers of interactions between the context–response pair. Experiment results on three public benchmarks indicate that the proposed model significantly outperforms previous conventional context-response matching models and achieve slightly better results than the BERT model for multi-turn response selection in retrieval-based dialogue systems.

We study knowledge-grounded dialogue generation with pre-trained language models. To leverage the redundant external knowledge under capacity constraint, we propose equipping response generation defined by a pre-trained language model with a knowledge selection module, and an unsupervised approach to jointly optimizing knowledge selection and response generation with unlabeled dialogues. Empirical results on two benchmarks indicate that our model can significantly outperform state-of-the-art methods in both automatic evaluation and human judgment.

The paper describes Streetside Building Search-Retrieve System (SBSRS) - a system for recognizing buildings from steetside images. SBSRS powers several distinct applications: 1) it improves map-search by enriching its streetview service with semantic information, such as location, business name, open hours, etc.; 2) it enables search by image and location - a novel form of visual image search where both visual and location signals are used to identify the most relevant result to a query image of a building.

While neural conversation models have shown great potentials towards generating informative and engaging responses via introducing external knowledge, learning such a model often requires knowledge-grounded dialogues that are difficult to obtain. To overcome the data challenge and reduce the cost of building a knowledge-grounded dialogue system, we explore the problem under a zero-resource setting by assuming no context-knowledge-response triples are needed for training. To this end, we propose representing the knowledge that bridges a context and a response and the way that the knowledge is expressed as latent variables, and devise a variational approach that can effectively estimate a generation model from independent dialogue corpora and knowledge corpora. Evaluation results on three benchmarks of knowledge-grounded dialogue generation indicate that our model can achieve comparable performance with state-of-the-art methods that rely on knowledge-grounded dialogues for training, and exhibits a good generalization ability over different datasets.

In this paper, we study context-response matching with pre-trained contextualized representations for multi-turn response selection in retrieval-based chatbots. Existing models, such as Cove and ELMo, are trained with limited context (often a single sentence or paragraph), and may not work well on multi-turn conversations, due to the hierarchical nature, informal language, and domain-specific words. To address the challenges, we propose pre-training hierarchical contextualized representations, including contextual word-level and sentence-level representations, by learning a dialogue generation model from large-scale conversations with a hierarchical encoder-decoder architecture. Then the two levels of representations are blended into the input and output layer of a matching model respectively. Experimental results on two benchmark conversation datasets indicate that the proposed hierarchical contextualized representations can bring significantly and consistently improvement to existing matching models for response selection.

We study learning of a matching model for response selection in retrieval-based dialogue systems. The problem is equally important with designing the architecture of a model, but is less explored in existing literature. To learn a robust matching model from noisy training data, we propose a general co-teaching framework with three specific teaching strategies that cover both teaching with loss functions and teaching with data curriculum. Under the framework, we simultaneously learn two matching models with independent training sets. In each iteration, one model transfers the knowledge learned from its training set to the other model, and at the same time receives the guide from the other model on how to overcome noise in training. Through being both a teacher and a student, the two models learn from each other and get improved together. Evaluation results on two public data sets indicate that the proposed learning approach can generally and significantly improve the performance of existing matching models.

Currently, researchers have paid great attention to retrieval-based dialogues in opendomain.In particular, people study the problem by investigating context-response matching for multi-turn response selection based on publicly recognized benchmark data sets.State-of-the-art methods require a response to interact with each utterance in a context from the beginning, but the interaction is performed in a shallow way.In this work, we let utterance-response interaction go deep by proposing an interaction-over-interaction network (IoI).The model performs matching by stacking multiple interaction blocks in which residual information from one time of interaction initiates the interaction process again.Thus, matching information within an utterance-response pair is extracted from the interaction of the pair in an iterative fashion, and the information flows along the chain of the blocks via representations.Evaluation results on three benchmark data sets indicate that IoI can significantly outperform state-of-theart methods in terms of various matching metrics.Through further analysis, we also unveil how the depth of interaction affects the performance of IoI.

We present open domain response generation with meta-words.A meta-word is a structured record that describes various attributes of a response, and thus allows us to explicitly model the one-to-many relationship within open domain dialogues and perform response generation in an explainable and controllable manner.To incorporate meta-words into generation, we enhance the sequence-to-sequence architecture with a goal tracking memory network that formalizes meta-word expression as a goal and manages the generation process to achieve the goal with a state memory panel and a state controller.Experimental results on two large-scale datasets indicate that our model can significantly outperform several state-ofthe-art generation models in terms of response relevance, response diversity, accuracy of oneto-many modeling, accuracy of meta-word expression, and human evaluation.

Jia Li, Chongyang Tao, Wei Wu, Yansong Feng, Dongyan Zhao, Rui Yan. 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.

Ze Yang, Wei Wu, Jian Yang, Can Xu, Zhoujun Li. 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.

Ze Yang, Can Xu, Wei Wu, Zhoujun Li. 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.

Address parsing is a critical step for map search engines. This component annotates the terms of an address query, e.g. house numbers, road names, administrative units etc., so that the address search engine can resolve the expected result. Deep recurrent models achieve state of the art performance for address parsing; however, scaling such models is problematic. They require a significant amount of term-annotated data which is expensive to acquire. In this paper, active learning significantly reduces the amount of labeled data required to train accurate address parsing models. We demonstrate the efficiency of our approach when cold-starting with human-labeled as well as synthetically-generated data.

Inferring user behavior patterns in their daily location visits, i.e., where people go and how long they stay there, enables a variety of useful applications such as time management systems, new location recommendations, and the opportunity for analytics. For example, digital assistants can use inferred daily patterns to automate calendar events for users, or notify users about anticipated traffic conditions to their predicted next location. Retailers, on the other hand, can use the patterns to do location-based recommendations of venues similar or in proximity of the ones anticipated to be visited.

We present a document-grounded matching network (DGMN) for response selection that can power a knowledge-aware retrieval-based chatbot system. The challenges of building such a model lie in how to ground conversation contexts with background documents and how to recognize important information in the documents for matching. To overcome the challenges, DGMN fuses information in a document and a context into representations of each other, and dynamically determines if grounding is necessary and importance of different parts of the document and the context through hierarchical interaction with a response at the matching step. Empirical studies on two public data sets indicate that DGMN can significantly improve upon state-of-the-art methods and at the same time enjoys good interpretability.

Predicting users daily location visits - when and where they will go, and how long they will stay - is key for making effective location-based recommendations. Knowledge of an upcoming day allows the suggestion of relevant alternatives (e.g., a new coffee shop on the way to work) in advance, prior to a visit. This helps users make informed decisions and plan accordingly.

The tutorial is based on our long-term research on open domain conversation, rich hands-on experience on development of Microsoft XiaoIce, and our previous tutorials on EMNLP 2018 and the Web Conference 2019. It starts from a summary of recent achievement made by both academia and industry on chatbots, and then performs a thorough and systematic introduction to state-of-the-art methods for open domain conversation modeling including both retrieval-based methods and generation-based methods. In addition to these, the tutorial also covers some new progress on both groups of methods, such as transition from model design to model learning, transition from knowledge agnostic conversation to knowledge aware conversation, and transition from single-modal conversation to multi-modal conversation. The tutorial is ended by some promising future directions such as how to combine non-task-oriented dialogue systems with task-oriented dialogue systems and how to enhance language learning with chatbots.

We consider context-response matching with multiple types of representations for multi-turn response selection in retrieval-based chatbots. The representations encode semantics of contexts and responses on words, n-grams, and sub-sequences of utterances, and capture both short-term and long-term dependencies among words. With such a number of representations in hand, we study how to fuse them in a deep neural architecture for matching and how each of them contributes to matching. To this end, we propose a multi-representation fusion network where the representations can be fused into matching at an early stage, at an intermediate stage, or at the last stage. We empirically compare different representations and fusing strategies on two benchmark data sets. Evaluation results indicate that late fusion is always better than early fusion, and by fusing the representations at the last stage, our model significantly outperforms the existing methods, and achieves new state-of-the-art performance on both data sets. Through a thorough ablation study, we demonstrate the effect of each representation to matching, which sheds light on how to select them in practical systems.

We study the problem of response selection for multi-turn conversation in retrieval-based chatbots. The task involves matching a response candidate with a conversation context, the challenges for which include how to recognize important parts of the context, and how to model the relationships among utterances in the context. Existing matching methods may lose important information in contexts as we can interpret them with a unified framework in which contexts are transformed to fixed-length vectors without any interaction with responses before matching. This motivates us to propose a new matching framework that can sufficiently carry important information in contexts to matching and model relationships among utterances at the same time. The new framework, which we call a sequential matching framework (SMF), lets each utterance in a context interact with a response candidate at the first step and transforms the pair to a matching vector. The matching vectors are then accumulated following the order of the utterances in the context with a recurrent neural network (RNN) that models relationships among utterances. Context-response matching is then calculated with the hidden states of the RNN. Under SMF, we propose a sequential convolutional network and sequential attention network and conduct experiments on two public data sets to test their performance. Experiment results show that both models can significantly outperform state-of-the-art matching methods. We also show that the models are interpretable with visualizations that provide us insights on how they capture and leverage important information in contexts for matching.

Long text brings a big challenge to neural network based text matching approaches due to their complicated structures. To tackle the challenge, we propose a knowledge enhanced hybrid neural network (KEHNN) that leverages prior knowledge to identify useful information and filter out noise in long text and performs matching from multiple perspectives. The model fuses prior knowledge into word representations by knowledge gates and establishes three matching channels with words, sequential structures of text given by Gated Recurrent Units (GRUs), and knowledge enhanced representations. The three channels are processed by a convolutional neural network to generate high level features for matching, and the features are synthesized as a matching score by a multilayer perceptron. In this paper, we focus on exploring the use of taxonomy knowledge for text matching. Evaluation results from extensive experiments on public data sets of question answering and conversation show that KEHNN can significantly outperform state-of-the-art matching models and particularly improve matching accuracy on pairs with long text.

We study response generation for open domain conversation in chatbots. Existing methods assume that words in responses are generated from an identical vocabulary regardless of their inputs, which not only makes them vulnerable to generic patterns and irrelevant noise, but also causes a high cost in decoding. We propose a dynamic vocabulary sequence-to-sequence (DVS2S) model which allows each input to possess their own vocabulary in decoding. In training, vocabulary construction and response generation are jointly learned by maximizing a lower bound of the true objective with a Monte Carlo sampling method. In inference, the model dynamically allocates a small vocabulary for an input with the word prediction model, and conducts decoding only with the small vocabulary. Because of the dynamic vocabulary mechanism, DVS2S eludes many generic patterns and irrelevant words in generation, and enjoys efficient decoding at the same time. Experimental results on both automatic metrics and human annotations show that DVS2S can significantly outperform state-of-the-art methods in terms of response quality, but only requires 60% decoding time compared to the most efficient baseline.

We study multi-turn response generation in chatbots where a response is generated according to a conversation context. Existing work has modeled the hierarchy of the context, but does not pay enough attention to the fact that words and utterances in the context are differentially important. As a result, they may lose important information in context and generate irrelevant responses. We propose a hierarchical recurrent attention network (HRAN) to model both the hierarchy and the importance variance in a unified framework. In HRAN, a hierarchical attention mechanism attends to important parts within and among utterances with word level attention and utterance level attention respectively.

The 20 Questions (Q20) game is a well known game which encourages deductive reasoning and creativity.In the game, the answerer first thinks of an object such as a famous person or a kind of animal.Then the questioner tries to guess the object by asking 20 questions.In a Q20 game system, the user is considered as the answerer while the system itself acts as the questioner which requires a good strategy of question selection to figure out the correct object and win the game.However, the optimal policy of question selection is hard to be derived due to the complexity and volatility of the game environment.In this paper, we propose a novel policy-based Reinforcement Learning (RL) method, which enables the questioner agent to learn the optimal policy of question selection through continuous interactions with users.To facilitate training, we also propose to use a reward network to estimate the more informative reward.Compared to previous methods, our RL method is robust to noisy answers and does not rely on the Knowledge Base of objects.Experimental results show that our RL method clearly outperforms an entropy-based engineering system and has competitive performance in a noisyfree simulation environment.

Attention mechanism has become a popular and widely used component in sequence-to-sequence models. However, previous research on neural generative dialogue systems always generates universal responses, and the attention distribution learned by the model always attends to the same semantic aspect. To solve this problem, in this paper, we propose a novel Multi-Head Attention Mechanism (MHAM) for generative dialog systems, which aims at capturing multiple semantic aspects from the user utterance. Further, a regularizer is formulated to force different attention heads to concentrate on certain aspects. The proposed mechanism leads to more informative, diverse, and relevant response generated. Experimental results show that our proposed model outperforms several strong baselines.

We consider incorporating topic information into a sequence-to-sequence framework to generate informative and interesting responses for chatbots. To this end, we propose a topic aware sequence-to-sequence (TA-Seq2Seq) model. The model utilizes topics to simulate prior human knowledge that guides them to form informative and interesting responses in conversation, and leverages topic information in generation by a joint attention mechanism and a biased generation probability. The joint attention mechanism summarizes the hidden vectors of an input message as context vectors by message attention and synthesizes topic vectors by topic attention from the topic words of the message obtained from a pre-trained LDA model, with these vectors jointly affecting the generation of words in decoding. To increase the possibility of topic words appearing in responses, the model modifies the generation probability of topic words by adding an extra probability item to bias the overall distribution. Empirical studies on both automatic evaluation metrics and human annotations show that TA-Seq2Seq can generate more informative and interesting responses, significantly outperforming state-of-the-art response generation models.

We study response selection for multiturn conversation in retrieval-based chatbots.Existing work either concatenates utterances in context or matches a response with a highly abstract context vector finally, which may lose relationships among utterances or important contextual information.We propose a sequential matching network (SMN) to address both problems.SMN first matches a response with each utterance in the context on multiple levels of granularity, and distills important matching information from each pair as a vector with convolution and pooling operations.The vectors are then accumulated in a chronological order through a recurrent neural network (RNN) which models relationships among utterances.The final matching score is calculated with the hidden states of the RNN.An empirical study on two public data sets shows that SMN can significantly outperform stateof-the-art methods for response selection in multi-turn conversation.

Image search is a popular application on web search engines. Issuing a location-related query on an image search engine often returns multiple images of maps among the top ranked results. Traditionally, clicking on such images either opens the image in a new browser tab or takes users to a web page containing the image. However, finding the area of intent on an interactive web map (e.g., Bing Maps) is a manual process. In this paper, we describe a novel system, LiveMaps, for analyzing and retrieving an appropriate map viewport for a given image of a map. This provides annotation of images of maps returned by image search engines, allowing users to directly open a link to an interactive map centered on the location of interest.

Image search is a popular application on web search engines. Issuing a location-related query in image search engines often returns multiple images of maps among the top ranked results. Traditionally, clicking on such images either opens the image in a new browser tab or takes users to a web page containing the image. However, finding the area of intent on an interactive web map is a manual process. In this paper, we describe a novel system, LiveMaps, for analyzing and retrieving an appropriate map viewport for a given image of a map. This allows annotation of images of maps returned by image search engines, allowing users to directly open a link to an interactive map centered on the location of interest.

We study tag recommendation for questions in community question answering (CQA). Tags represent the semantic summarization of questions are useful for navigation and expert finding in CQA and can facilitate content consumption such as searching and mining in these web sites. The task is challenging, as both questions and tags are short and a large fraction of tags are tail tags which occur very infrequently. To solve these problems, we propose matching questions and tags not only by themselves, but also by similar questions and similar tags. The idea is then formalized as a model in which we calculate question-tag similarity using a linear combination of similarity with similar questions and tags weighted by tag importance.Question similarity, tag similarity, and tag importance are learned in a supervised random walk framework by fusing multiple features. Our model thus can not only accurately identify question-tag similarity for head tags, but also improve the accuracy of recommendation of tail tags. Experimental results show that the proposed method significantly outperforms state-of-the-art methods on tag recommendation for questions. Particularly, it improves tail tag recommendation accuracy by a large margin.

While automatic response generation for building chatbot systems has drawn a lot of attention recently, there is limited understanding on when we need to consider the linguistic context of an input text in the generation process. The task is challenging, as messages in a conversational environment are short and informal, and evidence that can indicate a message is context dependent is scarce. After a study of social conversation data crawled from the web, we observed that some characteristics estimated from the responses of messages are discriminative for identifying context dependent messages. With the characteristics as weak supervision, we propose using a Long Short Term Memory (LSTM) network to learn a classifier. Our method carries out text representation and classifier learning in a unified framework. Experimental results show that the proposed method can significantly outperform baseline methods on accuracy of classification.

We study the problem of question retrieval in community question answering (CQA). The biggest challenge within this task is lexical gaps between questions since similar questions are usually expressed with different but semantically related words. To bridge the gaps, state-of-the-art methods incorporate extra information such as word-to-word translation and categories of questions into the traditional language models. We find that the existing language model based methods can be interpreted using a new framework, that is they represent words and question categories in a vector space and calculate question-question similarities with a linear combination of dot products of the vectors. The problem is that these methods are either heuristic on data representation or difficult to scale up. We propose a principled and efficient approach to learning representations of data in CQA. In our method, we simultaneously learn vectors of words and vectors of question categories by optimizing an objective function naturally derived from the framework. In question retrieval, we incorporate learnt representations into traditional language models in an effective and efficient way. We conduct experiments on large scale data from Yahoo! Answers and Baidu Knows, and compared our method with state-of-the-art methods on two public data sets. Experimental results show that our method can significantly improve on baseline methods for retrieval relevance. On 1 million training data, our method takes less than 50 minutes to learn a model on a single multicore machine, while the translation based language model needs more than 2 days to learn a translation table on the same machine.

This paper proposes mining query subtopics from questions in community question answering (CQA). The subtopics are represented as a number of clusters of questions with keywords summarizing the clusters. The task is unique in that the subtopics from questions can not only facilitate user browsing in CQA search, but also describe aspects of queries from a question-answering perspective. The challenges of the task include how to group semantically similar questions and how to find keywords capable of summarizing the clusters. We formulate the subtopic mining task as a non-negative matrix factorization (NMF) problem and further extend the model of NMF to incorporate question similarity estimated from metadata of CQA into learning. Compared with existing methods, our method can jointly optimize question clustering and keyword extraction and encourage the former task to enhance the latter. Experimental results on large scale real world CQA datasets show that the proposed method significantly outperforms the existing methods in terms of keyword extraction, while achieving a comparable performance to the state-of-the-art methods for question clustering.

Real-time geocoders help users find precise locations in online mapping systems. Geocoding unstructured queries can be difficult, as users may describe map locations by referencing several spatially co-located entities (e.g., a business near a street intersection). Serving these queries is important as it provides new capabilities and allows for expanding in markets with less structured postal systems. Traditionally, this problem poses significant difficulties for online systems where latency constraints prevent exhaustive join-based algorithms. Previous work in this area involved natural language processing to segment queries based on known rules, or purely spatial approaches that are difficult to maintain and may have high latency. In this paper, we present a new approach to geocoding - BingGC - that makes fulfillment of extremely diverse geocoding queries possible via a combination of traditional web search technologies and a novel algorithm that uses textual search and spatial joins to quickly find results. It allows resolution of up to s spatially co-located entities in a single query with no pre-computation or rule-based matching. We provide experimental analysis of our system compared against leading online geocoders.

This paper studies the problem of question retrieval in community question answering (CQA). To bridge lexical gaps in questions, which is regarded as the biggest challenge in retrieval, state-of-the-art methods learn translation models using answers under an assumption that they are parallel texts. In practice, however, questions and answers are far from "parallel". Indeed, they are heterogeneous for both the literal level and user behaviors. There are a particularly large number of low quality answers, to which the performance of translation models is vulnerable. To address these problems, we propose a supervised question-answer topic modeling approach. The approach assumes that questions and answers share some common latent topics and are generated in a "question language" and "answer language" respectively following the topics. The topics also determine an answer quality signal. Compared with translation models, our approach not only comprehensively models user behaviors on CQA portals, but also highlights the instinctive heterogeneity of questions and answers. More importantly, it takes answer quality into account and performs robustly against noise in answers. With the topic modeling approach, we propose a topic-based language model, which matches questions not only on a term level but also on a topic level. We conducted experiments on large scale data from Yahoo! Answers and Baidu Knows. Experimental results show that the proposed model can significantly outperform state-of-the-art retrieval models in CQA.

Relevant question retrieval and ranking is a typical task in community question answering (CQA). Existing methods mainly focus on long and syntactically structured queries. However, when an input query is short, the task becomes challenging, due to a lack information regarding user intent. In this paper, we mine different types of user intent from various sources for short queries. With these intent signals, we propose a new intent-based language model. The model takes advantage of both state-of-the-art relevance models and the extra intent information mined from multiple sources. We further employ a state-of-the-art learning-to-rank approach to estimate parameters in the model from training data. Experiments show that by leveraging user intent prediction, our model significantly outperforms the state-of-the-art relevance models in question search.

We consider learning query and document similarities from a click-through bipartite graph with metadata on the nodes. The metadata contains multiple types of features of queries and documents. We aim to leverage both the click-through bipartite graph and the features to learn query-document, document-document, and query-query similarities. The challenges include how to model and learn the similarity functions based on the graph data.

A new DAC circuit with multi-threshold voltage for large panel TFT-CLD source driver is proposed based on its binary-tree structural characters. Through setting different bulk voltages VBfor different CMOS analog switches, the threshold voltages and the on-resistance of CMOS analog switches are reduced and the signal transmission speed from resistance network to output buffer is increased greatly. Physical implementation of this structure is simple and no extra components are required. The proposed DAC circuit structure with 10-bit resolution is designed and simulated using 0.35 mum 13.5 V CMOS high-voltage process, the SPICE simulation results show that the step response delay time is reduced from 35 ns to 17.4 ns (by 50%), as compared to conventional DAC structure.