Large Language Models (LLMs) have shown improved generation performance through retrieval-augmented generation (RAG) following the retriever-reader paradigm, which supplements model inputs with externally retrieved knowledge. However, prior work often evaluates RAG holistically, assessing the retriever and reader jointly, making it difficult to isolate the true contribution of retrieval, particularly given the prompt sensitivity of LLMs used as readers. We move beyond perplexity and introduce Spectrum Projection Score (SPS), a lightweight and supervision-free metric that allows the reader to gauge the semantic alignment of a retrieved summary with its hidden representation by comparing the area formed by generated tokens from the summary, and the principal directions of subspace in the reader and to measure the relevance. Building on SPS we present xCompress, an inference‑time controller framework that dynamically samples, ranks, and compresses retrieval summary candidates. Extensive experiments on five QA benchmarks with four open-sourced LLMs show that SPS not only enhances performance across a range of tasks but also provides a principled perspective on the interaction between retrieval and generation.

Large language models (LLMs) have demonstrated strong reasoning capabilities, and as existing approaches for enhancing LLM reasoning continue to mature, increasing attention has shifted toward meta-reasoning as a promising direction for further improvement.However, most existing meta-reasoning methods remain episodic: they focus on executing complex meta-reasoning routines within individual instances, but ignore the accumulation of reusable meta-reasoning skills across instances, leading to recurring failure modes and repeatedly high metacognitive effort.In this paper, we introduce Metacognitive Consolidation, a novel framework in which a model consolidates metacognitive experience from past reasoning episodes into reusable knowledge that improves future meta-reasoning.We instantiate this framework by structuring instance-level problem solving into distinct roles for reasoning, monitoring, and control to generate rich, attributable meta-level traces.These traces are then consolidated through a hierarchical, multi-timescale update mechanism that gradually forms evolving metaknowledge.Experimental results demonstrate consistent performance gains across benchmarks and backbone models, and show that performance improves as metacognitive experience accumulates over time.

Abstract Large Reasoning Models (LRMs) have demonstrated impressive capabilities in many tasks, yet they struggle with reliably following multiple instructions, either by failing to satisfy individual constraints or by struggling to balance competing constraints simultaneously. We formalize this challenge as the Constraint Adherence Problem (CAP). This paper introduces a novel framework that addresses CAP by representing instructions as a structured knowledge graph of constraints. Our approach, Constraint Relationship Graph Completion (CRGC), explicitly models relationships between constraints, identifies adherence challenges, and discovers “bridge constraints” that help the model better focus on and reconcile requirements. Bridge constraints act as auxiliary instructions that make primary constraints more salient and compatible. Unlike existing approaches that enhance instruction following through general training methods, CRGC specifically improves constraint satisfaction by leveraging the model’s own knowledge to create better pathways for generation. Experiments across three popular instruction following datasets demonstrate that our approach reduces constraint violations by 39% compared to standard prompting while maintaining reasoning abilities of large reasoning models.

Learning medical visual representations from image-report pairs through joint learning has garnered increasing research attention due to its potential for transferring acquired knowledge to various downstream medical tasks. Previous works have predominantly focused on instance-wise or token-wise cross-modal alignment, often neglecting the importance of pathological-level consistency. This paper presents a novel framework PLACE that promotes the Pathological-Level Alignment and enriches the fine-grained details via Correlation Exploration without additional human annotations. Specifically, we propose a novel pathological-level cross-modal alignment (PCMA) approach to maximize the consistency of pathology observations from both images and reports. To facilitate this, a Visual Pathology Observation Extractor is introduced to extract visual pathological observation representations from localized tokens. The PCMA module operates independently of any external disease annotations, enhancing the generalizability and robustness of our methods. Furthermore, we design a proxy task that enforces the model to identify correlations among image patches, thereby enriching the fine-grained details crucial for various downstream tasks. Experimental results demonstrate that our proposed framework achieves new state-of-the-art performance on multiple downstream tasks, including classification, image-to-text retrieval, semantic segmentation, object detection and report generation.

In this demo, we present AERA Chat, an automated and explainable educational assessment system designed for interactive and visual evaluations of student responses. This system leverages large language models (LLMs) to generate automated marking and rationale explanations, addressing the challenge of limited explainability in automated educational assessment and the high costs associated with annotation. Our system allows users to input questions and student answers, providing educators and researchers with insights into assessment accuracy and the quality of LLM-assessed rationales. Additionally, it offers advanced visualization and robust evaluation tools, enhancing the usability for educational assessment and facilitating efficient rationale verification.

Predicting unknown drug-drug interactions (DDIs) is crucial for improving medication safety. Previous efforts in DDI prediction have typically focused on binary classification or predicting DDI categories, with the absence of explanatory insights that could enhance trust in these predictions. In this work, we propose to generate natural language explanations for DDI predictions, enabling the model to reveal the underlying pharmacodynamics and pharmacokinetics mechanisms simultaneously as making the prediction. To do this, we have collected DDI explanations from DDInter and DrugBank and developed various models for extensive experiments and analysis. Our models can provide accurate explanations for unknown DDIs between known drugs. This paper contributes new tools to the field of DDI prediction and lays a solid foundation for further research on generating explanations for DDI predictions.

Despite the notable advancements of existing prompting methods, such as In-Context Learning and Chain-of-Thought for Large Language Models (LLMs), they still face challenges related to various biases. Traditional debiasing methods primarily focus on the model training stage, including approaches based on data augmentation and reweighting, yet they struggle with the complex biases inherent in LLMs. To address such limitations, the causal relationship behind the prompting methods is uncovered using a structural causal model, and a novel causal prompting method based on front-door adjustment is proposed to effectively mitigate LLMs biases. In specific, causal intervention is achieved by designing the prompts without accessing the parameters and logits of LLMs. The chain-of-thought generated by LLM is employed as the mediator variable and the causal effect between input prompts and output answers is calculated through front-door adjustment to mitigate model biases. Moreover, to accurately represent the chain-of-thoughts and estimate the causal effects, contrastive learning is used to fine-tune the encoder of chain-of-thought by aligning its space with that of the LLM. Experimental results show that the proposed causal prompting approach achieves excellent performance across seven natural language processing datasets on both open-source and closed-source LLMs.

As Large Language Models (LLMs) are increasingly applied to complex reasoning tasks, achieving both accurate task performance and faithful explanations becomes crucial.However, LLMs often generate unfaithful explanations, partly because they do not consistently adhere closely to the provided context.Existing approaches to this problem either rely on superficial calibration methods, such as decomposed Chain-of-Thought prompting, or require costly retraining to improve model faithfulness.In this work, we propose a probabilistic inference paradigm that leverages taskspecific and lookahead rewards to ensure that LLM-generated rationales are more faithful to model decisions and align better with input context.These rewards are derived from a domainspecific proposal distribution, allowing for optimized sequential Monte Carlo approximations.Our evaluations across three different reasoning tasks show that this method, which allows for controllable generation during inference, improves both accuracy and faithfulness of LLMs.This method offers a promising path towards making LLMs more reliable for reasoning tasks without sacrificing performance.

Key-Value (KV) cache has become a bottleneck of LLMs for long-context generation.Despite the numerous efforts in this area, the optimization for the decoding phase is generally ignored.However, we believe such optimization is crucial, especially for long-output generation tasks based on the following two observations: (i) Excessive compression during the prefill phase which requires specific full context, impairs the comprehension of the reasoning task; (ii) Deviation of heavy hitters 1 occurs in the reasoning tasks with long outputs.Therefore, SCOPE, a simple yet efficient framework that separately performs KV cache optimization during the prefill and decoding phases, is introduced.Specifically, the KV cache during the prefill phase is preserved to maintain the essential information, while a novel strategy based on sliding is proposed to select essential heavy hitters for the decoding phase.Memory usage and memory transfer are further optimized using adaptive and discontinuous strategies.Extensive experiments on LONGGENBENCH show the effectiveness and generalization of SCOPE and its compatibility as a plug-in to other prefill-only KV compression methods. 2

Retrieval-augmented generation (RAG) demonstrates remarkable performance across tasks in open-domain question-answering. However, traditional search engines may retrieve shallow content, limiting the ability of LLMs to handle complex, multi-layered information. To address this, we introduce WebWalkerQA, a benchmark designed to assess the ability of LLMs to perform web traversal. It evaluates the capacity of LLMs to traverse a website’s subpages to extract high-quality data systematically. We propose WebWalker, which is a multi-agent framework that mimics human-like web navigation through an explore-critic paradigm. Extensive experimental results show that WebWalkerQA is challenging and demonstrates the effectiveness of RAG combined with WebWalker, through this horizontal and vertical integration in real-world scenarios.

Large language models (LLMs) have recently pushed open-domain question answering (ODQA) to new frontiers.However, prevailing retriever-reader pipelines often depend on multiple rounds of prompt-level instructions, leading to high computational overhead, instability, and suboptimal retrieval coverage.In this paper, we propose EmbQA, an embedding-level framework that alleviates these shortcomings by enhancing both the retriever and the reader.Specifically, we refine query representations via lightweight linear layers under an unsupervised contrastive learning objective, thereby reordering retrieved passages to highlight those most likely to contain correct answers.Additionally, we introduce an exploratory embedding that broadens the model's latent semantic space to diversify candidate generation and employs an entropy-based selection mechanism to choose the most confident answer automatically.Extensive experiments across three opensource LLMs, three retrieval methods, and four ODQA benchmarks demonstrate that EmbQA substantially outperforms recent baselines in both accuracy and efficiency.

Personalized large language models (LLMs) aim to tailor their outputs to user preferences.Recent advances in parameter-efficient finetuning (PEFT) methods have highlighted the effectiveness of adapting population-level LLMs to personalized LLMs by fine-tuning userspecific parameters with user history.However, user data is typically sparse, making it challenging to adapt LLMs to specific user patterns.To address this challenge, we propose PROgressive PERsonalization (PROPER), a novel progressive learning framework inspired by meso-level theory in social science.PROPER bridges population-level and user-level models by grouping users based on preferences and adapting LLMs in stages.It combines a Mixture-of-Experts (MoE) structure with Low Ranked Adaptation (LoRA), using a user-aware router to assign users to appropriate groups automatically.Additionally, a LoRA-aware router is proposed to facilitate the integration of individual user LoRAs with group-level LoRAs.Experimental results show that PROPER significantly outperforms SOTA models across multiple tasks, demonstrating the effectiveness of our approach.Our code is available at https://github.com/callanwu/PROPER.

As more than 70% of reviews in the existing opinion summary data set are positive, current opinion summarization approaches are hesitant to generate negative summaries given the input of negative texts. To address such sentiment bias, a direct approach without the reliance on a specific structure is to generate additional data based on large language models to balance the emotional distribution of the dataset. However, large-scale data augmentation based on large language models faces an apparent disadvantage, the expensive costs. Therefore, in this paper, we propose LASS, a novel data augmentation framework based on both LArge and Small language models for debiaSing opinion summarization. Specifically, a small number of synthesized negative reviews is obtained by rewriting the positive text via a large language model. Then, a disentangle reconstruction model is trained based on the generated data. After training, a large amount of synthetic data can be obtained by decoding the new representation obtained from the combination of different sample representations and filtering based on perplexity degree and sentiment classification. Experiments have proved that LASS can effectively alleviate emotional bias, similar to using only large models, but in a more economical way.

We address the task of CoREFerence resolution (CoREF) in chunked long narratives. Existing approaches remain either focused on supervised fine-tuning or limited to one-off prediction, which poses a challenge where the context is long. We develop a dynamic approach to cope with this: by deploying dual Large Language Models (LLMs), we assign specialised LLMs to local named entity recognition and distant CoREF tasks, respectively, while ensuring their exchange of information. Utilising our novel memorisation schemes, the coreference resolution LLM would memorise characters and their associated descriptions, thereby reducing token consumption compared with storing previous messages. To alleviate hallucinations of LLMs, we employ an automatic prompt optimisation method, with the LLM ranker modified to leverage annotations. Our approach achieves performance gains over other LLM-based models and fine-tuning approaches on long narrative datasets, significantly reducing the resources required for inference and training.

Although preference optimization methods have improved reasoning performance in Large Language Models (LLMs), they often lack transparency regarding why one reasoning outcome is preferred over another.This limitation is especially critical in Automated Student Answer Scoring (ASAS), where explainability is essential to justify assessment outcomes.Verbal reinforcement learning offers the potential to generate explicit reflection, but it tends to produce superficial critiques that can harm assessment performance.Existing LLMs also struggle to reliably detect subtle reasoning errors in ASAS tasks.Moreover, manually identifying intermediate reasoning errors is expensive and difficult to scale.To address these challenges, we introduce a contrastive reflection synthesis pipeline that generates precise verbal feedback by identifying discrepancies in structure reasoning graph paths.Leveraging these synthetic reflection data, we propose DARS, a Dual-model Reflective Scoring framework featuring a dedicated Critic model trained for effective reflection.DARS achieves strong performance and consistently outperforms existing ASAS baselines across all evaluation metrics.Extensive experiments further provide novel insights into the value of reflection data, framework design, and the scaling behavior of DARS. 1 * Now at Google DeepMind. 1 We release the DARS code at https://github.com/ lijiazheng99/DARS.Student Answer: the mRNA, Is met by tRNA with anticodons, that link to the mRNA's codons.This tRNA carries an amino acid that goes with the mRNA's codon.This is done multiple times until there is a stranhd of amino acids.Ribosomes attach these amino acids together.Reasoner: The student's answer scored 3 points according to the marking rubric, which awards 3 points for addressing four key elements.The student successfully described mRNA exiting the nucleus, tRNA carrying complementary anticodons recognizing the mRNA codons, the corresponding amino acids on tRNA being bonded, and the continuation of amino acid linkage until a stop codon is reached, as indicated by '...tRNA with anticodons, that link to the mRNA's codons.' and '...until there is a strand of amino acids.... Critic: It appears there was a misunderstanding in your assessment of the student's answer.You credited the student for mentioning that the'corresponding amino acids on tRNA are bonded to adjacent tRNA's amino acids' and that 'amino acids continue to be linked until a STOP codon is read on the mRNA.'However, upon reviewing the student's response, these elements were not explicitly covered.The student's description primarily focused on the interaction between mRNA and tRNA and the eventual formation of a peptide bond, but did not detail the process of amino acids being bonded to adjacent tRNA's amino acids nor the specific mechanism of elongation ... Reasoner: The student's answer scored 2 points for achieving three key answer elements.The student accurately described the roles of mRNA and tRNA in protein synthesis, mentioning that 'the mRNA, Is met by tRNA with anticodons, that link to the mRNA's codons' which covers the key elements of mRNA bases being read in triplets and tRNA carrying the complementary anticodon recognizing the complementary codon of the mRNA.Additionally, the student implied the bonding of amino acids by stating 'Ribosomes attach these amino acids together.'However, the answer did not explicitly mention the exit of mRNA from the nucleus, or the elongation and termination at a STOP codon which resulted in losing points for these key elements. Critic: Rationale Looks Good! [STOP]

Recent advances, such as DeepSeek R1-Zero, highlight the effectiveness of incentive training, a reinforcement learning paradigm that computes rewards solely based on the final answer part of a language model's output, thereby encouraging the generation of intermediate reasoning steps.However, these methods fundamentally rely on external verifiers, which limits their applicability to domains like mathematics and coding, where such verifiers are readily available.Although reward models can serve as verifiers, they require high-quality annotated data and are costly to train.In this work, we propose NOVER, NO-VERifier Reinforcement Learning, a general reinforcement learning framework that requires only standard supervised fine-tuning data with no need for an external verifier.NOVER enables incentive training across a wide range of text-to-text tasks and outperforms the model of the same size distilled from large reasoning models such as DeepSeek R1 671B by 7.7%.Moreover, the flexibility of NOVER enables new possibilities for optimizing large language models, such as inverse incentive training.

Chain-of-Thought (CoT) reasoning enhances Large Language Models (LLMs) by encouraging step-by-step reasoning in natural language.However, leveraging a latent continuous space for reasoning may offer benefits in terms of both efficiency and robustness.Prior implicit CoT methods attempt to bypass language completely by reasoning in continuous space but have consistently underperformed compared to the standard explicit CoT approach.We introduce CODI (Continuous Chain-of-Thought via Self-Distillation), a novel training framework that effectively compresses natural language CoT into continuous space.CODI jointly trains a teacher task (Explicit CoT) and a student task (Implicit CoT), distilling the reasoning ability from language into continuous space by aligning the hidden states of a designated token.Our experiments show that CODI is the first implicit CoT approach to match the performance of explicit CoT on GSM8k at the GPT-2 scale, achieving a 3.1x compression rate and outperforming the previous stateof-the-art by 28.2% in accuracy.CODI also demonstrates robustness, generalizable to complex datasets, and interpretability.These results validate that LLMs can reason effectively not only in natural language, but also in a latent continuous space.Code is available at https://github.com/zhenyi4/codi.

Complex narrative contexts often challenge language models' ability to follow instructions, and existing benchmarks fail to capture these difficulties.To address this, we propose Concise-SAE, a training-free framework that improves instruction following by identifying and editing instruction-relevant neurons using only natural language instructions, without requiring labelled data.To thoroughly evaluate our method, we introduce FREEIN-STRUCT, a diverse and realistic benchmark of 1,212 examples that highlights the challenges of instruction following in narrative-rich settings.While initially motivated by complex narratives, Concise-SAE demonstrates stateof-the-art instruction adherence across varied tasks without compromising generation quality.

The alignment of large language models (LLMs) with human preferences remains a key challenge. While post-training techniques like Reinforcement Learning from Human Feedback (RLHF) and Direct Preference Optimization (DPO) have achieved notable success, they often experience computational inefficiencies and training instability. In this paper, we propose Feature-level constrained Preference Optimization (FPO), a novel method designed to simplify the alignment process while ensuring stability. FPO leverages pre-trained Sparse Autoencoders (SAEs) and introduces feature-level constraints, allowing for efficient, sparsity-enforced alignment. Our approach enjoys efficiency by using sparse features activated in a well-trained sparse autoencoder and the quality of sequential KL divergence by using the feature-level offline reference. Experimental results on benchmark datasets demonstrate that FPO achieves an above 5% absolute improvement in win rate with much lower computational cost compared to state-of-the-art baselines, making it a promising solution for efficient and controllable LLM alignments.

Large Language Models (LLMs) struggle with complex reasoning due to limited diversity and inefficient search. We propose Soft Reasoning, an embedding-based search framework that optimises the embedding of the first token to guide generation. It combines (1) embedding perturbation for controlled exploration and (2) Bayesian optimisation to refine embeddings via a verifier-guided objective, balancing exploration and exploitation. This approach improves reasoning accuracy and coherence while avoiding reliance on heuristic search. Experiments demonstrate superior correctness with minimal computation, making it a scalable, model-agnostic solution.

Xingwei Tan, Yuxiang Zhou, Gabriele Pergola, Yulan He. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Sexism has become a pressing issue, driven by the rapid-spreading influence of societal norms, media portrayals, and online platforms that perpetuate and amplify gender biases. Curbing sexism has emerged as a critical challenge globally. Being capable of recognizing sexist statements and behaviors is of particular importance since it is the first step in mind change. This survey provides an extensive overview of recent advancements in sexism detection. We present details of the various resources used in this field and methodologies applied to the task, covering different languages, modalities, models, and approaches. Moreover, we examine the specific challenges these models encounter in accurately identifying and classifying sexism. Additionally, we highlight areas that require further research and propose potential new directions for future exploration in the domain of sexism detection. Through this comprehensive exploration, we strive to contribute to the advancement of interdisciplinary research, fostering a collective effort to combat sexism in its multifaceted manifestations.

In this demo, we present NarrativePlay -- an innovative system enabling users to role-play a fictional character and interact with dynamically generated narrative environments. Unlike existing predefined sandbox approaches, NarrativePlay centres around the main storyline events extracted from the narrative, allowing users to experience the story from the perspective of a character they chose. To design versatile AI agents for diverse scenarios, we employ a framework built on a Large Language Models (LLMs) to extract detailed character traits from text. We also incorporate automatically generated visual displays of narrative settings, character portraits, and character speech, greatly enhancing the overall user experience.

Neural Theory-of-Mind (N-ToM), machine's ability to understand and keep track of the mental states of others, is pivotal in developing socially intelligent agents.However, prevalent N-ToM benchmarks have several shortcomings, including the presence of ambiguous and artificial narratives, absence of personality traits and preferences, a lack of questions addressing characters' psychological mental states, and limited diversity in the questions posed.In response to these issues, we construct Open-ToM, a new benchmark for assessing N-ToM with (1) longer and clearer narrative stories, (2) characters with explicit personality traits, (3) actions that are triggered by character intentions, and (4) questions designed to challenge LLMs' capabilities of modeling characters' mental states of both the physical and psychological world.Using OpenToM, we reveal that state-of-the-art LLMs thrive at modeling certain aspects of mental states in the physical world but fall short when tracking characters' mental states in the psychological world. 1

While Large language models (LLMs) have the capability to iteratively reflect on their own outputs, recent studies have observed their struggles with knowledge-rich problems without access to external resources.In addition to the inefficiency of LLMs in self-assessment, we also observe that LLMs struggle to revisit their predictions despite receiving explicit negative feedback.Therefore, We propose Mirror, a Multiple-perspective self-reflection method for knowledge-rich reasoning, to avoid getting stuck at a particular reflection iteration.Mirror enables LLMs to reflect from multipleperspective clues, achieved through a heuristic interaction between a Navigator and a Reasoner.It guides agents toward diverse yet plausibly reliable reasoning trajectory without access to ground truth by encouraging (1) diversity of directions generated by Navigator and (2) agreement among strategically induced perturbations in responses generated by the Reasoner.The experiments on five reasoning datasets demonstrate that Mirror's superiority over several contemporary self-reflection approaches.Additionally, the ablation study studies clearly indicate that our strategies alleviate the aforementioned challenges.The code is released at https://github.com/hanqi-qi/Mirror.git.

Most existing topic models rely on bag-of-words (BOW) representation, which limits their ability to capture word order information and leads to challenges with out-of-vocabulary (OOV) words in new documents. Contextualized word embeddings, however, show superiority in word sense disambiguation and effectively address the OOV issue. In this work, we introduce a novel neural topic model called the Contextlized Word Topic Model (CWTM), which integrates contextualized word embeddings from BERT. The model is capable of learning the topic vector of a document without BOW information. In addition, it can also derive the topic vectors for individual words within a document based on their contextualized word embeddings. Experiments across various datasets show that CWTM generates more coherent and meaningful topics compared to existing topic models, while also accommodating unseen words in newly encountered documents.

Despite recent progress in automated rumour verification, little has been done on evaluating rumours in a real-world setting. We advance the state-of-the-art on the PHEME dataset, which consists of Twitter response threads collected as a rumour was unfolding. We automatically collect evidence relevant to PHEME and use it to construct knowledge graphs in a time-sensitive manner, excluding information post-dating rumour emergence. We identify discrepancies between the evidence retrieved and PHEME’s labels, which are discussed in detail and amended to release an updated dataset. We develop a novel knowledge graph approach which finds paths linking disjoint fragments of evidence. Our rumour verification model which combines evidence from the graph outperforms the state-of-the-art on PHEME and has superior generisability when evaluated on a temporally distant rumour verification dataset.

As in the existing opinion summary data set, more than 70% are positive texts, the current opinion summarization approaches are reluctant to generate the negative opinion summary given the input of negative opinions. To address such sentiment bias, two approaches are proposed through two perspectives: model-specific and model-agnostic. For the model-specific approach, a variational autoencoder is proposed to disentangle the input representation into sentiment-relevant and sentiment-irrelevant components through adversarial loss. Therefore, the sentiment information in the input is kept and employed for the following decoding which avoids interference of content information with emotional signals. To further avoid relying on some specific opinion summarization frameworks, a model-agnostic approach based on counterfactual data augmentation is proposed. A dataset with a more balanced emotional polarity distribution is constructed using a large pre-trained language model based on some pairwise and mini-edited principles. Experimental results show that the sentiment consistency of the generated summaries is significantly improved using the proposed approaches, while their semantics quality is unaffected.

Direct Preference Optimization (DPO) has emerged as a prominent algorithm for the direct and robust alignment of Large Language Models (LLMs) with human preferences, offering a more straightforward alternative to the complex Reinforcement Learning from Human Feedback (RLHF).Despite its promising efficacy, DPO faces a notable drawback: "verbosity", a common over-optimization phenomenon also observed in RLHF.While previous studies mainly attributed verbosity to biased labels within the data, we propose that the issue also stems from an inherent algorithmic length reliance in DPO.Specifically, we suggest that the discrepancy between sequencelevel Kullback-Leibler (KL) divergences between chosen and rejected sequences, used in DPO, results in overestimated or underestimated rewards due to varying token lengths.Empirically, we utilize datasets with different label lengths to demonstrate the presence of biased rewards.We then introduce an effective downsampling approach, named SamPO, to eliminate potential length reliance.Our experimental evaluations, conducted across three LLMs of varying scales and a diverse array of conditional and open-ended benchmarks, highlight the efficacy of SamPO in mitigating verbosity, achieving improvements of 5% to 12% over DPO through debaised rewards 1 .

The inherent ambiguity of cause and effect boundaries poses a challenge in evaluating causal event extraction tasks.Traditional metrics like Exact Match and BertScore poorly reflect model performance, so we trained evaluation models to approximate human evaluation, achieving high agreement.We used them to perform Reinforcement Learning with extraction models to align them with human preference, prioritising semantic understanding.We successfully explored our approach through multiple datasets, including transferring an evaluator trained on one dataset to another as a way to decrease the reliance on human-annotated data.In that vein, we also propose a weak-to-strong supervision method that uses a fraction of the annotated data to train an evaluation model while still achieving high performance in training an RL model. 1

To better interpret the intrinsic mechanism of large language models (LLMs), recent studies focus on monosemanticity on its basic units.A monosemantic neuron is dedicated to a single and specific concept, which forms a oneto-one correlation between neurons and concepts.Despite extensive research in monosemanticity probing, it remains unclear whether monosemanticity is beneficial or harmful to model capacity.To explore this question, we revisit monosemanticity from the feature decorrelation perspective and advocate for its encouragement.We experimentally observe that the current conclusion by Wang et al. ( 2024), which suggests that decreasing monosemanticity enhances model performance, does not hold when the model changes.Instead, we demonstrate that monosemanticity consistently exhibits a positive correlation with model capacity, in the preference alignment process.Consequently, we apply feature correlation as a proxy for monosemanticity and incorporate a feature decorrelation regularizer into the dynamic preference optimization process.The experiments show that our method not only enhances representation diversity but also improves preference alignment performance 1 .

Understanding in-context learning (ICL) capability that enables large language models (LLMs) to excel in proficiency through demonstration examples is of utmost importance.This importance stems not only from the better utilization of this capability across various tasks, but also from the proactive identification and mitigation of potential risks, including concerns regarding truthfulness, bias, and toxicity, that may arise alongside the capability.In this paper, we present a thorough survey on the interpretation and analysis of in-context learning.First, we provide a concise introduction to the background and definition of in-context learning.Then, we give an overview of advancements from two perspectives: 1) the theoretical perspective, emphasizing studies on mechanistic interpretability and delving into the mathematical foundations behind ICL; and 2) the empirical perspective, concerning studies that empirically analyze factors associated with ICL.We conclude by discussing open questions and the challenges encountered and, by suggesting potential avenues for future research.We believe that our work establishes the basis for further exploration into the interpretation of incontext learning.To aid this effort, we have created a repository 1 containing resources that will be continually updated.1 https://github.com/zyxnlp

The approach of Reinforcement Learning from Human Feedback (RLHF) is widely used for enhancing pre-trained Language Models (LM), enabling them to better align with human preferences. Existing RLHF-based LMs however require complete retraining whenever new queries or feedback are introduced, as human preferences may differ across different domains or topics. LM retraining is often impracticable in most real-world scenarios, due to the substantial time and computational costs involved, as well as data privacy concerns. To address this limitation, we propose Continual Proximal Policy Optimization (CPPO), a novel method that is able to continually align LM with dynamic human preferences. Specifically, CPPO adopts a weighting strategy to decide which samples should be utilized for enhancing policy learning and which should be used for solidifying past experiences. This seeks a good trade-off between policy learning and knowledge retention. Our experimental results show that CPPO outperforms strong Continuous learning (CL) baselines when it comes to consistently aligning with human preferences. Furthermore, compared to PPO, CPPO offers more efficient and stable learning in non-continual scenarios.

Xingwei Tan, Yuxiang Zhou, Gabriele Pergola, Yulan He. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

To seek reliable information sources for news events, we introduce a novel task of expert recommendation, which aims to identify trustworthy sources based on their previously quoted statements. To achieve this, we built a novel dataset, called NewsQuote, consisting of 23,571 quote-speaker pairs sourced from a collection of news articles. We formulate the recommendation task as the retrieval of experts based on their likelihood of being associated with a given query. We also propose a multi-layer ranking framework employing Large Language Models to improve the recommendation performance. Our results show that employing an in-context learning based LLM ranker and a multi-layer ranking-based filter significantly improve both the predictive quality and behavioural quality of the recommender system.

Identifying sarcastic clues from both textual and visual information has become an important research issue, called Multimodal Sarcasm Detection. In this paper, we investigate multimodal sarcasm detection from a novel perspective, where a multimodal graph contrastive learning strategy is proposed to fuse and distinguish the sarcastic clues for textual modality and visual modality. Specifically, we first utilize object detection to derive the crucial visual regions accompanied by their captions of the images, which allows better learning of the key visual regions of visual modality. In addition, to make full use of the semantic information of the visual modality, we employ optical character recognition to extract the textual content in the images. Then, based on image regions, the textual content of visual modality, and the context of the textual modality, we build a multimodal graph for each sample to model the intricate sarcastic relations between modalities. Furthermore, we devise a graph-oriented contrastive learning strategy to leverage the correlations in the same label and differences between different labels, so as to capture better multimodal representations for multimodal sarcasm detection. Extensive experiments show that our method outperforms the previous best baseline models (with a 2.47% improvement in Accuracy, a 1.99% improvement in F-score, and a 2.20% improvement in Macro F-score). The ablation study shows that both multimodal graph structure and graph-oriented contrastive learning are important to our framework. Further, the experiments of using different pre-trained methods show that the proposed multimodal graph contrastive learning framework can directly work with various pre-trained models and achieve outstanding performance in multimodal sarcasm detection.

Radiology report generation (RRG) has gained increasing research attention because of its huge potential to mitigate medical resource shortages and aid the process of disease decision making by radiologists. Recent advancements in Radiology Report Generation (RRG) are largely driven by improving a model's capabilities in encoding single-modal feature representations, while few studies explicitly explore the cross-modal alignment between image regions and words. Radiologists typically focus first on abnormal image regions before composing the corresponding text descriptions, thus cross-modal alignment is of great importance to learn a RRG model which is aware of abnormalities in the image. Motivated by this, we propose a Class Activation Map guided Attention Network (CAMANet) which explicitly promotes cross-modal alignment by employing aggregated class activation maps to supervise cross-modal attention learning, and simultaneously enrich the discriminative information. Experimental results demonstrate that CAMANet outperforms previous SOTA methods on two commonly used RRG benchmarks.

Explainable recommender systems can explain their recommendation decisions, enhancing user trust in the systems. Most explainable recommender systems either rely on human-annotated rationales to train models for explanation generation or leverage the attention mechanism to extract important text spans from reviews as explanations. The extracted rationales are often confined to an individual review and may fail to identify the implicit features beyond the review text. To avoid the expensive human annotation process and to generate explanations beyond individual reviews, we propose to incorporate a geometric prior learnt from user-item interactions into a variational network which infers latent factors from user-item reviews. The latent factors from an individual user-item pair can be used for both recommendation and explanation generation, which naturally inherit the global characteristics encoded in the prior knowledge. Experimental results on three e-commerce datasets show that our model significantly improves the interpretability of a variational recommender using the Wasserstein distance while achieving performance comparable to existing content-based recommender systems in terms of recommendation behaviours.

Document-level multi-event extraction aims to extract the structural information from a given document automatically.Most recent approaches usually involve two steps: (1) modeling entity interactions; (2) decoding entity interactions into events.However, such approaches ignore a global view of inter-dependency of multiple events.Moreover, an event is decoded by iteratively merging its related entities as arguments, which might suffer from error propagation and is computationally inefficient.In this paper, we propose an alternative approach for document-level multi-event extraction with event proxy nodes and Hausdorff distance minimization.The event proxy nodes, representing pseudo-events, are able to build connections with other event proxy nodes, essentially capturing global information.The Hausdorff distance makes it possible to compare the similarity between the set of predicted events and the set of ground-truth events.By directly minimizing Hausdorff distance, the model is trained towards the global optimum directly, which improves performance and reduces training time.Experimental results show that our model outperforms previous state-of-the-art method in F1-score on two datasets with only a fraction of training time. 1 1 Code is available at https://github.com/xnyuwg/procnet... [5] Shenkai Petrochemical Co., Ltd. received the receipt from the company's shareholder, Yexiang Investment Management Co., Ltd. on the evening of November 15, 2016 Regarding the notice of the shares being frozen.... [8] On November 14, 2016, Yexiang Investment received the Notice of Litigation Preservation from the People's Court of Binjiang District, and granted a total of 47,577,481 shares held by Yexiang Investment will be frozen, and the freezing period is from October 31, 2016 to October 30, 2019 ... [10] Yexiang Investment is … holding 47,577,481 shares of the company, accounting for 13.07% of the company's total share capital.... [

In recent years, we have witnessed the shift of paradigms in Natural Language Processing (NLP) from fine-tuning large-scale pre-trained language models (PLMs) on task-specific data to prompt-based learning. In the latter, the task description is embedded into the PLM input, enabling the same model to handle multiple tasks. While both approaches have demonstrated impressive performance in various NLP tasks, their opaque nature makes comprehending their inner workings and decision-making processes challenging for humans.

Human-in-the-loop topic modelling incorporates users' knowledge into the modelling process, enabling them to refine the model iteratively.Recent research has demonstrated the value of user feedback, but there are still issues to consider, such as the difficulty in tracking changes, comparing different models and the lack of evaluation based on real-world examples of use.We developed a novel, interactive human-in-the-loop topic modeling system with a user-friendly interface that enables users compare and record every step they take, and a novel topic words suggestion feature to help users provide feedback that is faithful to the ground truth.Our system also supports not only what traditional topic models can do, i.e., learning the topics from the whole corpus, but also targeted topic modelling, i.e., learning topics for specific aspects of the corpus.In this article, we provide an overview of the system and present the results of a series of user studies designed to assess the value of the system in progressively more realistic applications of topic modelling.

Existing models to extract temporal relations between events lack a principled method to incorporate external knowledge.In this study, we introduce Bayesian-Trans, a Bayesian learningbased method that models the temporal relation representations as latent variables and infers their values via Bayesian inference and translational functions.Compared to conventional neural approaches, instead of performing point estimation to find the best set parameters, the proposed model infers the parameters' posterior distribution directly, enhancing the model's capability to encode and express uncertainty about the predictions.Experimental results on the three widely used datasets show that Bayesian-Trans outperforms existing approaches for event temporal relation extraction.We additionally present detailed analyses on uncertainty quantification, comparison of priors, and ablation studies, illustrating the benefits of the proposed approach.1

Table of contents (ToC) extraction centres on structuring documents in a hierarchical manner.In this paper, we propose a new dataset, ESGDoc, comprising 1,093 ESG annual reports from 563 companies spanning from 2001 to 2022.These reports pose significant challenges due to their diverse structures and extensive length.To address these challenges, we propose a new framework for Toc extraction, consisting of three steps: (1) Constructing an initial tree of text blocks based on reading order and font sizes; (2) Modelling each tree node (or text block) independently by considering its contextual information captured in node-centric subtree; (3) Modifying the original tree by taking appropriate action on each tree node (Keep, Delete, or Move).This construction-modellingmodification (CMM) process offers several benefits.It eliminates the need for pairwise modelling of section headings as in previous approaches, making document segmentation practically feasible.By incorporating structured information, each section heading can leverage both local and long-distance context relevant to itself.Experimental results show that our approach outperforms the previous state-of-theart baseline with a fraction of running time.Our framework proves its scalability by effectively handling documents of any length.1

Automatic multi-hop fact verification task has gained significant attention in recent years.Despite impressive results, these well-designed models perform poorly on out-of-domain data.One possible solution is to augment the training data with counterfactuals, which are generated by minimally altering the causal features of the original data.However, current counterfactual data augmentation techniques fail to handle multi-hop fact verification due to their incapability to preserve the complex logical relationships within multiple correlated texts.In this paper, we overcome this limitation by developing a rationale-sensitive method to generate linguistically diverse and label-flipping counterfactuals while preserving logical relationships.In specific, the diverse and fluent counterfactuals are generated via an Explain-Edit-Generate architecture.Moreover, the checking and filtering modules are proposed to regularize the counterfactual data with logical relations and flipped labels.Experimental results show that the proposed approach outperforms the SOTA baselines and can generate linguistically diverse counterfactual data without disrupting their logical relationships 1 .

Existing solutions to lightweight human pose estimation typically adopt a depthwise separable strategy, i.e., a normal 2D convolution is factorized into channel aggregation and spatial aggregation. However, this strategy cannot well capture multi-scale Effective Receptive Field (ERF), which is essential to dense prediction tasks like human pose estimation. To address this issue, we propose a novel lightweight network for human pose estimation, namely effective aggregation net (EANet). In EANet, we introduce two lightweight computational units: effective channel aggregating (ECA) and effective spatial aggregating (ESA), which are respectively responsible for channel-wise feature aggregation and pixel-wise feature aggregation. Unlike typical channel-wise aggregation using pointwise (1 × 1) convolution, the ECA aggregates few feature points that are estimated as effective ones. Moreover, the ESA is designed with re-parameterizing techniques, and it aggregates effective spatial feature points with multi-scale shared convolutions. Comprehensive experiments are conducted on three challenging datasets, i.e., COCO, Crowd-Pose, Wholebody-COCO. Our EANet demonstrates superior results on human pose estimation over previous lightweight methods, reaching a new state-of-the-art performance with a good trade-off. Our code and models are publicly available1.

For black-box attacks, we utilize the transferability of adversarial examples to attack the unseen model successfully. However, existing attack algorithms are easily trapped into a sharp maximum, where a small perturbation changes its value significantly, leading to the failure of the attack. To tackle this issue, we propose a novel Sharpness-Aware Attack seeking the adversarial example with a flat maximum. Specifically, we sample several poor adversarial examples from the neighborhoods of the current point and alleviate the sharpness between them. We also sample anticipatory neighborhoods examples to make the attack algorithm converge quickly to an excellent starting point. Extensive experiments on the ImageNet dataset show the effectiveness of our method, combined with existing gradient-based attacks, our method yields an average attack success rate of 70.0% for nine defense models.

Counterfactual generation lies at the core of various machine learning tasks, including image translation and controllable text generation. This generation process usually requires the identification of the disentangled latent representations, such as content and style, that underlie the observed data. However, it becomes more challenging when faced with a scarcity of paired data and labelling information. Existing disentangled methods crucially rely on oversimplified assumptions, such as assuming independent content and style variables, to identify the latent variables, even though such assumptions may not hold for complex data distributions. For instance, food reviews tend to involve words like “tasty”, whereas movie reviews commonly contain words such as “thrilling” for the same positive sentiment. This problem is exacerbated when data are sampled from multiple domains since the dependence between content and style may vary significantly over domains. In this work, we tackle the domain-varying dependence between the content and the style variables inherent in the counterfactual generation task. We provide identification guarantees for such latent-variable models by leveraging the relative sparsity of the influences from different latent variables. Our theoretical insights enable the development of a doMain AdapTive counTerfactual gEneration model, called (MATTE). Our theoretically grounded framework achieves state-of-the-art performance in unsupervised style transfer tasks, where neither paired data nor style labels are utilized, across four large-scale datasets.

This full-day tutorial introduces modern techniques for practical uncertainty quantification specifically in the context of multi-class and multi-label text classification. First, we explain the usefulness of estimating aleatoric uncertainty and epistemic uncertainty for text classification models. Then, we describe several state-of-the-art approaches to uncertainty quantification and analyze their scalability to big text data: Virtual Ensemble in GBDT, Bayesian Deep Learning (including Deep Ensemble, Monte-Carlo Dropout, Bayes by Backprop, and their generalization Epistemic Neural Networks), Evidential Deep Learning (including Prior Networks and Posterior Networks), as well as Distance Awareness (including Spectral-normalized Neural Gaussian Process and Deep Deterministic Uncertainty). Next, we talk about the latest advances in uncertainty quantification for pre-trained language models (including asking language models to express their uncertainty, interpreting uncertainties of text classifiers built on large-scale language models, uncertainty estimation in text generation, calibration of language models, and calibration for in-context learning). After that, we discuss typical application scenarios of uncertainty quantification in text classification (including in-domain calibration, cross-domain robustness, and novel class detection). Finally, we list popular performance metrics for the evaluation of uncertainty quantification effectiveness in text classification. Practical hands-on examples/exercises are provided to the attendees for them to experiment with different uncertainty quantification methods on a few real-world text classification datasets such as CLINC150.

Contrastive opinion extraction aims to extract a structured summary or key points organised as positive and negative viewpoints towards a common aspect or topic. Most recent works for unsupervised key point extraction is largely built on sentence clustering or opinion summarisation based on the popularity of opinions expressed in text. However, these methods tend to generate aspect clusters with incoherent sentences, conflicting viewpoints, redundant aspects. To address these problems, we propose a novel unsupervised Contrastive OpinioN Extraction model, called Cone, which learns disentangled latent aspect and sentiment representations based on pseudo aspect and sentiment labels by combining contrastive learning with iterative aspect/sentiment clustering refinement. Apart from being able to extract contrastive opinions, it is also able to quantify the relative popularity of aspects and their associated sentiment distributions. The model has been evaluated on both a hotel review dataset and a Twitter dataset about COVID vaccines. The results show that despite using no label supervision or aspect-denoted seed words, Cone outperforms a number of competitive baselines on contrastive opinion extraction. The results of Cone can be used to offer a better recommendation of products and services online.

Text classifiers built on Pre-trained Language Models (PLMs) have achieved remarkable progress in various tasks including sentiment analysis, natural language inference, and question-answering. However, the occurrence of uncertain predictions by these classifiers poses a challenge to their reliability when deployed in practical applications. Much effort has been devoted to designing various probes in order to understand what PLMs capture. But few studies have delved into factors influencing PLM-based classifiers’ predictive uncertainty. In this paper, we propose a novel framework, called CUE, which aims to interpret uncertainties inherent in the predictions of PLM-based models. In particular, we first map PLM-encoded representations to a latent space via a variational auto-encoder. We then generate text representations by perturbing the latent space which causes fluctuation in predictive uncertainty. By comparing the difference in predictive uncertainty between the perturbed and the original text representations, we are able to identify the latent dimensions responsible for uncertainty and subsequently trace back to the input features that contribute to such uncertainty. Our extensive experiments on four benchmark datasets encompassing linguistic acceptability classification, emotion classification, and natural language inference show the feasibility of our proposed framework. Our source code is available at https://github.com/lijiazheng99/CUE.

Research on automated social media rumour verification, the task of identifying the veracity of questionable information circulating on social media, has yielded neural models achieving high performance, with accuracy scores that often exceed 90%. However, none of these studies focus on the real-world generalisability of the proposed approaches, that is whether the models perform well on datasets other than those on which they were initially trained and tested. In this work we aim to fill this gap by assessing the generalisability of top performing neural rumour verification models covering a range of different architectures from the perspectives of both topic and temporal robustness. For a more complete evaluation of generalisability, we collect and release COVID-RV, a novel dataset of Twitter conversations revolving around COVID-19 rumours. Unlike other existing COVID-19 datasets, our COVID-RV contains conversations around rumours that follow the format of prominent rumour verification benchmarks, while being different from them in terms of topic and time scale, thus allowing better assessment of the temporal robustness of the models. We evaluate model performance on COVID-RV and three popular rumour verification datasets to understand limitations and advantages of different model architectures, training datasets and evaluation scenarios. We find a dramatic drop in performance when testing models on a different dataset from that used for training. Further, we evaluate the ability of models to generalise in a few-shot learning setup, as well as when word embeddings are updated with the vocabulary of a new, unseen rumour. Drawing upon our experiments we discuss challenges and make recommendations for future research directions in addressing this important problem.

Abstract Monitoring online customer reviews is important for business organizations to measure customer satisfaction and better manage their reputations. In this paper, we propose a novel dynamic Brand-Topic Model (dBTM) which is able to automatically detect and track brand-associated sentiment scores and polarity-bearing topics from product reviews organized in temporally ordered time intervals. dBTM models the evolution of the latent brand polarity scores and the topic-word distributions over time by Gaussian state space models. It also incorporates a meta learning strategy to control the update of the topic-word distribution in each time interval in order to ensure smooth topic transitions and better brand score predictions. It has been evaluated on a dataset constructed from MakeupAlley reviews and a hotel review dataset. Experimental results show that dBTM outperforms a number of competitive baselines in brand ranking, achieving a good balance of topic coherence and uniqueness, and extracting well-separated polarity-bearing topics across time intervals.1

The state-of-the-art approaches to targeted aspect-based sentiment analysis (TABSA) are mostly built on deep neural networks with attention mechanisms. One problem is that embeddings of targets and aspects are either pre-trained from large external corpora or randomly initialized. We argue that affective commonsense knowledge and words indicative of sentiment could be used to learn better target and aspect embeddings. We therefore propose an embedding refinement framework calledRAEC(RefiningAffectiveEmbedding fromContext), in which sentiment concepts extracted from affective commonsense knowledge and word relative location information are incorporated to derive context-affective embeddings. Furthermore, a sparse coefficient vector is exploited in refining the embeddings of targets and aspects separately. In this way, embeddings of targets and aspects can capture the highly relevant affective words. Experimental results on two benchmark datasets show that our framework can be easily integrated with existing embedding-based TABSA models and achieves state-of-the-art results compared to models relying on pre-trained word embeddings or built on other embedding refinement methods.

Existing aspect-based/category sentiment analysis methods have shown great success in detecting sentiment polarity toward a given aspect in a sentence with supervised learning, where the training and inference stages share the same pre-defined set of aspects. However, in practice, the aspect categories are changing rather than keeping fixed over time. Dealing with unseen aspect categories is under-explored in existing methods. In this article, we formulate a new few-shot aspect category sentiment analysis (FSACSA) task, which aims to effectively predict the sentiment polarity of previously unseen aspect categories. To this end, we propose a novel Aspect-Focused Meta-Learning (AFML) framework that constructs aspect-aware and aspect-contrastive representations from external knowledge to match the target aspect with aspects in the training set. Concretely, we first construct two auxiliary contrastive sentences for a given sentence with the incorporation of external knowledge, enabling the learning of sentence representations with a better generalization. Then, we devise an aspect-focused induction network to leverage the contextual sentiment toward a given aspect to refine the label vectors. Furthermore, we employ the episode-based meta-learning algorithm to train the whole network, so as to learn to generalize to novel aspects. Extensive experiments on multiple real-life datasets show that our proposed AFML framework achieves the state-of-the-art results for the FSACSA task.

Code summaries help developers comprehend programs and reduce their time to infer the program functionalities during software maintenance. Recent efforts resort to deep learning techniques such as sequence-to-sequence models for generating accurate code summaries, among which Transformer-based approaches have achieved promising performance. However, effectively integrating the code structure information into the Transformer is under-explored in this task domain. In this article, we propose a novel approach named SG-Trans to incorporate code structural properties into Transformer. Specifically, we inject the local symbolic information (e.g., code tokens and statements) and global syntactic structure (e.g., dataflow graph) into the self-attention module of Transformer as inductive bias. To further capture the hierarchical characteristics of code, the local information and global structure are designed to distribute in the attention heads of lower layers and high layers of Transformer. Extensive evaluation shows the superior performance of SG-Trans over the state-of-the-art approaches. Compared with the best-performing baseline, SG-Trans still improves 1.4% and 2.0% on two benchmark datasets, respectively, in terms of METEOR score, a metric widely used for measuring generation quality.

Bin Liang, Chenwei Lou, Xiang Li, Min Yang, Lin Gui, Yulan He, Wenjie Pei, Ruifeng Xu. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2022.

Zero-shot stance detection (ZSSD) aims to detect the stance for an unseen target during the inference stage.In this paper, we propose a joint contrastive learning (JointCL) framework, which consists of stance contrastive learning and target-aware prototypical graph contrastive learning.Specifically, a stance contrastive learning strategy is employed to better generalize stance features for unseen targets.Further, we build a prototypical graph for each instance to learn the target-based representation, in which the prototypes are deployed as a bridge to share the graph structures between the known targets and the unseen ones.Then a novel target-aware prototypical graph contrastive learning strategy is devised to generalize the reasoning ability of target-based stance representations to the unseen targets.Extensive experiments on three benchmark datasets show that the proposed approach achieves state-ofthe-art performance in the ZSSD task 1 .

The primary goal of drug safety researchers and regulators is to promptly identify adverse drug reactions.Doing so may in turn prevent or reduce the harm to patients and ultimately improve public health.Evaluating and monitoring drug safety (i.e., pharmacovigilance) involves analyzing an ever growing collection of spontaneous reports from health professionals, physicians, and pharmacists, and information voluntarily submitted by patients.In this scenario, facilitating analysis of such reports via automation has the potential to rapidly identify safety signals.Unfortunately, public resources for developing natural language models for this task are scant.We present PHEE, a novel dataset for pharmacovigilance comprising over 5000 annotated events from medical case reports and biomedical literature, making it the largest such public dataset to date.We describe the hierarchical event schema designed to provide coarse and fine-grained information about patients' demographics, treatments and (side) effects.Along with the discussion of the dataset, we present a thorough experimental evaluation of current state-of-the-art approaches for biomedical event extraction, point out their limitations, and highlight open challenges to foster future research in this area 1 .

In this paper, we study the task of improving the cohesion and coherence of long-form text generated by language models.To this end, we propose RSTGen, a framework that utilises Rhetorical Structure Theory (RST), a classical language theory, to control the discourse structure, semantics and topics of generated text.Firstly, we demonstrate our model's ability to control structural discourse and semantic features of generated text in open generation evaluation.Then we experiment on the two challenging long-form text tasks of argument generation and story generation.Evaluation using automated metrics and a metric with high correlation to human evaluation, shows that our model performs competitively against existing models, while offering significantly more controls over generated text than alternative methods.

Miguel Arana-Catania, Elena Kochkina, Arkaitz Zubiaga, Maria Liakata, Robert Procter, Yulan He. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Lixing Zhu, Zheng Fang, Gabriele Pergola, Robert Procter, Yulan He. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Token uniformity is commonly observed in transformer-based models, in which different tokens share a large proportion of similar information after going through stacked multiple self-attention layers in a transformer. In this paper, we propose to use the distribution of singular values of outputs of each transformer layer to characterise the phenomenon of token uniformity and empirically illustrate that a less skewed singular value distribution can alleviate the token uniformity problem. Base on our observations, we define several desirable properties of singular value distributions and propose a novel transformation function for updating the singular values. We show that apart from alleviating token uniformity, the transformation function should preserve the local neighbourhood structure in the original embedding space. Our proposed singular value transformation function is applied to a range of transformer-based language models such as BERT, ALBERT, RoBERTa and DistilBERT, and improved performance is observed in semantic textual similarity evaluation and a range of GLUE tasks.

Zero-shot stance detection (ZSSD) is challenging as it requires detecting the stance of previously unseen targets during the inference stage. Being able to detect the target-related transferable stance features from the training data is arguably an important step in ZSSD. Generally speaking, stance features can be grouped into target-invariant and target-specific categories. Target-invariant stance features carry the same stance regardless of the targets they are associated with. On the contrary, target-specific stance features only co-occur with certain targets. As such, it is important to distinguish these two types of stance features when learning stance features of unseen targets. To this end, in this paper, we revisit ZSSD from a novel perspective by developing an effective approach to distinguish the types (target-invariant/-specific) of stance features, so as to better learn transferable stance features. To be specific, inspired by self-supervised learning, we frame the stance-feature-type identification as a pretext task in ZSSD. Furthermore, we devise a novel hierarchical contrastive learning strategy to capture the correlation and difference between target-invariant and -specific features and further among different stance labels. This essentially allows the model to exploit transferable stance features more effectively for representing the stance of previously unseen targets. Extensive experiments on three benchmark datasets show that the proposed framework achieves the state-of-the-art performance in ZSSD.

Abstract Recent years have witnessed increasing interest in developing interpretable models in Natural Language Processing (NLP). Most existing models aim at identifying input features such as words or phrases important for model predictions. Neural models developed in NLP, however, often compose word semantics in a hierarchical manner. As such, interpretation by words or phrases only cannot faithfully explain model decisions in text classification. This article proposes a novel Hierarchical Interpretable Neural Text classifier, called HINT, which can automatically generate explanations of model predictions in the form of label-associated topics in a hierarchical manner. Model interpretation is no longer at the word level, but built on topics as the basic semantic unit. Experimental results on both review datasets and news datasets show that our proposed approach achieves text classification results on par with existing state-of-the-art text classifiers, and generates interpretations more faithful to model predictions and better understood by humans than other interpretable neural text classifiers.1

Recently, advances in neural network approaches have achieved many successes in both sentiment classification and probabilistic topic modeling. On the one hand, latent topics derived from the global context of documents could be helpful in capturing more accurate word semantics and hence could potentially improve the sentiment classification accuracy. On the other hand, the word-level attention vectors obtained during the learning of sentiment classifiers could carry word-level polarity information and can be used to guide the discovery of topics in topic modeling. This paper proposes a multi-task learning framework which jointly learns a sentiment classifier and a topic model by making the word-level latent topic distributions in the topic model to be similar to the word-level attention vectors in sentiment classifiers through mutual learning. Experimental results on the Yelp and IMDB datasets verify the superior performance of the proposed framework over strong baselines on both sentiment classification and topic modeling. The proposed framework also extracts more interpretable topics compared to other conventional topic models and neural topic models.

Previous work has shown the effectiveness of using event representations for tasks such as script event prediction and stock market prediction. It is however still challenging to learn the subtle semantic differences between events based solely on textual descriptions of events often represented as (subject, predicate, object) triples. As an alternative, images offer a more intuitive way of understanding event semantics. We observe that event described in text and in images show different abstraction levels and therefore should be projected onto heterogeneous embedding spaces, as opposed to what have been done in previous approaches which project signals from different modalities onto a homogeneous space. In this paper, we propose a Multimodal Event Representation Learning framework (MERL) to learn event representations based on both text and image modalities simultaneously. Event textual triples are projected as Gaussian density embeddings by a dual-path Gaussian triple encoder, while event images are projected as point embeddings by a visual event component-aware image encoder. Moreover, a novel score function motivated by statistical hypothesis testing is introduced to coordinate two embedding spaces. Experiments are conducted on various multimodal event-related tasks and results show that MERL outperforms a number of unimodal and multimodal baselines, demonstrating the effectiveness of the proposed framework.

Sentiment analysis on user-generated content has achieved notable progress by introducing user information to consider each individual’s preference and language usage. However, most existing approaches ignore the data sparsity problem, where the content of some users is limited and the model fails to capture discriminative features of users. To address this issue, we hypothesize that users could be grouped together based on their rating biases as well as degree of rating consistency and the knowledge learned from groups could be employed to analyze the users with limited data. Therefore, in this paper, a neural group-wise sentiment analysis model with data sparsity awareness is proposed. The user-centred document representations are generated by incorporating a group-based user encoder. Furthermore, a multi-task learning framework is employed to jointly modelusers’ rating biases and their degree of rating consistency. One task is vanilla populationlevel sentiment analysis and the other is groupwise sentiment analysis. Experimental results on three real-world datasets show that the proposed approach outperforms some state-of the-art methods. Moreover, model analysis and case study demonstrate its effectiveness of modeling user rating biases and variances.

Jiasheng Si, Deyu Zhou, Tongzhe Li, Xingyu Shi, Yulan He. 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.

Hanqi Yan, Lin Gui, Gabriele Pergola, Yulan He. 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.

Lixing Zhu, Gabriele Pergola, Lin Gui, Deyu Zhou, Yulan He. 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.

As the digital news industry becomes the main channel of information dissemination, the adverse impact of fake news is explosively magnified. The credibility of a news report should not be considered in isolation. Rather, previously published news articles on the similar event could be used to assess the credibility of a news report. Inspired by this, we propose a BERT-based multimodal unreliable news detection framework, which captures both textual and visual information from unreliable articles utilising the contrastive learning strategy. The contrastive learner interacts with the unreliable news classifier to push similar credible news (or similar unreliable news) closer while moving news articles with similar content but opposite credibility labels away from each other in the multimodal embedding space. Experimental results on a COVID-19 related dataset, ReCOVery, show that our model outperforms a number of competitive baseline in unreliable news detection.

Biomedical question-answering (QA) has gained increased attention for its capability to provide users with high-quality information from a vast scientific literature.Although an increasing number of biomedical QA datasets has been recently made available, those resources are still rather limited and expensive to produce.Transfer learning via pre-trained language models (LMs) has been shown as a promising approach to leverage existing general-purpose knowledge.However, finetuning these large models can be costly and time consuming, often yielding limited benefits when adapting to specific themes of specialised domains, such as the COVID-19 literature.To bootstrap further their domain adaptation, we propose a simple yet unexplored approach, which we call biomedical entity-aware masking (BEM).We encourage masked language models to learn entity-centric knowledge based on the pivotal entities characterizing the domain at hand, and employ those entities to drive the LM fine-tuning.The resulting strategy is a downstream process applicable to a wide variety of masked LMs, not requiring additional memory or components in the neural architectures.Experimental results show performance on par with state-of-the-art models on several biomedical QA datasets.

In this paper, we propose the Brand-Topic Model (BTM) which aims to detect brandassociated polarity-bearing topics from product reviews.Different from existing models for sentiment-topic extraction which assume topics are grouped under discrete sentiment categories such as 'positive', 'negative' and 'neural', BTM is able to automatically infer real-valued brand-associated sentiment scores and generate fine-grained sentiment-topics in which we can observe continuous changes of words under a certain topic (e.g., 'shaver' or 'cream') while its associated sentiment gradually varies from negative to positive.BTM is built on the Poisson factorisation model with the incorporation of adversarial learning.It has been evaluated on a dataset constructed from Amazon reviews.Experimental results show that BTM outperforms a number of competitive baselines in brand ranking, achieving a better balance of topic coherence and uniqueness, and extracting better-separated polaritybearing topics.

Detecting events and their evolution through time is a crucial task in natural language understanding.Recent neural approaches to event temporal relation extraction typically map events to embeddings in the Euclidean space and train a classifier to detect temporal relations between event pairs.However, embeddings in the Euclidean space cannot capture richer asymmetric relations such as event temporal relations.We thus propose to embed events into hyperbolic spaces, which are intrinsically oriented at modeling hierarchical structures.We introduce two approaches to encode events and their temporal relations in hyperbolic spaces.One approach leverages hyperbolic embeddings to directly infer event relations through simple geometrical operations.In the second one, we devise an end-to-end architecture composed of hyperbolic neural units tailored for the temporal relation extraction task.Thorough experimental assessments on widely used datasets have shown the benefits of revisiting the tasks on a different geometrical space, resulting in state-of-the-art performance on several standard metrics.Finally, the ablation study and several qualitative analyses highlighted the rich event semantics implicitly encoded into hyperbolic spaces. 1

Multi-label document classification, associating one document instance with a set of relevant labels, is attracting more and more research attention.Existing methods explore the incorporation of information beyond text, such as document metadata or label structure.These approaches however either simply utilize the semantic information of metadata or employ the predefined parent-child label hierarchy, ignoring the heterogeneous graphical structures of metadata and labels, which we believe are crucial for accurate multi-label document classification.Therefore, in this paper, we propose a novel neural network based approach for multi-label document classification, in which two heterogeneous graphs are constructed and learned using heterogeneous graph transformers.One is metadata heterogeneous graph, which models various types of metadata and their topological relations.The other is label heterogeneous graph, which is constructed based on both the labels' hierarchy and their statistical dependencies.Experimental results on two benchmark datasets show the proposed approach outperforms several stateof-the-art baselines.

Implicit sentiment analysis, aiming at detecting the sentiment of a sentence without sentiment words, has become an attractive research topic in recent years.In this paper, we focus on event-centric implicit sentiment analysis that utilizes the sentiment-aware event contained in a sentence to infer its sentiment polarity.Most existing methods in implicit sentiment analysis simply view noun phrases or entities in text as events or indirectly model events with sophisticated models.Since events often trigger sentiments in sentences, we argue that this task would benefit from explicit modeling of events and event representation learning.To this end, we represent an event as the combination of its event type and the event triplet .Based on such event representation, we further propose a novel model with hierarchical tensor-based composition mechanism to detect sentiment in text.In addition, we present a dataset 1 for event-centric implicit sentiment analysis where each sentence is labeled with the event representation described above.Experimental results on our constructed dataset and an existing benchmark dataset show the effectiveness of the proposed approach.

The flexibility of the inference process in Variational Autoencoders (VAEs) has recently led to revising traditional probabilistic topic models giving rise to Neural Topic Models (NTM).Although these approaches have achieved significant results, surprisingly very little work has been done on how to disentangle the latent topics.Existing topic models when applied to reviews may extract topics associated with writers' subjective opinions mixed with those related to factual descriptions such as plot summaries in movie and book reviews.It is thus desirable to automatically separate opinion topics from plot/neutral ones enabling a better interpretability.In this paper, we propose a neural topic model combined with adversarial training to disentangle opinion topics from plot and neutral ones.We conduct an extensive experimental assessment introducing a new collection of movie and book reviews paired with their plots, namely MOBO dataset, showing an improved coherence and variety of topics, a consistent disentanglement rate, and sentiment classification performance superior to other supervised topic models.

Detecting sarcastic expressions could promote the understanding of natural language in social media. In this paper, we revisit sarcasm detection from a novel perspective, so as to account for the long-range literal sentiment inconsistencies. More concretely, we explore a novel scenario of constructing an affective graph and a dependency graph for each sentence based on the affective information retrieved from external affective commonsense knowledge and the syntactical information of the sentence. Based on it, an Affective Dependency Graph Convolutional Network (ADGCN) framework is proposed to draw long-range incongruity patterns and inconsistent expressions over the context for sarcasm detection by means with interactively modeling the affective and dependency information. Experimental results on multiple benchmark datasets show that our proposed approach outperforms the current state-of-the-art methods in sarcasm detection.

Target plays an essential role in stance detection of an opinionated review/claim, since the stance expressed in the text often depends on the target. In practice, we need to deal with targets unseen in the annotated training data. As such, detecting stance for an unknown or unseen target is an important research problem. This paper presents a novel approach that automatically identifies and adapts the target-dependent and target-independent roles that a word plays with respect to a specific target in stance expressions, so as to achieve cross-target stance detection. More concretely, we explore a novel solution of constructing heterogeneous target-adaptive pragmatics dependency graphs (TPDG) for each sentence towards a given target. An in-target graph is constructed to produce inherent pragmatics dependencies of words for a distinct target. In addition, another cross-target graph is constructed to develop the versatility of words across all targets for boosting the learning of dominant word-level stance expressions available to an unknown target. A novel graph-aware model with interactive Graphical Convolutional Network (GCN) blocks is developed to derive the target-adaptive graph representation of the context for stance detection. The experimental results on a number of benchmark datasets show that our proposed model outperforms state-of-the-art methods in cross-target stance detection.

Abstract Climate models (CM) are used to evaluate the impact of climate change on the risk of floods and heavy precipitation events. However, these numerical simulators produce outputs with low spatial resolution that exhibit difficulties representing precipitation events accurately. This is mainly due to computational limitations on the spatial resolution used when simulating multi-scale weather dynamics in the atmosphere. To improve the prediction of high resolution precipitation we apply a Deep Learning (DL) approach using input data from a reanalysis product, that is comparable to a climate model’s output, but can be directly related to precipitation observations at a given time and location. Further, our input excludes local precipitation, but includes model fields (weather variables) that are more predictable and generalizable than local precipitation. To this end, we present TRU-NET (Temporal Recurrent U-Net), an encoder-decoder model featuring a novel 2D cross attention mechanism between contiguous convolutional-recurrent layers to effectively model multi-scale spatio-temporal weather processes. We also propose a non-stochastic variant of the conditional-continuous (CC) loss function to capture the zero-skewed patterns of rainfall. Experiments show that our models, trained with our CC loss, consistently attain lower RMSE and MAE scores than a DL model prevalent in precipitation downscaling and outperform a state-of-the-art dynamical weather model. Moreover, by evaluating the performance of our model under various data formulation strategies, for the training and test sets, we show that there is enough data for our deep learning approach to output robust, high-quality results across seasons and varying regions.

Recent years have witnessed a surge of interests of using neural topic models for automatic topic extraction from text, since they avoid the complicated mathematical derivations for model inference as in traditional topic models such as Latent Dirichlet Allocation (LDA).However, these models either typically assume improper prior (e.g.Gaussian or Logistic Normal) over latent topic space or could not infer topic distribution for a given document.To address these limitations, we propose a neural topic modeling approach, called Bidirectional Adversarial Topic (BAT) model, which represents the first attempt of applying bidirectional adversarial training for neural topic modeling.The proposed BAT builds a twoway projection between the document-topic distribution and the document-word distribution.It uses a generator to capture the semantic patterns from texts and an encoder for topic inference.Furthermore, to incorporate word relatedness information, the Bidirectional Adversarial Topic model with Gaussian (Gaussian-BAT) is extended from BAT.To verify the effectiveness of BAT and Gaussian-BAT, three benchmark corpora are used in our experiments.The experimental results show that BAT and Gaussian-BAT obtain more coherent topics, outperforming several competitive baselines.Moreover, when performing text clustering based on the extracted topics, our models outperform all the baselines, with more significant improvements achieved by Gaussian-BAT where an increase of near 6% is observed in accuracy.

Opinion prediction on Twitter is challenging due to the transient nature of tweet content and neighbourhood context.In this paper, we model users' tweet posting behaviour as a temporal point process to jointly predict the posting time and the stance label of the next tweet given a user's historical tweet sequence and tweets posted by their neighbours.We design a topic-driven attention mechanism to capture the dynamic topic shifts in the neighbourhood context.Experimental results show that the proposed model predicts both the posting time and the stance labels of future tweets more accurately compared to a number of competitive baselines.

In most previous studies, the aspect-related text is considered an important clue for the Aspect-based Sentiment Analysis (ABSA) task, and thus various attention mechanisms have been proposed to leverage the interactions between aspects and context. However, it is observed that some sentiment expressions carry the same polarity regardless of the aspects they are associated with. In such cases, it is not necessary to incorporate aspect information for ABSA. More observations on the experimental results show that blindly leveraging interactions between aspects and context as features may introduce noises when analyzing those aspect-invariant sentiment expressions, especially when the aspect-related annotated data is insufficient. Hence, in this paper, we propose an Adversarial Multi-task Learning framework to identify the aspect-invariant/dependent sentiment expressions without extra annotations. In addition, we adopt a gating mechanism to control the contribution of representations derived from aspect-invariant and aspect-dependent hidden states when generating the final contextual sentiment representations for the given aspect. This essentially allows the exploitation of aspect-invariant sentiment features for better ABSA results. Experimental results on two benchmark datasets show that extending existing neural models using our proposed framework achieves superior performance. In addition, the aspect-invariant data extracted by the proposed framework can be considered as pivot features for better transfer learning of the ABSA models on unseen aspects.

We introduce CHIME, a cross-passage hierarchical memory network for question answering (QA) via text generation.It extends XLNet (Yang et al., 2019) introducing an auxiliary memory module consisting of two components: the context memory collecting cross-passage evidence, and the answer memory working as a buffer continually refining the generated answers.Empirically, we show the efficacy of the proposed architecture in the multi-passage generative QA, outperforming the state-of-the-art baselines with better syntactically well-formed answers and increased precision in addressing the questions of the AmazonQA review dataset.An additional qualitative analysis revealed the interpretability introduced by the memory module.

In our world with full of uncertainty, debates and argumentation contribute to the progress of science and society. Despite of the increasing attention to characterize human arguments, most progress made so far focus on the debate outcome, largely ignoring the dynamic patterns in argumentation processes. This paper presents a study that automatically analyzes the key factors in argument persuasiveness, beyond simply predicting who will persuade whom. Specifically, we propose a novel neural model that is able to dynamically track the changes of latent topics and discourse in argumentative conversations, allowing the investigation of their roles in influencing the outcomes of persuasion. Extensive experiments have been conducted on argumentative conversations on both social media and supreme court. The results show that our model outperforms state-of-the-art models in identifying persuasive arguments via explicitly exploring dynamic factors of topic and discourse. We further analyze the effects of topics and discourse on persuasiveness, and find that they are both useful — topics provide concrete evidence while superior discourse styles may bias participants, especially in social media arguments. In addition, we draw some findings from our empirical results, which will help people better engage in future persuasive conversations.

Storyline generation aims to produce a concise summary of related events unfolding over time from a collection of news articles. It can be cast into an evolutionary clustering problem by separating news articles into different epochs. Existing unsupervised approaches to storyline generation are typ ically based on probabilistic graphical models. They assume that the storyline distribution at the current epoch depends on the weighted combination of storyline distributions in the latest previous M epochs. The evolutionary parameters of such long-term dependency are typically set by a fixed exponential decay function to capture the intuition that events in more recent epochs have stronger influence to the storyline generation in the current epoch. However, we argue that the amount of relevant historical contextual information should vary for different storylines. Therefore, in this paper, we propose a new Dynamic Dependency Storyline Extraction Model (D2SEM) in which the dependencies among events in different epochs but belonging to the same storyline are dynamically updated to track the time-varying distributions of storylines over time. The proposed model has been evaluated on three news corpora and the experimental results show that it outperforms the state-of-the-art approaches and is able to capture the dependency on historical contextual information dynamically.

Research into cyberbullying detection has increased in recent years, due in part to the proliferation of cyberbullying across social media and its detrimental effect on young people. A growing body of work is emerging on automated approaches to cyberbullying detection. These approaches utilise machine learning and natural language processing techniques to identify the characteristics of a cyberbullying exchange and automatically detect cyberbullying by matching textual data to the identified traits. In this paper, we present a systematic review of published research (as identified via Scopus, ACM and IEEE Xplore bibliographic databases) on cyberbullying detection approaches. On the basis of our extensive literature review, we categorise existing approaches into 4 main classes, namely supervised learning, lexicon-based, rule-based, and mixed-initiative approaches. Supervised learning-based approaches typically use classifiers such as SVM and Naıve Bayes to develop predictive models for cyberbullying detection. Lexicon-based systems utilise word lists and use the presence of words within the lists to detect cyberbullying. Rule-based approaches match text to predefined rules to identify bullying, and mixed-initiatives approaches combine human-based reasoning with one or more of the aforementioned approaches. We found lack of labelled datasets and non-holistic consideration of cyberbullying by researchers when developing detection systems are two key challenges facing cyberbullying detection research. This paper essentially maps out the state-of-the-art in cyberbullying detection research and serves as a resource for researchers to determine where to best direct their future research efforts in this field.

In this paper, we propose a novel variational generator framework for conditional GANs to catch semantic details for improving the generation quality and diversity. Traditional generators in conditional GANs simply concatenate the conditional vector with the noise as the input representation, which is directly employed for upsampling operations. However, the hidden condition information is not fully exploited, especially when the input is a class label. Therefore, we introduce a variational inference into the generator to infer the posterior of latent variable only from the conditional input, which helps achieve a variable augmented representation for image generation. Qualitative and quantitative experimental results show that the proposed method outperforms the state-of-the-art approaches and achieves the realistic controllable images.

Lin Gui, Jia Leng, Gabriele Pergola, Yu Zhou, Ruifeng Xu, Yulan He. 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.

Rui Wang, Deyu Zhou, Yulan He. 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.

Yang Yang, Deyu Zhou, Yulan He, Meng Zhang. 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.

This paper presents an unsupervised framework for jointly modeling topic content and discourse behavior in microblog conversations. Concretely, we propose a neural model to discover word clusters indicating what a conversation concerns (i.e., topics) and those reflecting how participants voice their opinions (i.e., discourse). 1 Extensive experiments show that our model can yield both coherent topics and meaningful discourse behavior. Further study shows that our topic and discourse representations can benefit the classification of microblog messages, especially when they are jointly trained with the classifier. Our data sets and code are available at: http://github.com/zengjichuan/Topic_Disc .

Presents corrections to author information from the paper, “Characterizing and predicting early reviewers for effective product marketing on e-commerce websites,” (Bai, T., et al), IEEE Trans. Knowl. Data Eng., vol. 30, no. 12, pp. 2271–2284, Dec. 2018.

Combining the virtues of probability graphic models and neural networks, Conditional Variational Auto-encoder (CVAE) has shown promising performance in many applications such as response generation.However, existing CVAE-based models often generate responses from a single latent variable which may not be sufficient to model high variability in responses.To solve this problem, we propose a novel model that sequentially introduces a series of latent variables to condition the generation of each word in the response sequence.In addition, the approximate posteriors of these latent variables are augmented with a backward Recurrent Neural Network (RNN), which allows the latent variables to capture long-term dependencies of future tokens in generation.To facilitate training, we supplement our model with an auxiliary objective that predicts the subsequent bag of words.Empirical experiments conducted on the OpenSubtitle and Reddit datasets show that the proposed model leads to significant improvements on both relevance and diversity over state-of-the-art baselines.

Text might express or evoke multiple emotions with varying intensities.As such, it is crucial to predict and rank multiple relevant emotions by their intensities.Moreover, as emotions might be evoked by hidden topics, it is important to unveil and incorporate such topical information to understand how the emotions are evoked.We proposed a novel interpretable neural network approach for relevant emotion ranking.Specifically, motivated by transfer learning, the neural network is initialized to make the hidden layer approximate the behavior of topic models.Moreover, a novel error function is defined to optimize the whole neural network for relevant emotion ranking.Experimental results on three real-world corpora show that the proposed approach performs remarkably better than the state-of-theart emotion detection approaches and multilabel learning methods.Moreover, the extracted emotion-associated topic words indeed represent emotion-evoking events and are in line with our common-sense knowledge.

Deyu Zhou, Linsen Guo, Yulan He. Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers). 2018.

Deyu Zhou, Yang Yang, Yulan He. Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers). 2018.

Weakly supervised part-of-speech (POS) tagging is to learn to predict the POS tag for a given word in context by making use of partial annotated data instead of the fully tagged corpora. Weakly supervised POS tagging would benefit various natural language processing applications in such languages where tagged corpora are mostly unavailable. In this article, we propose a novel framework for weakly supervised POS tagging based on a dictionary of words with their possible POS tags. In the constrained error-correcting output codes (ECOC)-based approach, a unique L -bit vector is assigned to each POS tag. The set of bitvectors is referred to as a coding matrix with value { 1, -1}. Each column of the coding matrix specifies a dichotomy over the tag space to learn a binary classifier. For each binary classifier, its training data is generated in the following way: each pair of words and its possible POS tags are considered as a positive training example only if the whole set of its possible tags falls into the positive dichotomy specified by the column coding and similarly for negative training examples. Given a word in context, its POS tag is predicted by concatenating the predictive outputs of the L binary classifiers and choosing the tag with the closest distance according to some measure. By incorporating the ECOC strategy, the set of all possible tags for each word is treated as an entirety without the need of performing disambiguation. Moreover, instead of manual feature engineering employed in most previous POS tagging approaches, features for training and testing in the proposed framework are automatically generated using neural language modeling. The proposed framework has been evaluated on three corpora for English, Italian, and Malagasy POS tagging, achieving accuracies of 93.21%, 90.9%, and 84.5% individually, which shows a significant improvement compared to the state-of-the-art approaches.

Online reviews have become an important source of information for users before making an informed purchase decision. Early reviews of a product tend to have a high impact on the subsequent product sales. In this paper, we take the initiative to study the behavior characteristics of early reviewers through their posted reviews on two real-world large e-commerce platforms, i.e., Amazon and Yelp. In specific, we divide product lifetime into three consecutive stages, namelyearly,majority, andlaggards. A user who has posted a review in the early stage is considered as an early reviewer. We quantitatively characterize early reviewers based on their rating behaviors, the helpfulness scores received from others and the correlation of their reviews with product popularity. We have found that (1) an early reviewer tends to assign a higher average rating score; and (2) an early reviewer tends to post more helpful reviews. Our analysis of product reviews also indicates that early reviewers’ ratings and their received helpfulness scores are likely to influence product popularity. By viewing the review posting process as a multiplayer competition game, we propose a novel margin-based embedding model for early reviewer prediction. Extensive experiments on two different e-commerce datasets have shown that our proposed approach outperforms a number of competitive baselines.

Online Judge (OJ) systems have been widely used in many areas, including programming, mathematical problems solving, and job interviews. Unlike other online learning systems, such as Massive Open Online Course, most OJ systems are designed for self-directed learning without the intervention of teachers. Also, in most OJ systems, problems are simply listed in volumes and there is no clear organization of them by topics or difficulty levels. As such, problems in the same volume are mixed in terms of topics or difficulty levels. By analyzing large-scale users’ learning traces, we observe that there are two major learning modes (or patterns). Users either practice problems in a sequential manner from the same volume regardless of their topics or they attempt problems about the same topic, which may spread across multiple volumes. Our observation is consistent with the findings in classic educational psychology. Based on our observation, we propose a novel two-mode Markov topic model to automatically detect the topics of online problems by jointly characterizing the two learning modes. For further predicting the difficulty level of online problems, we propose a competition-based expertise model using the learned topic information. Extensive experiments on three large OJ datasets have demonstrated the effectiveness of our approach in three different tasks, including skill topic extraction, expertise competition prediction and problem recommendation.

To extract structured representations of newsworthy events from Twitter, unsupervised models typically assume that tweets involving the same named entities and expressed using similar words are likely to belong to the same event.Hence, they group tweets into clusters based on the cooccurrence patterns of named entities and topical keywords.However, there are two main limitations.First, they require the number of events to be known beforehand, which is not realistic in practical applications.Second, they don't recognise that the same named entity might be referred to by multiple mentions and tweets using different mentions would be wrongly assigned to different events.To overcome these limitations, we propose a nonparametric Bayesian mixture model with word embeddings for event extraction, in which the number of events can be inferred automatically and the issue of lexical variations for the same named entity can be dealt with properly.Our model has been evaluated on three datasets with sizes ranging between 2,499 and over 60 million tweets.Experimental results show that our model outperforms the baseline approach on all datasets by 5-8% in F-measure.

Emotion cause extraction aims to identify the reasons behind a certain emotion expressed in text.It is a much more difficult task compared to emotion classification.Inspired by recent advances in using deep memory networks for question answering (QA), we propose a new approach which considers emotion cause identification as a reading comprehension task in QA.Inspired by convolutional neural networks, we propose a new mechanism to store relevant context in different memory slots to model context information.Our proposed approach can extract both word level sequence features and lexical features.Performance evaluation shows that our method achieves the state-of-the-art performance on a recently released emotion cause dataset, outperforming a number of competitive baselines by at least 3.01% in F-measure.

We explore how to detect people's perspectives that occupy a certain proposition.We propose a Bayesian modelling approach where topics (or propositions) and their associated perspectives (or viewpoints) are modeled as latent variables.Words associated with topics or perspectives follow different generative routes.Based on the extracted perspectives, we can extract the top associated sentences from text to generate a succinct summary which allows a quick glimpse of the main viewpoints in a document.The model is evaluated on debates from the House of Commons of the UK Parliament, revealing perspectives from the debates without the use of labelled data and obtaining better results than previous related solutions under a variety of evaluations.

Stance classification, which aims at detecting the stance expressed in text towards a specific target, is an emerging problem in sentiment analysis. A major difference between stance classification and traditional aspect-level sentiment classification is that the identification of stance is dependent on target which might not be explicitly mentioned in text. This indicates that apart from text content, the target information is important to stance detection. To this end, we propose a neural network-based model, which incorporates target-specific information into stance classification by following a novel attention mechanism. In specific, the attention mechanism is expected to locate the critical parts of text which are related to target. Our evaluations on both the English and Chinese Stance Detection datasets show that the proposed model achieves the state-of-the-art performance.

BACKGROUND: Prediction of DNA-binding residue is important for understanding the protein-DNA recognition mechanism. Many computational methods have been proposed for the prediction, but most of them do not consider the relationships of evolutionary information between residues. RESULTS: In this paper, we first propose a novel residue encoding method, referred to as the Position Specific Score Matrix (PSSM) Relation Transformation (PSSM-RT), to encode residues by utilizing the relationships of evolutionary information between residues. PDNA-62 and PDNA-224 are used to evaluate PSSM-RT and two existing PSSM encoding methods by five-fold cross-validation. Performance evaluations indicate that PSSM-RT is more effective than previous methods. This validates the point that the relationship of evolutionary information between residues is indeed useful in DNA-binding residue prediction. An ensemble learning classifier (EL_PSSM-RT) is also proposed by combining ensemble learning model and PSSM-RT to better handle the imbalance between binding and non-binding residues in datasets. EL_PSSM-RT is evaluated by five-fold cross-validation using PDNA-62 and PDNA-224 as well as two independent datasets TS-72 and TS-61. Performance comparisons with existing predictors on the four datasets demonstrate that EL_PSSM-RT is the best-performing method among all the predicting methods with improvement between 0.02-0.07 for MCC, 4.18-21.47% for ST and 0.013-0.131 for AUC. Furthermore, we analyze the importance of the pair-relationships extracted by PSSM-RT and the results validates the usefulness of PSSM-RT for encoding DNA-binding residues. CONCLUSIONS: We propose a novel prediction method for the prediction of DNA-binding residue with the inclusion of relationship of evolutionary information and ensemble learning. Performance evaluation shows that the relationship of evolutionary information between residues is indeed useful in DNA-binding residue prediction and ensemble learning can be used to address the data imbalance issue between binding and non-binding residues. A web service of EL_PSSM-RT ( http://hlt.hitsz.edu.cn:8080/PSSM-RT_SVM/ ) is provided for free access to the biological research community.

Different types of sentences express sentiment in very different ways. Traditional sentence-level sentiment classification research focuses on one-technique-fits-all solution or only centers on one special type of sentences. In this paper, we propose a divide-and-conquer approach which first classifies sentences into different types, then performs sentiment analysis separately on sentences from each type. Specifically, we find that sentences tend to be more complex if they contain more sentiment targets. Thus, we propose to first apply a neural network based sequence model to classify opinionated sentences into three types according to the number of targets appeared in a sentence. Each group of sentences is then fed into a one-dimensional convolutional neural network separately for sentiment classification. Our approach has been evaluated on four sentiment classification datasets and compared with a wide range of baselines. Experimental results show that: (1) sentence type classification can improve the performance of sentence-level sentiment analysis; (2) the proposed approach achieves state-of-the-art results on several benchmarking datasets.

Social media provides unprecedented opportunities for people to disseminate information and share their opinions and views online. Extracting events from social media platforms such as Twitter could help in understanding what is being discussed. However, event extraction from social text streams po ses huge challenges due to the noisy nature of social media posts and dynamic evolution of language. We propose a generic unsupervised framework for exploring events on Twitter which consists of four major steps, filtering, pre-processing, extraction and categorization, and post-processing. Tweets published in a certain time period are aggregated and noisy tweets which do not contain newsworthy events are filtered by the filtering step. The remaining tweets are pre-processed by temporal resolution, part-of-speech tagging and named entity recognition in order to identify the key elements of events. An unsupervised Bayesian model is proposed to automatically extract the structured representations of events in the form of quadruples < entity, keyword, date, location > and further categorize the extracted events into event types. Finally, the categorized events are assigned with the event type labels without human intervention. The proposed framework has been evaluated on over 60 million tweets which were collected for one month in December 2010. A precision of 78.01% is achieved for event extraction using our proposed Bayesian model, outperforming a competitive baseline by nearly 13.6%. Moreover, events are also clustered into coherence groups with the automatically assigned event type labels with an accuracy of 42.57%.

In clinical documents, medical terms are often expressed in multi-word phrases. Traditional topic modelling approaches relying on the "bag-of-words" assumption are not effective in extracting topic themes from clinical documents. This paper proposes to first extract medical phrases using an off-the-shelf tool for medical concept mention extraction, and then train a topic model which takes a hierarchy of Pitman-Yor processes as prior for modelling the generation of phrases of arbitrary length. Experimental results on patients' discharge summaries show that the proposed approach outperforms the state-of-the-art topical phrase extraction model on both perplexity and topic coherence measure and finds more interpretable topics.

Conventional topic models are ineffective for topic extraction from microblog messages since the lack of structure and context among the posts renders poor message-level word co-occurrence patterns.In this work, we organize microblog posts as conversation trees based on reposting and replying relations, which enrich context information to alleviate data sparseness.Our model generates words according to topic dependencies derived from the conversation structures.In specific, we differentiate messages as leader messages, which initiate key aspects of previously focused topics or shift the focus to different topics, and follower messages that do not introduce any new information but simply echo topics from the messages that they repost or reply.Our model captures the different extents that leader and follower messages may contain the key topical words, thus further enhances the quality of the induced topics.The results of thorough experiments demonstrate the effectiveness of our proposed model.

Event extraction from texts aims to detect structured information such as what has happened, to whom, where and when.Event extraction and visualization are typically considered as two different tasks.In this paper, we propose a novel approach based on probabilistic modelling to jointly extract and visualize events from tweets where both tasks benefit from each other.We model each event as a joint distribution over named entities, a date, a location and event-related keywords.Moreover, both tweets and event instances are associated with coordinates in the visualization space.The manifold assumption that the intrinsic geometry of tweets is a low-rank, non-linear manifold within the high-dimensional space is incorporated into the learning framework using a regularization.Experimental results show that the proposed approach can effectively deal with both event extraction and visualization and performs remarkably better than both the state-of-the-art event extraction method and a pipeline approach for event extraction and visualization.

We study the problem of detecting sentences describing adverse drug reactions (ADRs) and frame the problem as binary classification. We investigate different neural network (NN) architectures for ADR classification. In particular, we propose two new neural network models, Convolutional Recurrent Neural Network (CRNN) by concatenating convolutional neural networks with recurrent neural networks, and Convolutional Neural Network with Attention (CNNA) by adding attention weights into convolutional neural networks. We evaluate various NN architectures on a Twitter dataset containing informal language and an Adverse Drug Effects (ADE) dataset constructed by sampling from MEDLINE case reports. Experimental results show that all the NN architectures outperform the traditional maximum entropy classifiers trained from n-grams with different weighting strategies considerably on both datasets. On the Twitter dataset, all the NN architectures perform similarly. But on the ADE dataset, CNN performs better than other more complex CNN variants. Nevertheless, CNNA allows the visualisation of attention weights of words when making classification decisions and hence is more appropriate for the extraction of word subsequences describing ADRs.

We propose a novel template matching approach for the discrimination of handwritten and machine-printed text. We first pre-process the scanned document images by performing denoising, circles/lines exclusion and word-block level segmentation. We then align and match characters in a flexible sized gallery with the segmented regions, using parallelised normalised cross-correlation. The experimental results over the Pattern Recognition & Image Analysis Research Lab-Natural History Museum (PRImA-NHM) dataset show remarkably high robustness of the algorithm in classifying cluttered, occluded and noisy samples, in addition to those with significant high missing data. The algorithm, which gives 84.0% classification rate with false positive rate 0.16 over the dataset, does not require training samples and generates compelling results as opposed to the training-based approaches, which have used the same benchmark.

Persuasive communication is the process of shaping, reinforcing and changing others' responses.In political debates, speakers express their views towards the debated topics by choosing both the content of their discourse and the argumentation process.In this work we study the use of semantic frames for modelling argumentation in speakers' discourse.We investigate the impact of a speaker's argumentation style and their effect in influencing an audience in supporting their candidature.We model the influence index of each candidate based on their relative standings in the polls released prior to the debate and present a system which ranks speakers in terms of their relative influence using a combination of content and persuasive argumentation features.Our results show that although content alone is predictive of a speaker's influence rank, persuasive argumentation also affects such indices.

We present in this article an automated framework that extracts product adopter information from online reviews and incorporates the extracted information into feature-based matrix factorization for more effective product recommendation. In specific, we propose a bootstrapping approach for the extraction of product adopters from review text and categorize them into a number of different demographic categories. The aggregated demographic information of many product adopters can be used to characterize both products and users in the form of distributions over different demographic categories. We further propose a graph-based method to iteratively update user- and product-related distributions more reliably in a heterogeneous user--product graph and incorporate them as features into the matrix factorization approach for product recommendation. Our experimental results on a large dataset crawled from J ing D ong , the largest B2C e-commerce website in China, show that our proposed framework outperforms a number of competitive baselines for product recommendation.

In recent years, the boundaries between e-commerce and social networking have become increasingly blurred. Many e-commerce Web sites support the mechanism of social login where users can sign on the Web sites using their social network identities such as their Facebook or Twitter accounts. Users can also post their newly purchased products on microblogs with links to the e-commerce product Web pages. In this paper, we propose a novel solution for cross-site cold-start product recommendation, which aims to recommend products from e-commerce Web sites to users at social networking sites in “cold-start” situations, a problem which has rarely been explored before. A major challenge is how to leverage knowledge extracted from social networking sites for cross-site cold-start product recommendation. We propose to use the linked users across social networking sites and e-commerce Web sites (users who have social networking accounts and have made purchases on e-commerce Web sites) as a bridge to map users' social networking features to another feature representation for product recommendation. In specific, we propose learning both users' and products' feature representations (called user embeddings and product embeddings, respectively) from data collected from e-commerce Web sites using recurrent neural networks and then apply a modified gradient boosting trees method to transform users' social networking features into user embeddings. We then develop a feature-based matrix factorization approach which can leverage the learnt user embeddings for cold-start product recommendation. Experimental results on a large dataset constructed from the largest Chinese microblogging service Sina Weibo and the largest Chinese B2C e-commerce website JingDong have shown the effectiveness of our proposed framework.

Twitter, as a popular microblogging service, has become a new information channel for users to receive and exchange the mostup-to-date information on current events. However, since there is no control on how users can publish messages on Twitter, finding newsworthy events from Twitter becomes a difficult task like "finding a needle in a haystack". In this paper we propose a general unsupervised framework to explore events from tweets, which consists of a pipeline process of filtering, extraction and categorization. To filter out noisy tweets, the filtering step exploits a lexicon-based approach to separate tweets that are event-related from those that are not. Then, based on these event-related tweets, the structured representations of events are extracted and categorized automatically using an unsupervised Bayesian model without the use of any labelled data. Moreover, the categorized events are assigned with the event type labels without human intervention. The proposed framework has been evaluated on over 60 millions tweets which were collected for one month in December 2010. A precision of 70.49% is achieved in event extraction, outperforming a competitive baseline by nearly 6%. Events are also clustered into coherence groups with the automatically assigned event type label.

Storyline detection from news articles aims at summarizing events described under a certain news topic and revealing how those events evolve over time.It is a difficult task because it requires first the detection of events from news articles published in different time periods and then the construction of storylines by linking events into coherent news stories.Moreover, each storyline has different hierarchical structures which are dependent across epochs.Existing approaches often ignore the dependency of hierarchical structures in storyline generation.In this paper, we propose an unsupervised Bayesian model, called dynamic storyline detection model, to extract structured representations and evolution patterns of storylines.The proposed model is evaluated on a large scale news corpus.Experimental results show that our proposed model outperforms several baseline approaches.

In this paper, we explore the idea of social role theory (SRT) and propose a novel regularized topic model which incorporates SRT into the generative process of social media content. We assume that a user can play multiple social roles, and each social role serves to fulfil different duties and is associated with a role-driven distribution over latent topics. In particular, we focus on social roles corresponding to the most common social activities on social networks. Our model is instantiated on microblogs, i.e., Twitter and community question-answering (cQA), i.e., Yahoo!Answers, where social roles on Twitter include “originators” and “propagators”, and roles on cQA are “askers” and “answerers”. Both explicit and implicit interactions between users are taken into account and modeled as regularization factors. To evaluate the performance of our proposed method, we have conducted extensive experiments on two Twitter datasets and two cQA datasets. Furthermore, we also consider multi-role modeling for scientific papers where an author's research expertise area is considered as a social role. A novel application of detecting users' research interests through topical keyword labeling based on the results of our multi-role model has been presented. The evaluation results have shown the feasibility and effectiveness of our model.

Existing approaches of social influence analysis usually focus on how to develop effective algorithms to quantize users' influence scores. They rarely consider a person's expertise levels which are arguably important to influence measures. In this paper, we propose a computational approach to measuring the correlation between expertise and social media influence, and we take a new perspective to understand social media influence by incorporating expertise into influence analysis. We carefully constructed a large dataset of 13,684 Chinese celebrities from Sina Weibo (literally “Sina microblogging”). We found that there is a strong correlation between expertise levels and social media influence scores. In addition, different expertise levels showed influence variation patterns: high-expertise celebrities have stronger influence on the “audience” in their expertise domains.

Product recommender systems are often deployed by e-commerce websites to improve user experience and increase sales. However, recommendation is limited by the product information hosted in those e-commerce sites and is only triggered when users are performing e-commerce activities. In this paper, we develop a novel product recommender system called METIS, a MErchanT Intelligence recommender System, which detects users' purchase intents from their microblogs in near real-time and makes product recommendation based on matching the users' demographic information extracted from their public profiles with product demographics learned from microblogs and online reviews. METIS distinguishes itself from traditional product recommender systems in the following aspects: 1) METIS was developed based on a microblogging service platform. As such, it is not limited by the information available in any specific e-commerce website. In addition, METIS is able to track users' purchase intents in near real-time and make recommendations accordingly. 2) In METIS, product recommendation is framed as a learning to rank problem. Users' characteristics extracted from their public profiles in microblogs and products' demographics learned from both online product reviews and microblogs are fed into learning to rank algorithms for product recommendation. We have evaluated our system in a large dataset crawled from Sina Weibo. The experimental results have verified the feasibility and effectiveness of our system. We have also made a demo version of our system publicly available and have implemented a live system which allows registered users to receive recommendations in real time.

MOTIVATION: In molecular biology, molecular events describe observable alterations of biomolecules, such as binding of proteins or RNA production. These events might be responsible for drug reactions or development of certain diseases. As such, biomedical event extraction, the process of automatically detecting description of molecular interactions in research articles, attracted substantial research interest recently. Event trigger identification, detecting the words describing the event types, is a crucial and prerequisite step in the pipeline process of biomedical event extraction. Taking the event types as classes, event trigger identification can be viewed as a classification task. For each word in a sentence, a trained classifier predicts whether the word corresponds to an event type and which event type based on the context features. Therefore, a well-designed feature set with a good level of discrimination and generalization is crucial for the performance of event trigger identification. RESULTS: In this article, we propose a novel framework for event trigger identification. In particular, we learn biomedical domain knowledge from a large text corpus built from Medline and embed it into word features using neural language modeling. The embedded features are then combined with the syntactic and semantic context features using the multiple kernel learning method. The combined feature set is used for training the event trigger classifier. Experimental results on the golden standard corpus show that >2.5% improvement on F-score is achieved by the proposed framework when compared with the state-of-the-art approach, demonstrating the effectiveness of the proposed framework.

Learning user interests from online social networks helps to better understand user behaviors and provides useful guidance to design user-centric applications. Apart from analyzing users' online content, it is also important to consider users' social connections in the social Web. Graph regularization methods have been widely used in various text mining tasks, which can leverage the graph structure information extracted from data. Previously, graph regularization methods operate under the cluster assumption that nearby nodes are more similar and nodes on the same structure (typically referred to as a cluster or a manifold) are likely to be similar. We argue that learning user interests from complex, sparse, and dynamic social networks should be based on the link structure assumption under which node similarities are evaluated based on the local link structures instead of explicit links between two nodes. We propose a regularization framework based on the relation bipartite graph, which can be constructed from any type of relations. Using Twitter as our case study, we evaluate our proposed framework from social networks built from retweet relations. Both quantitative and qualitative experiments show that our proposed method outperforms a few competitive baselines in learning user interests over a set of predefined topics. It also gives superior results compared to the baselines on retweet prediction and topical authority identification.

A large number of studies have been devoted to modeling the contents and interactions between users on Twitter. In this paper, we propose a method inspired from Social Role Theory (SRT), which assumes that a user behaves differently with different roles in the generation process of Twitter content. We consider the two most distinctive social roles on Twitter: originator and propagator, who respectively posts original messages and retweets or forwards the messages from others. In addition, we also consider role-specific social interactions, especially implicit interactions between users who share some common interests. All the above elements are integrated into a novel regularized topic model. We evaluate the proposed method on real Twitter data. The results show that our method is more effective than the existing ones which do not distinguish social roles.

Social learning analytics are concerned with the process of knowledge construction as learners build knowledge together in their social and cultural environments. One of the most important tools employed during this process is language. In this paper we take exploratory dialogue, a joint form of co-reasoning, to be an external indicator that learning is taking place. Using techniques developed within the field of computational linguistics, we build on previous work using cue phrases to identify exploratory dialogue within online discussion. Automatic detection of this type of dialogue is framed as a binary classification task that labels each contribution to an online discussion as exploratory or non-exploratory. We describe the development of a self-training framework that employs discourse features and topical features for classification by integrating both cue-phrase matching and k-nearest neighbour classification. Experiments with a corpus constructed from the archive of a two-day online conference show that our proposed framework outperforms other approaches. A classifier developed using the self-training framework is able to make useful distinctions between the learning dialogue taking place at different times within an online conference as well as between the contributions of individual participants.

Timeline generation is an important research task which can help users to have a quick understanding of the overall evolution of any given topic. It thus attracts much attention from research communities in recent years. Nevertheless, existing work on timeline generation often ignores an important factor, the attention attracted to topics of interest (hereafter termed "social attention"). Without taking into consideration social attention, the generated timelines may not reflect users' collective interests. In this paper, we study how to incorporate social attention in the generation of timeline summaries. In particular, for a given topic, we capture social attention by learning users' collective interests in the form of word distributions from Twitter, which are subsequently incorporated into a unified framework for timeline summary generation. We construct four evaluation sets over six diverse topics. We demonstrate that our proposed approach is able to generate both informative and interesting timelines. Our work sheds light on the feasibility of incorporating social attention into traditional text mining tasks.

Social media data are produced continuously by a large and uncontrolled number of users. The dynamic nature of such data requires the sentiment and topic analysis model to be also dynamically updated, capturing the most recent language use of sentiments and topics in text. We propose a dynamic Joint Sentiment-Topic model (dJST) which allows the detection and tracking of views of current and recurrent interests and shifts in topic and sentiment. Both topic and sentiment dynamics are captured by assuming that the current sentiment-topic-specific word distributions are generated according to the word distributions at previous epochs. We study three different ways of accounting for such dependency information: (1) sliding window where the current sentiment-topic word distributions are dependent on the previous sentiment-topic-specific word distributions in the last S epochs; (2) skip model where history sentiment topic word distributions are considered by skipping some epochs in between; and (3) multiscale model where previous long- and short- timescale distributions are taken into consideration. We derive efficient online inference procedures to sequentially update the model with newly arrived data and show the effectiveness of our proposed model on the Mozilla add-on reviews crawled between 2007 and 2011.

This article presents two novel approaches for incorporating sentiment prior knowledge into the topic model for weakly supervised sentiment analysis where sentiment labels are considered as topics. One is by modifying the Dirichlet prior for topic-word distribution (LDA-DP), the other is by augmenting the model objective function through adding terms that express preferences on expectations of sentiment labels of the lexicon words using generalized expectation criteria (LDA-GE). We conducted extensive experiments on English movie review data and multi-domain sentiment dataset as well as Chinese product reviews about mobile phones, digital cameras, MP3 players, and monitors. The results show that while both LDA-DP and LDA-GE perform comparably to existing weakly supervised sentiment classification algorithms, they are much simpler and computationally efficient, rendering them more suitable for online and real-time sentiment classification on the Web. We observed that LDA-GE is more effective than LDA-DP, suggesting that it should be preferred when considering employing the topic model for sentiment analysis. Moreover, both models are able to extract highly domain-salient polarity words from text.

Sentiment analysis or opinion mining aims to use automated tools to detect subjective information such as opinions, attitudes, and feelings expressed in text. This paper proposes a novel probabilistic modeling framework called joint sentiment-topic (JST) model based on latent Dirichlet allocation (LDA), which detects sentiment and topic simultaneously from text. A reparameterized version of the JST model called Reverse-JST, obtained by reversing the sequence of sentiment and topic generation in the modeling process, is also studied. Although JST is equivalent to Reverse-JST without a hierarchical prior, extensive experiments show that when sentiment priors are added, JST performs consistently better than Reverse-JST. Besides, unlike supervised approaches to sentiment classification which often fail to produce satisfactory performance when shifting to other domains, the weakly supervised nature of JST makes it highly portable to other domains. This is verified by the experimental results on data sets from five different domains where the JST model even outperforms existing semi-supervised approaches in some of the data sets despite using no labeled documents. Moreover, the topics and topic sentiment detected by JST are indeed coherent and informative. We hypothesize that the JST model can readily meet the demand of large-scale sentiment analysis from the web in an open-ended fashion.

Natural language understanding (NLU) aims to map sentences to their semantic mean representations. Statistical approaches to NLU normally require fully-annotated training data where each sentence is paired with its word-level semantic annotations. In this paper, we propose a novel learning framework which trains the Hidden Markov Support Vector Machines (HM-SVMs) without the use of expensive fully-annotated data. In particular, our learning approach takes as input a training set of sentences labeled with abstract semantic annotations encoding underlying embedded structural relations and automatically induces derivation rules that map sentences to their semantic meaning representations. The proposed approach has been tested on the DARPA Communicator Data and achieved 93.18% in F-measure, which outperforms the previously proposed approaches of training the hidden vector state model or conditional random fields from unaligned data, with a relative error reduction rate of 43.3% and 10.6% being achieved.

We propose a novel framework where an initial classifier is learned by incorporating prior information extracted from an existing sentiment lexicon. Preferences on expectations of sentiment labels of those lexicon words are expressed using generalized expectation criteria. Documents classified with high confidence are then used as pseudo-labeled examples for automatical domain-specific feature acquisition. The word-class distributions of such self-learned features are estimated from the pseudo-labeled examples and are used to train another classifier by constraining the model's predictions on unlabeled instances. Experiments on both the movie review data and the multi-domain sentiment dataset show that our approach attains comparable or better performance than exiting weakly-supervised sentiment classification methods despite using no labeled documents.

Sentiment analysis or opinion mining aims to use automated tools to detect subjective information such as opinions, attitudes, and feelings expressed in text. This paper proposes a novel probabilistic modeling framework based on Latent Dirichlet Allocation (LDA), called joint sentiment/topic model (JST), which detects sentiment and topic simultaneously from text. Unlike other machine learning approaches to sentiment classification which often require labeled corpora for classifier training, the proposed JST model is fully unsupervised. The model has been evaluated on the movie review dataset to classify the review sentiment polarity and minimum prior information have also been explored to further improve the sentiment classification accuracy. Preliminary experiments have shown promising results achieved by JST.

In this paper, we discuss how discriminative training can be applied to the hidden vector state (HVS) model in different task domains. The HVS model is a discrete hidden Markov model (HMM) in which each HMM state represents the state of a push-down automaton with a finite stack size. In previous applications, maximum-likelihood estimation (MLE) is used to derive the parameters of the HVS model. However, MLE makes a number of assumptions and unfortunately some of these assumptions do not hold. Discriminative training, without making such assumptions, can improve the performance of the HVS model by discriminating the correct hypothesis from the competing hypotheses. Experiments have been conducted in two domains: the travel domain for the semantic parsing task using the DARPA Communicator data and the Air Travel Information Services (ATIS) data and the bioinformatics domain for the information extraction task using the GENIA corpus. The results demonstrate modest improvements of the performance of the HVS model using discriminative training. In the travel domain, discriminative training of the HVS model gives a relative error reduction rate of 31 percent in F-measure when compared with MLE on the DARPA Communicator data and 9 percent on the ATIS data. In the bioinformatics domain, a relative error reduction rate of 4 percent in F-measure is achieved on the GENIA corpus.

A major challenge in text mining for biology and biomedicine is automatically extracting protein-protein interactions from the vast amount of biological literature since most knowledge about them still hides in biological publications. Existing approaches can be broadly categorized as rule-based or statistical-based. Rule-based approaches require heavy manual efforts. On the other hand, statistical-based approaches require large-scale, richly annotated corpora in order to reliably estimate model parameters. This is normally difficult to obtain in practical applications. The hidden vector state (HVS) model, an extension of the basic discrete Markov model, has been successfully applied to extract protein-protein interactions. In this paper, we propose a novel approach to train the HVS model on both annotated and un-annotated corpus. Sentences selection algorithm is designed to utilize the semantic parsing results of the un-annotated corpus generated by the HVS model. Experimental results show that the performance of the initial HVS model trained on a small amount of the annotated data can be improved by employing this approach

The paper presents a hidden vector state (HVS) model for hierarchical semantic parsing. The model associates each state of push-down automata with the state of an HMM. State transitions are factored into separate stack pop and push operations and then constrained to give a tractable search space. The result is a model which is complex enough to capture hierarchical structure but which can be trained automatically from unannotated data. Experiments have been conducted on ATIS-3 1993 and 1994 test sets. The results show that the HVS model outperforms a general finite state tagger (FST) by 19% to 32% in error reduction.