Most commonsense reasoning models overlook the influence of personality traits, limiting their effectiveness in personalized systems such as dialogue generation. To address this limitation, we introduce the Personality-aware Commonsense Knowledge Graph (PCoKG), a structured dataset comprising 521,316 quadruples. We begin by employing three evaluators to score and filter events from the ATOMIC dataset, selecting those that are likely to elicit diverse reasoning patterns across different personality types. For knowledge graph construction, we leverage the role-playing capabilities of large language models (LLMs) to perform reasoning tasks. To enhance the quality of the generated knowledge, we incorporate a debate mechanism consisting of a proponent, an opponent, and a judge, which iteratively refines the outputs through feedback loops. We evaluate the dataset from multiple perspectives and conduct fine-tuning and ablation experiments using multiple LLM backbones to assess PCoKG's robustness and the effectiveness of its construction pipeline. Our LoRA-based fine-tuning results indicate a positive correlation between model performance and the parameter scale of the base models. Finally, we apply PCoKG to persona-based dialogue generation, where it demonstrates improved consistency between generated responses and reference outputs. This work bridges the gap between commonsense reasoning and individual cognitive differences, enabling the development of more personalized and context-aware AI systems.

In the literature, existing human-centric emotional motion generation methods primarily focus on boosting performance within a single scale-fixed dataset, largely neglecting the flexible and scale-increasing motion scenarios (e.g., sports, dance), whereas effectively learning these newly emerging scenarios can significantly enhance the model’s real-world generalization ability. Inspired by this, this paper proposes a new LLM-Centric Lifelong Empathic Motion Generation (L2-EMG) task, which aims to equip LLMs with the capability to continually acquire emotional motion generation knowledge across different unseen scenarios, potentially contributing to building a closed-loop and self-evolving embodied agent equipped with both empathy and intelligence. Further, this paper poses two key challenges in the L2-EMG task, i.e., the emotion decoupling challenge and the scenario adapting challenge. To this end, this paper proposes an Emotion-Transferable and Scenario-Adapted Mixture of Experts (ES-MoE) approach which designs a causal-guided emotion decoupling block and a scenario-adapted expert constructing block to address the two challenges, respectively. Especially, this paper constructs multiple L2-EMG datasets to validate the effectiveness of the ES-MoE approach. Extensive evaluations show that ES-MoE outperforms advanced baselines.

Large language models (LLMs) often generate hallucinations in knowledge-intensive QA due to parametric knowledge limitations. While existing methods like KG-CoT improve reliability by integrating knowledge graph (KG) paths, they suffer from rigid hop-count selection (solely question-driven) and underutilization of reasoning paths (lack of guidance). To address this, we propose RFKG-CoT: First, it replaces the rigid hop-count selector with a relation-driven adaptive hop-count selector that dynamically adjusts reasoning steps by activating KG relations (e.g., 1-hop for direct ''brother" relations, 2-hop for indirect ''father-son" chains), formalized via a relation mask. Second, it introduces a few-shot in-context learning path guidance mechanism with CoT (think) that constructs examples in a ''question-paths-answer" format to enhance LLMs' ability to understand reasoning paths. Experiments on four KGQA benchmarks show RFKG-CoT improves accuracy by up to 14.7 pp (Llama2-7B on WebQSP) over KG-CoT. Ablations confirm the hop-count selector and the path prompt are complementary, jointly transforming KG evidence into more faithful answers.

The semantic gap between colloquial user queries and professional legal documents presents a fundamental challenge in Legal Case Retrieval (LCR).Existing dense retrieval methods typically treat LCR as a black-box semantic matching process, neglecting the explicit juridical logic that underpins legal relevance.To address this, we propose GLIER (Generative Legal Inference and Evidence Ranking), a framework that reformulates retrieval as an inference process over latent legal variables.GLIER decomposes the task into two interpretability-driven stages: (1) A Joint Generative Inference module that translates raw queries into latent legal indicators (Charges and Legal Elements), employing a unified sequenceto-sequence strategy where charges and elements are generated jointly to enforce logical consistency; and (2) A Multi-View Evidence Fusion mechanism that aggregates generative confidence with structural and lexical signals for precise ranking.Extensive experiments on LeCaRD and LeCaRDv2 demonstrate that GLIER outperforms strong baselines like SAILER and KELLER.Notably, our framework exhibits exceptional data efficiency, maintaining robust performance even when trained with only 10% of the data. * Equal contribution.† Corresponding author.Code is available at https://github.com/SUGAR-NLP/ GLIER.(Query)On November 17, 2016, at 9 PM... due to a dispute over someone else's matter, he had an argument with Shao Huapeng over the phone.Shao Huapeng arranged to fight him at Zhaoh Bridge, but he did not go to the agreed location.Shao Huapeng

Retrieval-Augmented Generation (RAG) grounds language models in external evidence, but multi-hop question answering remains difficult because iterative pipelines must control what to retrieve next and when the available evidence is adequate.In practice, systems may answer from incomplete evidence chains, or they may accumulate redundant or distractor-heavy text that interferes with later retrieval and reasoning.We propose S2G-RAG (Structured Sufficiency and Gap-judging RAG), an iterative framework with an explicit controller, S2G-Judge.At each turn, S2G-Judge predicts whether the current evidence memory supports answering and, if not, outputs structured gap items that describe the missing information.We map these gap items into the next retrieval query, producing stable multi-turn retrieval trajectories.To reduce noise accumulation, we maintain a sentence-level Evidence Context by extracting a compact set of relevant sentences from retrieved documents.Experiments on TriviaQA, HotpotQA, and 2WikiMultiHopQA show that S2G-RAG improves multi-hop QA performance and robustness under multi-turn retrieval.Furthermore

In the literature, prior research on Security-oriented Video Understanding (SVU) has predominantly focused on detecting and locating the threats (e.g., shootings, robberies) in videos, while largely lacking the effective capability to generate and evaluate the threat causes. Motivated by these gaps, this paper introduces a new chat paradigm SVU task, i.e., In-depth Security-oriented Video Understanding (DeepSVU), which aims to not only identify and locate the threats but also attribute and evaluate the causes of threatening segments in detail. Furthermore, this paper reveals two key challenges in the proposed task: 1) how to effectively model the coarse-to-fine physical-world information (e.g., human behavior, object interactions and background context) to boost the DeepSVU task, and 2) how to adaptively trade off these factors. Addressing these challenges is crucial for improving VAD, especially for identifying, locating, and attributing anomalies. To tackle these challenges, this paper proposes a new Unified Physical-world Regularized MoE (UPRM) approach. Specifically, UPRM incorporates two key components: the Unified Physical-world Enhanced MoE (UPE) Block and the Physical-world Trade-off Regularizer (PTR), to address the above two challenges, respectively. Extensive experiments conduct on our DeepSVU instructions datasets (i.e., UCF-C instructions and CUVA instructions) demonstrate that UPRM outperforms several advanced Video-LLMs as well as non-LLM approaches. such information.These justify the importance of the coarse-to-fine physical-world information in the DeepSVU task and demonstrate the effectiveness of our UPRM in capturing such information.

Current stance detection models predominantly rely on text analysis, often overlooking the wealth of information embedded in user profiles. This oversight is largely attributed to the scarcity of stance detection datasets that encompass detailed user profiles. Bridging this gap, we have created the first dataset for Chinese stance detection that includes user profile information (PC-STANCE). The dataset contains 31,033 Chinese microblogs annotated with stances toward 8 targets, as well as user gender, age, and location information. We have conducted a detailed analysis of the dataset and performed extensive empirical experiments using classic neural network models. Our experiments achieve state-of-the-art results, surpassing several large language models like Llama and ChatGPT. This confirms the effectiveness of integrating user profile information in stance detection and underscores the challenging nature of the dataset. To facilitate further research in stance detection, we have made the dataset publicly available. 1

Modern information retrieval (IR) must reconcile short, ambiguous queries with increasingly diverse and dynamic corpora. Query expansion (QE) remains a core technique for mitigating vocabulary mismatch, but its design space has been reshaped by pre-trained and large language models (PLMs/LLMs). This survey reviews QE methods in the PLM/LLM era and provides a unified view of the emerging landscape. We first summarize how different model families enable new expansion behaviors, including stronger contextualization, more controllable generation, and instruction-following. We then organize recent techniques along four complementary design dimensions: where expansion is injected in the pipeline, how it is grounded and interacts with corpus evidence, how it is learned or aligned, and how structured knowledge, such as knowledge graphs, is incorporated. Beyond taxonomy, we synthesize application patterns and deployment considerations across representative retrieval settings, highlighting practical tradeoffs among effectiveness, controllability, grounding quality, and operating cost. Finally, we outline open challenges and future directions toward more reliable, safe, efficient, and continually adaptive QE under real-world constraints (resources are available at https://github.com/lmh0921/QueryExpansion-PLM-LLM-Survey-paperList ).

With the emergence of social media and e-commerce platforms, accurate user profiling has become increasingly vital for recommendation systems and personalized services. Recent studies have focused on generating detailed user profiles by extracting various aspects of user attributes from textual reviews. Nevertheless, these investigations have not fully exploited the potential of the abundant multimodal data at hand. In this study, we propose a novel task called multimodal user profiling. This task emphasizes the utilization of both review texts and their accompanying images to create comprehensive user profiles. By integrating textual and visual data, we leverage their complementary strengths, enabling the generation of more holistic user representations. Additionally, we explore a unified joint training framework with various multimodal training strategies that incorporate users’ historical review texts and images for user profile generation. Our experimental results underscore the significance of multimodal data in enhancing user profile generation and demonstrate the effectiveness of the proposed unified joint training approach.

Multimodal sentiment analysis for fashion-related social media is essential for understanding how consumers appraise fashion products across platforms like Instagram and Twitter, where both textual and visual elements contribute to sentiment expression. However, a notable challenge in this task is the modality gap, where the different information density between text and images hinders effective sentiment analysis. In this paper, we propose a novel multimodal framework that addresses this challenge by introducing pseudo data generated by a two-stage framework. We further utilize a multimodal fusion approach that efficiently integrates the information from various modalities for sentiment classification of fashion posts. Experiments conducted on a comprehensive dataset demonstrate that our framework significantly outperforms existing unimodal and multimodal baselines, highlighting its effectiveness in bridging the modality gap for more accurate sentiment classification in fashion-related social media posts.

Stance detection is a task aimed at identifying and analyzing the author’s stance from text. Previous studies have primarily focused on the text, which may not fully capture the implicit stance conveyed by the author. To address this limitation, we propose a novel approach that transforms original texts into artificially generated images and uses the visual representation to enhance stance detection. Our approach first employs a text-to-image model to generate candidate images for each text. These images are carefully crafted to adhere to three specific criteria: textual relevance, target consistency, and stance consistency. Next, we introduce a comprehensive evaluation framework to select the optimal image for each text from its generated candidates. Subsequently, we introduce a multimodal stance detection model that leverages both the original textual content and the generated image to identify the author’s stance. Experiments demonstrate the effectiveness of our approach and highlight the importance of artificially generated images for stance detection.

Large language models have achieved remarkable performance on various question answering (QA) tasks, but the explainability of their output remains challenging. To address this issue, recent work has proposed symbolic explanations called entailment trees as a way to explain and inspect a QA system’s answer. In order to improve the generation of such symbolic explanations and enhance the interpretability of the model, we introduce a new task called Question Answering Explanation Generation. This task retains the tree structure and maps symbols to natural language, aiming to generate natural language explanations from question-answer pairs. The advantage is that by generating natural language explanations, the reasoning process of the model becomes more explicit. Additionally, it abandons the requirement to precisely match relevant knowledge for all question-answer pairs, with only the question and answer considered as inputs. This makes the framework more general and applicable to real-world scenarios. Furthermore, we collected some question-answer data and used a bidirectional generation framework to construct a pseudo-dataset, which allows the model to gain a more comprehensive understanding of the association between QA and explanation. Moreover, we integrated context generation into the model’s generation process instead of providing the context directly as input to consider broader aspects and domain-specific knowledge relevant to the question and answer pair. Applying our method to a small-parameter model outperformed direct fine-tuning of a large-parameter model. The results of the experiment demonstrate the importance of the proposed explanation generation task and also indicate the effectiveness of our approach.

Large language models have achieved remarkable performance on various question answering (QA) tasks, but the explainability of their output remains challenging. To address this issue, recent work has proposed symbolic explanations called entailment trees as a way to explain and inspect a QA system’s answer. In order to improve the generation of such symbolic explanations and enhance the interpretability of the model, we introduce a new task called Question Answering Explanation Generation. This task retains the tree structure and maps symbols to natural language, aiming to generate natural language explanations from question-answer pairs. The advantage is that by generating natural language explanations, the reasoning process of the model becomes more explicit. Additionally, it abandons the requirement to precisely match relevant knowledge for all question-answer pairs, with only the question and answer considered as inputs. This makes the framework more general and applicable to real-world scenarios. Furthermore, we collected some question-answer data and used a bidirectional generation framework to construct a pseudo-dataset, which allows the model to gain a more comprehensive understanding of the association between QA and explanation. Moreover, we integrated context generation into the model’s generation process instead of providing the context directly as input to consider broader aspects and domain-specific knowledge relevant to the question and answer pair. Applying our method to a small-parameter model outperformed direct fine-tuning of a large-parameter model. The results of the experiment demonstrate the importance of the proposed explanation generation task and also indicate the effectiveness of our approach.

Previous multimodal emotion-cause analysis in conversations (MEC-AC) has predominantly focused on English, overlooking the applicability of existing methods in multilingual contexts. To bridge this gap, we construct a Chinese contextual dataset (MEC4) to investigate how language and culture diversity influences existing MECAC approaches. Moreover, prior studies often rely on average pooling or frame sampling to extract visual and acoustic features from video and audio of long dialogues, which inevitably results in the loss of temporal dynamics and emotionally salient cues. To overcome these limitations, we propose a memory-inspired multilingual multimodal framework (M3F) based on large language model (LLM), which can effectively capture the temporal and global informative features of non-linguistic modalities through memory bank module. This module simulates the way memory is stored in human cognitive processes and incrementally aggregates past visual and acoustic features in an autoregressive manner, enabling effective reference during future sequence modeling. Through rigorous experiments and insightful analyses, we find that cultural differences cause variations in how emotional expressions in English and Chinese languages rely on modalities.

Prior studies on Visual Sentiment Understanding (VSU) primarily rely on the explicit scene information (e.g., facial expression) to judge visual sentiments, which largely ignore implicit scene information (e.g., human action, objection relation and visual background), while such information is critical for precisely discovering visual sentiments. Motivated by this, this paper proposes a new Omni-scene driven visual Sentiment Identifying, Locating and Attributing in videos (Omni-SILA) task, aiming to interactively and precisely identify, locate and attribute visual sentiments through both explicit and implicit scene information. Furthermore, this paper believes that this Omni-SILA task faces two key challenges: modeling scene and highlighting implicit scene beyond explicit. To this end, this paper proposes an Implicit-enhanced Causal MoE (ICM) approach for addressing the Omni-SILA task. Specifically, a Scene-Balanced MoE (SBM) and an Implicit-Enhanced Causal (IEC) blocks are tailored to model scene information and highlight the implicit scene information beyond explicit, respectively. Extensive experimental results on our constructed explicit and implicit Omni-SILA datasets demonstrate the great advantage of the proposed ICM approach over advanced Video-LLMs.

Prior studies on Video Anomaly Detection (VAD) mainly focus on detecting whether each video frame is abnormal or not in the video, which largely ignore the structured video semantic information (i.e., what, when, and where does the abnormal event happen). With this in mind, this paper propose a new chat-paradigm Multi-scene Video Abnormal Event Extraction and Localization (M-VAE) task, aiming to extract the abnormal event quadruples (i.e., subject, event type, object, scene) and localize such event. Further, this paper believes that this new task faces two key challenges, i.e., global-local spatial modeling and global-local spatial balancing. To this end, this paper proposes a Global-local Spatial-sensitive Large Language Model (LLM) named Sherlock, i.e., acting like Sherlock Holmes to track down the criminal events, for this M-VAE task. Specifically, this model designs a Global-local Spatial-enhanced MoE (GSM) module and a Spatial Imbalance Regulator (SIR) to address the two challenges respectively. Extensive experiments on our M-VAE instruction dataset show the significant advantages of Sherlock over several advanced Video-LLMs. This justifies the importance of global-local spatial information for the M-VAE task and the effectiveness of Sherlock in capturing such information.

This paper concentrates on Document-level Event Factuality Identification (DEFI) that predicts event factuality values from the viewpoint of the document. At present, the shortcomings of previous studies are multi-fold, including data limitation and scarcity, coarsegrained interpretability without span-level factuality clues, no unified model for different datasets. This paper is devoted to address the above problems by building unified Machine Reading Comprehension (MRC) frameworks comprised of both span-extraction and multiple-choice styles, which exploit Heterogeneous Spectral Attention Networks (HSAN) with spectral networks and hypergraph attention networks as the fine-grained encoders, especially for span-level encoding. Moreover, we integrate Transfer Learning (TL) as cross-domain data augmentation to learn more span-level information from classical MRC datasets by source and target adapters. Experimental performance on ExDLEF corpus, which contains both English and Chinese documents, shows that our span-extraction MRC model is superior to several state-of-the-art baselines, and proves the effectiveness of transfer learning under MRC paradigms.

Cross-domain Aspect Sentiment Triplet Extraction (ASTE) aims to extract the triplets from the review of a target domain, utilizing knowledge from a source domain. As a newly proposed task, limited work has been devoted to it. Except for solving it in a zero-shot manner with in-domain models, recent work explores a bidirectional generative framework to generate pseudo-labeled target data. However, such a method suffers from low efficiency with two-stage training and unstable pseudo-label quality. In this paper, we propose a Hybrid Prompts Mixture (HiPM) method for cross-domain ASTE to fully utilize domain-independent knowledge. Within this method, given that syntax information is an essential linguistic feature for triplet extraction, we design a syntax-related hard prompt to transfer the structures. Additionally, aspects from different domains exhibit similarities in their respective categories. We take this shared information as the prototypes and enrich them through a warm-up step. The resulting prototypes then act as the source of soft prompts. We further mix the hard and soft prompts with the original sequence into a generative model to extract triplets. Experimental results show that our method outperforms baselines on twelve transfer pairs, and obtains a 1.48% average F1 score improvement over the state-of-the-art cross-domain ASTE model.

Emotion recognition in conversation (ERC) presents a significant challenge in natural language processing. In this study, we propose the Prompt-based Fusion Model (DialoguePFM), which innovatively introduces an emotion representation that conveys the emotion label via mask token in a pre-defined prompt. We then refine both emotion and utterance representations by capturing comprehensive dialogue information using a novel speaker-aware attention mechanism, which distinguishes between self and other speakers. Subsequently, these refined representations are merged before being inputted into the classifier. Empirical evaluations conducted on three English ERC datasets and one Chinese ERC dataset reveal that our proposed model either outperforms or matches the performance of state-of-the-art baselines, underscoring its effectiveness across different languages.

Emotion-Cause Pair Extraction (ECPE) in conversations aims to identify the emotional utterances (even their categories) along with their corresponding causal utterances, which is crucial in understanding the cause-effect relationship in dialogues. While prior studies of ECPE have predominantly focused on purely textual dialogues and neglected the exploration on the natural scenario of the dialogues with multimodal features, i.e., Multimodal Emotion-Cause Pair Extraction (MECPE) in conversations. To attempt this scenario, we propose a Generative approach for Multimodal Emotion-Cause pair extraction (GMEC) with a single stage, thus effectively reducing errors associated with the propagation and accumulation for MECPE. This approach can not only uniformly handle the information of diverse modalities, but also address all emotion and cause analysis tasks uniformly. Additionally, instead of utilizing the fixed commonsense knowledge base as previously, we resort to the Large Language Models (LLMs), which possess a powerful ability to emerge new knowledge, thereby acting as implicit knowledge engines for MECPE. We refer to this approach as enhanced GMEC. Extensive experimental results and detailed analysis demonstrate a notable improvement in the generative approach. Moreover, the integration of external knowledge from LLMs optimizes the efficiency of data utilization, particularly in few-shot scenarios. The integration of the generative model with LLMs has resulted in a cumulative enhancement of 4.94%, 10.90% on MECPE and MECPE-C (with emotion Category).

Cognitive reasoning holds a significant place within Natural Language Processing (NLP). Yet, the exploration of zero-shot scenarios, which align more closely with real-life situations than supervised scenarios, has been relatively limited. While a few studies have employed Large Language Models (LLMs) to tackle zero-shot cognitive reasoning tasks, they still grapple with two key challenges: 1) Traditional approaches rely on the chain-of-thought (CoT) mechanism, wherein LLMs are provided with a "Let's think step by step'' prompt. However, this schema may not accurately understand the meaning of a given question and ignores the possible learned knowledge (e.g., background or commonsense) of the LLMs about the questions, leading to incorrect answers. 2) Previous CoT methods normally exploit a single Large Language Model (LLM) and design many strategies to augment this LLM. We argue that the power of a single LLM is typically finite since it may not have learned some relevant knowledge about the question. To address these issues, we propose a Multi-LLM Knowledge Fusion (MLKF) approach, which resorts to heterogeneous knowledge emerging from multiple LLMs, for zero-shot cognitive reasoning tasks. Through extensive experiments and detailed analysis, we demonstrate that our MLKF can outperform the existing zero-shot or unsupervised state-of-the-art methods on four kinds of zero-shot tasks: aspect sentiment analysis, named entity recognition, question answering, and mathematical reasoning. Our code is available at https://github.com/trueBatty/MLKF

Thanks to the development of pre-trained sequence-to-sequence (seq2seq) models (e.g., BART), recent studies on AMR parsing often regard this task as a seq2seq translation problem by linearizing AMR graphs into AMR token sequences in pre-processing and recovering AMR graphs from sequences in post-processing. Seq2seq AMR parsing is a relatively simple paradigm but it unavoidably loses structural information among AMR tokens. To compensate for the loss of structural information, in this paper we explicitly leverage AMR structure in the decoding phase. Given an AMR graph, we first project the structure in the graph into an AMR token graph, i.e., structure among AMR tokens in the linearized sequence. The structures for an AMR token could be divided into two parts: structure in prediction history and structure in future. Then we propose to model structure in prediction history via a graph attention network (GAT) and learn structure in future via a multi-task scheme, respectively. Experimental results show that our approach significantly outperforms a strong baseline and achieves performance with 85.5 ±0.1 and 84.2 ±0.1 Smatch scores on AMR 2.0 and AMR 3.0, respectively

We tackle Event Argument Extraction (EAE) in the manner of template-based generation. Based on our exploration of generative EAE, it suffers from several issues, such as multiple arguments of one role, generating words out of context and inconsistency with prescribed format. We attribute it to the weakness of following complex input prompts. To address these problems, we propose the demonstration retrieval-augmented generative EAE (DRAGEAE), containing two components: event knowledge-injected generator (EKG) and demonstration retriever (DR). EKG employs event knowledge prompts to capture role dependencies and semantics. DR aims to search informative demonstrations from training data, facilitating the conditional generation of EKG. To train DR, we use the probability-based rankings from large language models (LLMs) as supervised signals. Experimental results on ACE-2005, RAMS and WIKIEVENTS demonstrate that our method outperforms all strong baselines and it can be generalized to various datasets. Further analysis is conducted to discuss the impact of diverse LLMs and prove that our model alleviates the above issues.

Employing pre-trained generation models for cross-domain aspect-based sentiment classification has recently led to large improvements. However, they ignore the importance of syntactic structures, which have shown appealing effectiveness in classification based models. Different from previous studies, efficiently encoding the syntactic structure in generation model is challenging because such models are pretrained on natural language, and modeling structured data may lead to catastrophic forgetting of distributional knowledge. In this study, we propose a novel structure-aware generation model to tackle this challenge. In particular, a prompt-driven strategy is designed to bridge the gap between different domains, by capturing implicit syntactic information from the input and output sides. Furthermore, the syntactic structure is explicitly encoded into the structure-aware generation model, which can effectively learn domain-irrelevant features based on syntactic pivot features. Empirical results demonstrate the effectiveness of the proposed structure-aware generation model over several strong baselines. The results also indicate the proposed model is capable of leveraging the input syntactic structure into the generation model.

Aspect Sentiment Understanding (ASU) in interactive scenarios (e.g., Question-Answering and Dialogue) has attracted ever-more interest in recent years and achieved important progresses. However, existing studies on interactive ASU largely ignore the coreference issue for opinion targets (i.e., aspects), while this phenomenon is ubiquitous in interactive scenarios especially dialogues, limiting the ASU performance. Recently, large language models (LLMs) shows the powerful ability to integrate various NLP tasks with the chat paradigm. In this way, this paper proposes a new Chat-based Aspect Sentiment Understanding (ChatASU) task, aiming to explore LLMs’ ability in understanding aspect sentiments in dialogue scenarios. Particularly, this ChatASU task introduces a sub-task, i.e., Aspect Chain Reasoning (ACR) task, to address the aspect coreference issue. On this basis, we propose a Trusted Self-reflexion Approach (TSA) with ChatGLM as backbone to ChatASU. Specifically, this TSA treats the ACR task as an auxiliary task to boost the performance of the primary ASU task, and further integrates trusted learning into reflexion mechanisms to alleviate the LLMs-intrinsic factual hallucination problem in TSA. Furthermore, a high-quality ChatASU dataset is annotated to evaluate TSA, and extensive experiments show that our proposed TSA can significantly outperform several state-of-the-art baselines, justifying the effectiveness of TSA to ChatASU and the importance of considering the coreference and hallucination issues in ChatASU.

Multimodal Conversational Emotion (MCE) detection, generally spanning across the acoustic, vision and language modalities, has attracted increasing interest in the multimedia community. Previous studies predominantly focus on learning contextual information in conversations with only a few considering the topic information in single language modality, while always neglecting the acoustic and vision topic information. On this basis, we propose a model-agnostic Topic-enriched Diffusion (TopicDiff) approach for capturing multimodal topic information in MCE tasks. Particularly, we integrate the diffusion model into neural topic model to alleviate the diversity deficiency problem of neural topic model in capturing topic information. Detailed evaluations demonstrate the significant improvements of TopicDiff over the state-of-the-art MCE baselines, justifying the importance of multimodal topic information to MCE and the effectiveness of TopicDiff in capturing such information. Furthermore, we observe an interesting finding that the topic information in acoustic and vision is more discriminative and robust compared to the language.

Weakly-supervised Phrase Grounding (WPG) is an emerging task of inferring the fine-grained phrase-region matching, while merely leveraging the coarse-grained sentence-image pairs for training. However, existing studies on WPG largely ignore the implicit phrase-region matching relations, which are crucial for evaluating the capability of models in understanding the deep multimodal semantics. To this end, this paper proposes an Implicit-Enhanced Causal Inference (IECI) approach to address the challenges of modeling the implicit relations and highlighting them beyond the explicit. Specifically, this approach leverages both the intervention and counterfactual techniques to tackle the above two challenges respectively. Furthermore, a high-quality implicit-enhanced dataset is annotated to evaluate IECI and detailed evaluations show the great advantages of IECI over the state-of-the-art baselines. Particularly, we observe an interesting finding that IECI outperforms the advanced multimodal LLMs by a large margin on this implicit-enhanced dataset, which may facilitate more research to evaluate the multimodal LLMs in this direction.

Cross-domain Named Entity Recognition (CD-NER) is crucial for Knowledge Graph (KG) construction and natural language processing (NLP), enabling learning from source to target domains with limited data.Previous studies often rely on manually collected entity-relevant sentences from the web or attempt to bridge the gap between tokens and entity labels across domains.These approaches are time-consuming and inefficient, as these data are often weakly correlated with the target task and require extensive pre-training.To address these issues, we propose automatically generating task-oriented knowledge (GTOK) using large language models (LLMs), focusing on the reasoning process of entity extraction.Then, we employ taskoriented pre-training (TOPT) to facilitate domain adaptation.Additionally, current crossdomain NER methods often lack explicit explanations for their effectiveness.Therefore, we introduce the concept of information density to better evaluate the model's effectiveness before performing entity recognition.We conduct systematic experiments and analyses to demonstrate the effectiveness of our proposed approach and the validity of using information density for model evaluation † .

Aspect term extraction (ATE) is an important Natural Language Processing task, which aims to extract aspect terms from reviews. Recently, data augmentation has emerged as a reliable approach for relieving data sparsity in the NLP area. For ATE, self-labeling and semi-generation methods have been proposed to implement effective data augmentation. However, they either rely on external data or a pretrained generation model. In this paper, we propose a simple and self-contained augmentation method, which produces new instances for augmentation by context decoupling and infrequent term refilling, without using external data and generation models. We conduct experiments on four benchmark SemEval datasets. The test results show that our method yields substantial improvements, and performs comparably to the state-of-the-art method which uses external data.

The growing importance of multi-modal humor detection within affective computing correlates with the expanding influence of short-form video sharing on social media platforms. In this paper, we propose a novel two-branch hierarchical model for short-form video humor detection (SVHD), named Comment-aided Video-Language Alignment (CVLA) via data-augmented multi-modal contrastive pre-training. Notably, our CVLA not only operates on raw signals across various modal channels but also yields an appropriate multi-modal representation by aligning the video and language components within a consistent semantic space. The experimental results on two humor detection datasets, including DY11k and UR-FUNNY, demonstrate that CVLA dramatically outperforms state-of-the-art and several competitive baseline approaches. Our dataset and code release at https://github.com/yliu-cs/CVLA.

Conversation is a common form of human communication that includes extensive emotional interaction. Traditional approaches focused on studying emotions and their underlying causes in conversations. They try to address two issues: what emotions are present in the dialogue and what causes these emotions. However, these works often overlook the bidirectional nature of emotional interaction in dialogue: utterances can evoke emotions (cause), and emotions can also lead to certain utterances (consequence). Therefore, we propose a new issue: what consequences arise from these emotions? This leads to the introduction of a new task called Emotion Consequence Forecasting in CONversations (ECFCON). In this work, we first propose a corresponding dialogue-level dataset. Specifically, we select 2,780 video dialogues for annotation, totaling 39,950 utterances. Out of these, 12,391 utterances contain emotions, and 8,810 of these have discernible consequences. Then, we benchmark this task by conducting experiments from the perspectives of traditional methods, generalized LLMs prompting methods, and clue-driven hybrid methods. Both our dataset and benchmark codes are openly accessible to the public.

In the literature, existing studies on text-to-motion generation (TMG) routinely focus on exploring the objective alignment of text and motion, which largely ignore the subjective emotion information, especially the limb-level emotion information. With this in mind, this paper proposes a new Emotion-enriched Text-to-Motion Generation (ETMG) task, aiming to generate motions with the subjective emotion information. Further, this paper believes that injecting emotions into limbs (named intra-limb emotion injection) and ensuring the coordination and coherence of emotional motions after injecting emotion information (named inter-limb emotion disturbance) is rather important and challenging in this ETMG task. To this end, this paper proposes an LL M-guided Limb-level Emotion Manipulating ( L3 EM) approach to ETMG. Specifically, this approach designs an LLM-guided intra-limb emotion modeling block to inject emotion into limbs, followed by a graph-structured inter-limb relation modeling block to ensure the coordination and coherence of emotional motions. Particularly, this paper constructs a coarse-grained Emotional Text-to-Motion (EmotionalT2M) dataset and a fine-grained Limb -level Emotional Text-to-Motion (Limb-ET2M) dataset to justify the effectiveness of the proposed L3EM approach. Detailed evaluation demonstrates the significant advantage of our L3EM approach to ETMG over the state-of-the-art baselines. This justifies the importance of the limb-level emotion information for ETMG and the effectiveness of our L3EM approach in coherently manipulating such information.

In real-world recon-videos such as surveillance and drone reconnaissance videos, commonly used explicit language, acoustic and facial expressions information is often missing. However, these videos are always rich in anomalous sentiments (e.g., criminal tendencies), which urgently requires the implicit scene information (e.g., actions and object relations) to fast and precisely identify these anomalous sentiments. Motivated by this, this paper proposes a new chat-paradigm Implicit anomalous sentiment Discovering and grounding (IasDig) task, aiming to interactively, fast discovering and grounding anomalous sentiments in recon-videos via leveraging the implicit scene information (i.e., actions and object relations). Furthermore, this paper believes that this IasDig task faces two key challenges, i.e., scene modeling and scene balancing. To this end, this paper proposes a new Scene-enhanced Video Large Language Model named Hawkeye, i.e., acting like a raptor (e.g., a Hawk) to discover and locate prey, for the IasDig task. Specifically, this approach designs a graph-structured scene modeling module and a balanced heterogeneous MoE module to address the above two challenges, respectively. Extensive experimental results on our constructed scene-sparsity and scene-density IasDig datasets demonstrate the great advantage of Hawkeye to IasDig over the advanced Video-LLM baselines, especially on the metric of false negative rates. This justifies the importance of the scene information for identifying implicit anomalous sentiments and the impressive practicality of Hawkeye for real-world applications.

Speculation and Negation Identification focuses on the extraction of speculative and negative cues and scopes. Previous work relied on complete syntactic trees or simply fed sentences into pre-trained language models , which were confronted with poor generalization within and across datasets, and the limitations of training samples. Accordingly, we build a complete pipeline framework that firstly detects cues and then extracts scopes, and propose unified Machine Reading Comprehension paradigms for both cue detection and scope resolution on several datasets. To tackle the insufficiency of training sets and produce more useful samples with appropriate amount of lexical and syntactic knowledge, we apply data augmentation integrating lexical and syntactic features for scope resolution. Experimental results show that our model achieves higher performance than baselines on several publicly accessible corpora.

Previous studies on cross-domain aspect-based sentiment classification depend on the pivot features or utilize the target data for representation learning, which ignores the correlations between instances and words. In this study, we employ two strategies to connect different domains through tripartite graphs. Firstly, we employ aword-topic-instance tripartite graphto bridge the gap between different domains with the cross-domain topic distribution. The cross-domain topic distribution is learned by a neural topic model on different domains. Secondly, we use aword-pivot-instance tripartite graphto connect instances and words in different domains. The pivot clauses are generated from the instances, and are domain-independent and sentimentally aligned with the original text. Afterward, we employ graph convolutional networks to model over these tripartite graphs for cross-domain sentiment classification respectively. The experimental result shows that our model with tripartite graphs outperforms several competitive models. The result also indicates the effectiveness of the proposed tripartite graphs for cross-domain aspect-based sentiment classification.

Aspect-based sentiment analysis (ABSA) aims to extract the aspect terms from review text, and to predict the polarity towards the aspect term. Although neural models have achieved competitive results, there are still many challenges in this task. Firstly, there is irrelevant and noise information in the review text, and the offsets of aspect term boundary are hard to decide. In addition, sentiment is usually either expressed implicitly or shifted due to the occurrence of negation and rhetorical words. To tackle the above limitations, we propose a scope detection model to distinguish whether the words from the review text are relevant with the aspect term, and to filter irrelevant and noise information. In addition, we investigate a biaffine-based model to constrain the scope detection process of aspect term extraction. We further generate a simplified clause based on the scope of aspect term, and predict the polarity based on the simplified clause. Empirical studies show the effectiveness of our proposed model over several strong baselines. These also justify the importance of scope detection for aspect-based sentiment analysis.

This paper presents our pioneering effort in addressing a new and realistic scenario in multi-modal dialogue systems called Multi-modal Real-time Emotion Pre-recognition in Conversations (MREPC). The objective is to predict the emotion of a forthcoming target utterance that is highly likely to occur. We believe that this task can enhance the dialogue system's understanding of the interlocutor's state of mind, enabling it to prepare an appropriate response in advance. However, addressing MREPC poses the following challenges:1) Previous studies on emotion elicitation typically focus on textual modality and perform sentiment forecasting within a fixed contextual scenario. 2) Previous studies on multi-modal emotion recognition aim to predict the emotion of existing utterances, making it difficult to extend these approaches to MREPC due to the absence of the target utterance. To tackle these challenges, we construct two benchmark multi-modal datasets for MREPC and propose a task-specific multi-modal contrastive pre-training approach. This approach leverages large-scale unlabeled multi-modal dialogues to facilitate emotion pre-recognition for potential utterances of specific target speakers. Through detailed experiments and extensive analysis, we demonstrate that our proposed multi-modal contrastive pre-training architecture effectively enhances the performance of multi-modal real-time emotion pre-recognition in conversations.

Conventional humor analysis normally focuses on text, text-image pair, and even long video (e.g., monologue) scenarios. However, with the recent rise of short-form video sharing, humor detection in this scenario has not yet gained much exploration. To the best of our knowledge, there are two primary issues associated with short-form video humor detection (SVHD): 1) At present, there are no ready-made humor annotation samples in this scenario, and it takes a lot of manpower and material resources to obtain a large number of annotation samples; 2) Unlike the more typical audio and visual modalities, the titles (as opposed to simultaneous transcription in the lengthy film) and associated interactive comments in short-form videos may convey apparent humorous clues. Therefore, in this paper, we first collect and annotate a video dataset from DouYin (aka. TikTok in the world), namely DY24h, with hierarchical comments. Then, we also design a novel approach with comment-aided multi-modal heterogeneous pre-training (CMHP) to introduce comment modality in SVHD. Extensive experiments and analysis demonstrate that our CMHP beats several existing video-based approaches on DY24h, and that the comments modality further aids a better comprehension of humor. Our dataset, code and pre-trained models are available at https://github.com/yliu-cs/CMHP.

Relation triplets extraction (RTE) aims to extract all potential triplets of subject-object entities and the corresponding relation in a sentence, which is an extremely essential procedure in information extraction and knowledge graph construction. In the literature, dominant studies normally focus on text only, which neglects the fact that other modalities may have effective information available for entities and relations (e.g., the image accompanying the text on Twitter). Therefore, in this paper, we introduce a multi-modal task for relation triplets extraction (MRTE), which considers both text and image modalities. To well handle the multi-modality and achieve all potential triplets extraction, we propose a novel multi-modal approach by resorting to a unified machine reading comprehension (MRC) framework with multiple queries, namely MUMRC. Specifically, we design a unified framework to conduct multi-modal entity extraction, and multi-modal relation classification depending on all potential subject-object entity pairs, which recasts both sub-tasks as a machine reading comprehension (MRC) problem. In this unified MRC model, we incorporate multiple questions into original text data and leverage visual information to enhance the textual semantics. Extensive experiments on the MNRE dataset demonstrate that the proposed approach significantly outperforms the state-of-the-art baselines of textual RTE and multi-modal information extraction.

Event Detection (ED) is a challenging but valuable task. It aims to identify the words that trigger the events in text and classify them into pre-defined types. The previous works utilize entity information as supplementary clues for ED. However, word semantics in different contexts always have subtle variations. Capturing the connection between entity information and changing semantics is challenging. To tackle the issue, we propose the Mixture of Experts (MoE) technique with context clues to adaptively model semantics. We term the framework as CMoE. Concretely, the CMoE simultaneously performs entity and event detection with the MoE technique which are flexible components embedded transformer block. Furthermore, we apply multi-task learning to mine the shared knowledge between entity and event. We conduct experiments on the public ACE 2005 and KBP 2017 datasets. The results show that our model achieves competitive performance on ACE 2005 without using external knowledge, yielding an improvement of about 2.3% F1-score for ED. More importantly, our model outperforms all State-of-The-Art models on KBP 2017.

Question Generation (QG) aims to automatically raise questions given contexts. Due to the discrepancy between the training (supervised learning) and inference (autoregressive) modes, the current QG task suffers from exposure bias, i.e., the phenomenon of error propagation induced by mistakenly-generated tokens in the preceding subsequence. To address this issue, we propose to utilize moment matching to progressively alleviate exposure biases, which is referred to EBAMQ for short in this paper. EBAMQ enhances QG by imposing positive mitigation effects on the distributed representation. This is implemented by two components, including (1) adaptive moment matching and (2) assistant-cooperating data augmentation. Within the two components, adaptive moment matching plays the dominant role. It enables the sharing of feature space among different samples that derive from binary channels (i.e., training and inference). More importantly, it provides more similar representations for synonymous or homologous tokens. In addition, we introduce the pseudo-masked technique to generate additional reference questions. It contributes to the reduction of inevitable constraints of ground-truth instances, and therefore, it helps to improve the robustness. Extensive experiments are conducted on the benchmark dataset SQuAD. Experimental results show that EBAMD produces substantial improvements for all the metrics (BLEU-4, ROUGE-L and METEOR) compared to the UniLM-based baseline, and it outperforms the state-of-the-art QG models.

Question Generation (QG) models suffer from exposure bias. Exposure bias refers to the problem that an autoregressive generator is trained given the observable preceding context, though it runs during test conditioned on the self-generated context. Exposure bias unavoidably causes performance degradation. In this paper, we enhance the GAN-based SeqGAN model which slightly contributes to alleviating exposure bias. Instead of RNN and CNN, we leverage UniLMv2 to construct the generator and discriminator in SeqGAN. Different from RNN who generates the current token in terms of the closely adjacent token, UniLMv2 depends on all the preceding contexts to predict the current token. This reduces the risk of severe token-by-token error propagation caused by exposure bias. During the representation learning process, we retain the adversarial mechanism of SeqGAN that drives the generation of human-like languages, as well as the rewarding mechanism which reduces the loss if subsequent tokens are perfectly generated. Though, we carry out a straight rewording using N-grams. The experiments on the benchmark dataset SQuAD show that our model outperforms the state-of-the-art QG models for the metrics BLEU-4, ROUGE-L and METEOR. Besides, our model is less time-consuming compared to the original SeqGAN. The time consumption of 2nd-stage learning is 1h47m on GeForce RTX 2080 Ti.

We tackle Aspect Term Extraction (ATE), a task that automatically recognizes aspect terms conditioned on the under-standing of word-level semantics. Due to the capacity of enriching linguistic phenomena for learning, data augmentation contributes to the establishment of robust ATE models. In this paper, we propose to leverage back translation to augment the training data for ATE. It is grounded on the potential advantages that the back-translated instances generally appear as paraphrases, providing diverse pragmatic modes for learning when semantics remains unchanged. This helps to enhance ATE models in recognizing aspect terms when varied contexts and morphologically-different words occur during test. In our experiments, we apply an off-the-shelf Neural Machine Translation (NMT) model for back translation, using French, Chinese and German as interlanguages, respectively. Besides, word alignment is conducted to designate aspect terms in the back-translated cases. Experimental results on SemEval benchmarks show that retraining with the augmented data produces substantial improvements, reaching up to 3.46% at best. In addition, the experiments suggest that 1) family languages are more beneficial than non-family for the aforementioned data augmentation, and 2) selective sampling produces positive effects in the low-resource settings, It is noteworthy that back translation has been explored for data augmentation in other fields, with the aim to enhance neural language modeling. Nevertheless, it hasn't yet been systematically studied towards the ATE task. Although a vest-pocket method is provided in this paper, the comprehensive analysis is conducted, including that on interlanguage selection, low-resource application, as well as compatibility with both conventional and pretrained neural models, besides that in the common comparison and ablation experiments. All the models and codes in the experiments will be made publicly available to support reproducible research.

Comparative Opinion Quintuple Extraction (COQE) is a task of recognizing comparative relationships in a sentence-level comment. It is additionally required to extract the quintuple constituents that attribute to a specific comparative relation, including a subject and object, as well as comparative aspect, opinion and preference. Previous study employ pipeline models in addressing the COQE task, which tend to suffer from error propagation. To address the issue, in this paper, we propose a BERT-based end-to-end neural model as the alternative. Specifically, we first boil COQE down to a set prediction problem, due to the finding that all the quintuple constituents fail to hold coherent or sequential relationships. In other word, we consider a group of quintuple components as a set, and intent to bag-and-drag them as a whole, instead individually and one-by-one. Furthermore, we leverage Graph Convolutional Network (GCN) to enhance the end-to-end model, which plays the role of perceiving and representing the relevant relations among the quintuple components. We experiment on three benchmark datasets, including Camera-COQE, Car-COQE and Ele-COQE. The experimental results show that our model (GCN-E2E) yields a significant improvement in most cases, compared to the BERT-based pipeline baseline. The performance reaches the F1-scores of 14.10%, 36.46% and 39.29%, with the improvements of 0.74%, 6.71% and 8.56%.

Causal relation extraction of biomedical entities is one of the most complex tasks in biomedical text mining, which involves two kinds of information: entity relations and entity functions. One feasible approach is to take relation extraction and function detection as two independent sub-tasks. However, this separate learning method ignores the intrinsic correlation between them and leads to unsatisfactory performance. In this paper, we propose a joint learning model, which combines entity relation extraction and entity function detection to exploit their commonality and capture their inter-relationship, so as to improve the performance of biomedical causal relation extraction. Experimental results on the BioCreative-V Track 4 corpus show that our joint learning model outperforms the separate models in BEL statement extraction, achieving the F1 scores of 57.0% and 37.3% on the test set in Stage 2 and Stage 1 evaluations, respectively. This demonstrates that our joint learning system reaches the state-of-the-art performance in Stage 2 compared with other systems.

Emotion cause extraction (ECE) seeks to find out what causes a given emotion, which has drawn much attention in natural language and signal processing. Conventional ECE normally focuses on single level (i.e., either word-level or clause-level) in a document scenario. However, as we known, single level ECE can not satisfy the wide applications, compared with both levels. Besides, the existing dialogue systems increasingly need empathy support. Therefore, in this paper, we propose to hierarchically extract both word and utterance-level emotion causes in the spoken dialogue scenario (HECE). We first construct two datasets (i.e., HECE-DD and HECE-IE) based on previous studies, then propose a hierarchical framework, which consists of a feature extractor, an utterance-level cause extractor, and a word-level cause extractor. In this framework, utterance and word levels can naturally preform interaction. Detailed experiments on two HECE datasets demonstrate that hierarchical extraction performs better than extraction on a single level.

Knowledge distillation is an effective method to transfer knowledge from a large pre-trained teacher model to a compacted student model. However, in previous studies, the distilled student models are still large and remain impractical in highly speed-sensitive systems (e.g., an IR system). In this study, we aim to distill a deep pre-trained model into an extremely compacted shallow model like CNN. Specifically, we propose a novel one-teacher and multiple-student knowledge distillation approach to distill a deep pre-trained teacher model into multiple shallow student models with ensemble learning. Moreover, we leverage large-scale unlabeled data to improve the performance of students. Empirical studies on three sentiment classification tasks demonstrate that our approach achieves better results with much fewer parameters (0.9%-18%) and extremely high speedup ratios (100X-1000X).

We leverage cross-language data expansion and retraining to enhance neural Event Detection (abbr., ED) on English ACE corpus. Machine translation is utilized for expanding English training set of ED from that of Chinese. However, experimental results illustrate that such strategy actually results in performance degradation. The survey of translations suggests that the mistakenly-aligned triggers in the expanded data negatively influences the retraining process. We refer this phenomenon to “trigger falsification”. To overcome the issue, we apply heuristic rules for regulating the expanded data, fixing the distracting samples that contain the falsified triggers. The supplementary experiments show that the rule-based regulation is beneficial, yielding the improvement of about 1.6% F1-score for ED. We additionally prove that, instead of transfer learning from the translated ED data, the straight data combination by random pouring surprisingly performs better.

Pre-trained masked language models have demonstrated remarkable ability as few-shot learners. In this paper, as an alternative, we propose a novel approach to few-shot learning with pre-trained token-replaced detection models like ELECTRA. In this approach, we reformulate a classification or a regression task as a token-replaced detection problem. Specifically, we first define a template and label description words for each task and put them into the input to form a natural language prompt. Then, we employ the pre-trained token-replaced detection model to predict which label description word is the most original (i.e., least replaced) among all label description words in the prompt. A systematic evaluation on 16 datasets demonstrates that our approach outperforms few-shot learners with pre-trained masked language models in both one-sentence and two-sentence learning tasks.

Document-level Event Factuality Identification (DEFI) predicts the factuality of a specific event based on a document from which the event can be derived, which is a fundamental and crucial task in Natural Language Processing (NLP). However, most previous studies only considered sentence-level task and did not adopt document-level knowledge. Moreover, they modelled DEFI as a typical text classification task depending on annotated information heavily, and limited to the task-specific corpus only, which resulted in data scarcity. To tackle these issues, we propose a new framework formulating DEFI as Machine Reading Comprehension (MRC) tasks considering both Span-Extraction (Ext) and Multiple-Choice (Mch). Our model does not employ any other explicit annotated information, and utilizes Transfer Learning (TL) to extract knowledge from universal large-scale MRC corpora for cross-domain data augmentation. The empirical results on DLEFM corpus demonstrate that the proposed model outperforms several state-of-the-arts.

Training Neural Machine Translation (NMT) models suffers from sparse parallel data, in the infrequent translation scenarios towards low-resource source languages. The existing solutions primarily concentrate on the utilization of Parent-Child (PC) transfer learning. It transfers well-trained NMT models on high-resource languages (namely Parent NMT) to low-resource languages, so as to produce Child NMT models by fine-tuning. It has been carefully demonstrated that a variety of PC variants yield significant improvements for low-resource NMT. In this paper, we intend to enhance PC-based NMT by a bidirectionally-adaptive learning strategy. Specifically, we divide inner constituents (6 transformers) of Parent encoder into two “teams”, i.e., T1 and T2. During representation learning, T1 learns to encode low-resource languages conditioned on bilingual shareable latent space. Generative adversarial network and masked language modeling are used for space-shareable encoding. On the other hand, T2 is straightforwardly transferred to low-resource languages, and fine-tuned together with T1 for low-resource translation. Briefly, T1 and T2 take actions separately for different goals. The former aims to adapt to characteristics of low-resource languages during encoding, while the latter adapts to translation experiences learned from high-resource languages. We experiment on benchmark corpora SETIMES, conducting low-resource NMT for Albanian (Sq), Macedonian (Mk), Croatian (Hr) and Romanian (Ro). Experimental results show that our method yields substantial improvements, which allows the NMT performance to reach BLEU4-scores of 62.24%, 56.93%, 50.53% and 54.65% for Sq, Mk, Hr and Ro, respectively.

Recent work on document-level sentiment classification has shown that the sentiment in the original text is often hard to capture, since the sentiment is usually either expressed implicitly or shifted due to the occurrences of negation and rhetorical words.To this end, we enhance the original text with a sentiment-driven simplified clause to intensify its sentiment.The simplified clause shares the same opinion with the original text but expresses the opinion much more simply.Meanwhile, we employ Abstract Meaning Representation (AMR) for generating simplified clauses, since AMR explicitly provides core semantic knowledge, and potentially offers core concepts and explicit structures of original texts.Empirical studies show the effectiveness of our proposed model over several strong baselines.The results also indicate the importance of simplified clauses for sentiment classification.

We tackle Aspect Term Extraction (ATE), an important natural language processing task that automatically identifies words of domain-specific aspects. Recently, a variety of sophisticated neural models and learning strategies have been explored for enhancing ATE, and significant improvements have been obtained. We intend to strengthen the neural models by involving features of Part-Of-Speech (POS) and syntactic dependency into the encoding process. It is motivated by the empirical findings that features of linguistic structure help to refine the understanding of semantics. Accordingly, we propose a stepwise encoding approach, where POS and syntactic dependency are successively leveraged step by step, including 1) joint encoding over both word and POS sequences using a pretrained language model; 2) BiGRU-based representational refinement conditioned on semantics-aware POS information and POS-aware semantic information; 3) representational augmentation by convolutional encoding of dependency graph. We conduct experiments on the four benchmark datasets of Semantic Evaluation (SemEval) for ATE. Experimental results show that our method obtains substantial improvements on all the considered datasets, and the performance ($F1$-score) reaches 87.69%, 89.76%, 77.94% and 83.96% for L-14, R-14, R-15 and R-16, respectively. All the models and source codes in the experiments will be made publicly available to support reproducible research.

Implicit discourse relation classification refers to a task of automatically determining relationships between arguments. It has been widely proven that, in a neural classification architecture, decoding discourse relations heavily relies on the reliable semantic representations of arguments. In addition, our previous survey shows that, for a target argument, the external semantic information hidden in the accompanying argument benefits the encoding of the target, either wholly or partially. Moreover, dependency structure appears as the crucial feature for synthesizing word senses of the entire words in arguments. Accordingly, we propose a novel method to enhance the current representation learning of pairwise arguments, which takes into consideration both external semantic information and internal dependency structure. In particular, we inject external semantic information into the Long-Short Term Memory (LSTM) unit of Recurrent Neural Network (RNN) through the input and forget gates. Different from the existing one-off interactive learning models, our method allows the neuronal memory of internal argument semantics to be affected by external information at each encoding step. On the basis, we apply the parser-based Graph Convolutional Networks (GCN) over the semantic presentations of words, so as to accumulate the closely-related semantic information in terms of dependency structures. We conduct experiments on Penn Discourse TreeBank Corpus of version 2.0 (PDTB 2.0). The test results illustrate that the proposed method enhances the baseline significantly, and it obtains comparable performance compared to the state of the art.

The goal of Event Detection (ED) task is to identify the words that mark the occurrence of events in text, and classify them into a set of event types. To model informative word semantics, some researchers apply Graph Convolutional Network (GCN) to exploit the syntactic graph transformed from the dependency tree, within one sentence. We are motivated to simultaneously leverage syntactic clues and context information across sentences. To this end, we propose a novel ED model with Cross-Sentence Graph Convolutional Networks (CSGCN). The CSGCN contains two main components, including a tree extension module and the syntax-aware graph convolution. Each sentence is parsed to a dependency tree by an automatic toolkit. The first module merges entity-specific subtrees from neighbor sentences into the dependency tree of current sentence, which constructs a cross-sentence dependency tree. On this basis, we transform the tree into an undirected graph. After that, a syntax-aware attention mechanism is employed in the computation of graph convolution. This mechanism dynamically captures syntax-relevant information from neighbor nodes via the graph structure. Finally, we devise an entity aggregation module to aggregate key entity information for trigger candidates. We conduct experiments on the ACE 2005 and KBP 2017 datasets. The results show that our model achieves satisfactory and competitive performance on ACE 2005, and outperforms all State-of-The-Art models on KBP 2017.

Document-level Event Factuality Identification (DEFI) predicts the event factuality according to the current document, and mainly depends on event-related tokens and sentences. However, previous studies relied on annotated information and did not filter irrelevant and noisy texts. Therefore, this paper proposes a novel end-to-end model, i.e., Reinforced Multi-Granularity Hierarchical Attention Network (RMHAN), which can learn information at different levels of granularity from tokens and sentences hierarchically. Moreover, with hierarchical reinforcement learning, RMHAN first selects relevant and meaningful tokens, and then selects useful sentences for document-level encoding. Experimental results on DLEF-v2 corpus show that RMHAN model outperforms several state-of-the-art baselines and achieves the best performance.

Current document-level neural machine translation (NMT) systems have achieved remarkable progress with document context. Nevertheless, discourse information that has been proven effective in many NLP tasks is ignored in most previous work. In this work, we aim at incorporating the coherence information hidden within the RST-style discourse structure into machine translation. To achieve it, we propose a document-level NMT system enhanced with the discourse-aware document context, which is named Disco2NMT. Specifically, Disco2NMT models document context based on the discourse dependency structures through a hierarchical architecture. We first convert the RST tree of an article into a dependency structure and then build the graph convolutional network (GCN) upon the segmented EDUs under the guidance of RST dependencies to capture the discourse-aware context for NMT incorporation. We conduct experiments on the document-level English-German and English-Chinese translation tasks with three domains (TED, News, and Europarl). Experimental results show that our Disco2NMT model significantly surpasses both context-agnostic and context-aware baseline systems on multiple evaluation indicators.

BACKGROUND: The COVID-19 pandemic has increasingly accelerated the publication pace of scientific literature. How to efficiently curate and index this large amount of biomedical literature under the current crisis is of great importance. Previous literature indexing is mainly performed by human experts using Medical Subject Headings (MeSH), which is labor-intensive and time-consuming. Therefore, to alleviate the expensive time consumption and monetary cost, there is an urgent need for automatic semantic indexing technologies for the emerging COVID-19 domain. RESULTS: In this research, to investigate the semantic indexing problem for COVID-19, we first construct the new COVID-19 Semantic Indexing dataset, which consists of more than 80 thousand biomedical articles. We then propose a novel semantic indexing framework based on the multi-probe attention neural network (MPANN) to address the COVID-19 semantic indexing problem. Specifically, we employ a k-nearest neighbour based MeSH masking approach to generate candidate topic terms for each input article. We encode and feed the selected candidate terms as well as other contextual information as probes into the downstream attention-based neural network. Each semantic probe carries specific aspects of biomedical knowledge and provides informatively discriminative features for the input article. After extracting the semantic features at both term-level and document-level through the attention-based neural network, MPANN adopts a linear multi-view classifier to conduct the final topic prediction for COVID-19 semantic indexing. CONCLUSION: The experimental results suggest that MPANN promises to represent the semantic features of biomedical texts and is effective in predicting semantic topics for COVID-19 related biomedical articles.

How to effectively model global context has been a critical challenge for document-level neural machine translation (NMT). Both preceding and global context have been carefully explored in the sequence-to-sequence (seq2seq) framework. However, previous studies generally map global context into one vector, which is not enough to well represent the entire document since this largely ignores the hierarchy between sentences and words within. In this article, we propose to model global context for source language from both sentence level and word level. Specifically at sentence level, we extract useful global context for the current sentence, while at word level, we compute global context against words within the current sentence. On this basis, both kinds of global context can be appropriately fused before being incorporated into the state-of-the-art seq2seq model, i.e., Transformer . Detailed experimentation on various document-level translation tasks shows that global context at both sentence level and word level significantly improve translation performance. More encouraging, both kinds of global context are complementary. This leads to more improvement when both kinds of global context are used.

End-to-end neural modeling with the encoder-decoder architecture has shown great promise in response generation. However, it often generates dull and generic responses due to its failure to effectively perceive various kinds of act, sentiment, and topic information. To address these challenges, we propose a response-generation model withstructure-aware constraintsto capture the structure of dialog and generate a better response with various constraints of the act, sentiment, and topic. In particular, given an utterance sequence, we first learn the representation of each utterance in the encoding stage. We then learn the turn, speaker, and dialog representation from the utterance representations and construct the structure of dialog. Third, we employ an attention mechanism to extract the constraints of act, sentiment, and topic based on the structure of the dialog. Finally, we utilize these structure-aware constraints to control the response-generation process in decoding stage. Extensive experimental results validate the superiority of our proposed model against the state-of-the-art baselines. In addition, the results also show that the proposed model can generate responses with more appropriate content based on the structure-aware constraints.

Emotion recognition in conversation is one of the essential tasks of natural language processing. However, this task’s annotation data is insufficient since such data is hard to collect and annotate. Meanwhile, there is large-scale data for conversational generation, and this data does not need annotation manually. But, whether the vector space between different datasets is similar will be a problem. Therefore, we utilize a same dataset to train the conversational generator and the classifier, and transfer knowledge between them. In particular, we propose an Emotion Recognition with Conversational Generation Transfer (ERCGT) framework to model the interaction among utterances by transfer learning. First, we train a conversational generator. In the second step, a transfer learning model is used to transfer the knowledge of generator to the emotion recognition model. Empirical studies illustrate the effectiveness of the proposed framework over several strong baselines on three benchmark emotion classification datasets.

As an important research issue in affective computing community, multi-modal emotion recognition has become a hot topic in the last few years. However, almost all existing studies perform multiple binary classification for each emotion with focus on complete time series data. In this paper, we focus on multi-modal emotion recognition in a multi-label scenario. In this scenario, we consider not only the label-to-label dependency, but also the feature-to-label and modality-to-label dependencies. Particularly, we propose a heterogeneous hierarchical message passing network to effectively model above dependencies. Furthermore, we propose a new multi-modal multi-label emotion dataset based on partial time-series content to show predominant generalization of our model. Detailed evaluation demonstrates the effectiveness of our approach.

Linqing Chen, Junhui Li, Zhengxian Gong, Boxing Chen, Weihua Luo, Min Zhang, Guodong Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Dongqin Xu, Junhui Li, Muhua Zhu, Min Zhang, Guodong Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Longyin Zhang, Fang Kong, Guodong Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Aspect terms extraction (ATE) and aspect sentiment classification (ASC) are two fundamental and fine-grained sub-tasks in aspect-level sentiment analysis (ALSA).In the textual analysis, jointly extracting both aspect terms and sentiment polarities has been drawn much attention due to the better applications than individual sub-task.However, in the multimodal scenario, the existing studies are limited to handle each sub-task independently, which fails to model the innate connection between the above two objectives and ignores the better applications.Therefore, in this paper, we are the first to jointly perform multi-modal ATE (MATE) and multi-modal ASC (MASC), and we propose a multi-modal joint learning approach with auxiliary cross-modal relation detection for multi-modal aspect-level sentiment analysis (MALSA).Specifically, we first build an auxiliary text-image relation detection module to control the proper exploitation of visual information.Second, we adopt the hierarchical framework to bridge the multi-modal connection between MATE and MASC, as well as separately visual guiding for each sub module.Finally, we can obtain all aspect-level sentiment polarities dependent on the jointly extracted specific aspects.Extensive experiments show the effectiveness of our approach against the joint textual approaches, pipeline and collapsed multi-modal approaches.

Natural language generation (NLG) tasks on pro-drop languages are known to suffer from zero pronoun (ZP) problems, and the problems remain challenging due to the scarcity of ZP-annotated NLG corpora.In this case, we propose a highly adaptive two-stage approach to couple context modeling with ZP recovering to mitigate the ZP problem in NLG tasks.Notably, we frame the recovery process in a task-supervised fashion where the ZP representation recovering capability is learned during the NLG task learning process, thus our method does not require NLG corpora annotated with ZPs.For system enhancement, we learn an adversarial bot to adjust our model outputs to alleviate the error propagation caused by mis-recovered ZPs.Experiments on three document-level NLG tasks, i.e., machine translation, question answering, and summarization, show that our approach can improve the performance to a great extent, and the improvement on pronoun translation is very impressive.

In this paper, we address a novel Cognition-driven Real-time Personality Detection (C-RPD) task, aiming to predict the personality trait (e.g., romantic and humorous) real-time shown by a human being from the perspective of cognitive psychology. Specifically, this task is motivated by a cognition difference phenomenon that humans with different personality traits tend to focus on different portions of the image and then give different personality-oriented language descriptions when observing an image. On this basis, we propose a new Language-guided Contrastive Visual Attention (L-CVA) approach to capture the above cognition difference information for addressing the C-RPD task. Experimental results on a real-world multimodal personality corpus verify the great advantage of our L-CVA approach to C-RPD over the state-of-the-art baselines. This justifies the importance of the cognition difference information to C-RPD and the effectiveness of our approach in capturing such information.

Document context-aware machine translation remains challenging due to the lack of large-scale document parallel corpora. To make full use of source-side monolingual documents for context-aware NMT, we propose a Pre-training approach with Global Context (PGC). In particular, we first propose a novel self-supervised pre-training task, which contains two training objectives: (1) reconstructing the original sentence from a corrupted version; (2) generating a gap sentence from its left and right neighbouring sentences. Then we design a universal model for PGC which consists of a global context encoder, a sentence encoder and a decoder, with similar architecture to typical context-aware NMT models. We evaluate the effectiveness and generality of our pre-trained PGC model by adapting it to various downstream context-aware NMT models. Detailed experimentation on four different translation tasks demonstrates that our PGC approach significantly improves the translation performance of context-aware NMT. For example, based on the state-of-the-art SAN model, we achieve an averaged improvement of 1.85 BLEU scores and 1.59 Meteor scores on the four translation tasks.

Sentiment analysis in dialogues plays a critical role in dialogue data analysis. However, previous studies on sentiment classification in dialogues largely ignore topic information, which is important for capturing overall information in some types of dialogues. In this study, we focus on the sentiment classification task in an important type of dialogue, namely customer service dialogue, and propose a novel approach which captures overall information to enhance the classification performance. Specifically, we propose a topic-aware multi-task learning (TML) approach which learns topic-enriched utterance representations in customer service dialogue by capturing various kinds of topic information. In the experiment, we propose a large-scale and high-quality annotated corpus for the sentiment classification task in customer service dialogue and empirical studies on the proposed corpus show that our approach significantly outperforms several strong baselines.

In the literature, existing studies always consider Aspect Sentiment Classification (ASC) as an independent sentence-level classification problem aspect by aspect, which largely ignore the document-level sentiment preference information, though obviously such information is crucial for alleviating the information deficiency problem in ASC. In this paper, we explore two kinds of sentiment preference information inside a document, i.e., contextual sentiment consistency w.r.t. the same aspect (namely intra-aspect sentiment consistency) and contextual sentiment tendency w.r.t. all the related aspects (namely inter-aspect sentiment tendency). On the basis, we propose a Cooperative Graph Attention Networks (CoGAN) approach for cooperatively learning the aspect-related sentence representation. Specifically, two graph attention networks are leveraged to model above two kinds of document-level sentiment preference information respectively, followed by an interactive mechanism to integrate the two-fold preference. Detailed evaluation demonstrates the great advantage of the proposed approach to ASC over the state-of-the-art baselines. This justifies the importance of the document-level sentiment preference information to ASC and the effectiveness of our approach capturing such information.

Due to its great importance in deep natural language understanding and various down-stream applications, text-level parsing of discourse rhetorical structure (DRS) has been drawing more and more attention in recent years.However, all the previous studies on text-level discourse parsing adopt bottom-up approaches, which much limit the DRS determination on local information and fail to well benefit from global information of the overall discourse.In this paper, we justify from both computational and perceptive points-of-view that the top-down architecture is more suitable for textlevel DRS parsing.On the basis, we propose a top-down neural architecture toward text-level DRS parsing.In particular, we cast discourse parsing as a recursive split point ranking task, where a split point is classified to different levels according to its rank and the elementary discourse units (EDUs) associated with it are arranged accordingly.In this way, we can determine the complete DRS as a hierarchical tree structure via an encoder-decoder with an internal stack.Experimentation on both the English RST-DT corpus and the Chinese CDTB corpus shows the great effectiveness of our proposed top-down approach towards textlevel DRS parsing.

From the perspective of health psychology, human beings with long-term and sustained negativity are highly probable to be diagnosed with depression.Inspired by this, we argue that the global topic information derived from user-generated contents (e.g., texts and images) is crucial to boost the performance of the depression detection task, though this information has been neglected by almost all previous studies on depression detection.To this end, we propose a new Multimodal Topic-enriched Auxiliary Learning (MTAL) approach, aiming at capturing the topic information inside different modalities (i.e., texts and images) for depression detection.Especially, in our approach, a modality-agnostic topic model is proposed to be capable of mining the topical clues from either the discrete textual signals or the continuous visual signals.On this basis, the topic modeling w.r.t. the two modalities are cast as two auxiliary tasks for improving the performance of the primary task (i.e., depression detection).Finally, the detailed evaluation demonstrates the great advantage of our MTAL approach to depression detection over the state-of-the-art baselines.This justifies the importance of the multimodal topic information to depression detection and the effectiveness of our approach in capturing such information.

Reading comprehension (RC) on social media such as Twitter is a critical and challenging task due to its noisy, informal, but informative nature. Most existing RC models are developed on formal datasets such as news articles and Wikipedia documents, which severely limit their performances when directly applied to the noisy and informal texts in social media. Moreover, these models only focus on a certain type of RC, extractive or generative, but ignore the integration of them. To well address these challenges, we come up with a noisy user-generated text-oriented RC model. In particular, we first introduce a set of text normalizers to transform the noisy and informal texts to the formal ones. Then, we integrate the extractive and the generative RC model by a multi-task learning mechanism and an answer selection module. Experimental results on TweetQA demonstrate that our NUT-RC model significantly outperforms the state-of-the-art social media-oriented RC models.

We tackle implicit discourse relation recognition.Both self-attention and interactive-attention mechanisms have been applied for attention-aware representation learning, which improves the current discourse analysis models.To take advantages of the two attention mechanisms simultaneously, we develop a propagative attention learning model using a cross-coupled two-channel network.We experiment on Penn Discourse Treebank.The test results demonstrate that our model yields substantial improvements over the baselines (BiLSTM and BERT).

Sentiment forecasting in dialog aims to predict the polarity of next utterance to come, and can help speakers revise their utterances in sentimental utterances generation.However, the polarity of next utterance is normally hard to predict, due to the lack of content of next utterance (yet to come).In this study, we propose a Neural Sentiment Forecasting (NSF) model to address inherent challenges.In particular, we employ a neural simulation model to simulate the next utterance based on the context (previous utterances encountered).Moreover, we employ a sequence influence model to learn both pair-wise and seq-wise influence.Empirical studies illustrate the importance of proposed sentiment forecasting task, and justify the effectiveness of our NSF model over several strong baselines.

In the literature, the research on abstract meaning representation (AMR) parsing is much restricted by the size of human-curated dataset which is critical to build an AMR parser with good performance.To alleviate such data size restriction, pre-trained models have been drawing more and more attention in AMR parsing.However, previous pre-trained models, like BERT, are implemented for general purpose which may not work as expected for the specific task of AMR parsing.In this paper, we focus on sequence-to-sequence (seq2seq) AMR parsing and propose a seq2seq pre-training approach to build pre-trained models in both single and joint way on three relevant tasks, i.e., machine translation, syntactic parsing, and AMR parsing itself.Moreover, we extend the vanilla fine-tuning method to a multi-task learning fine-tuning method that optimizes for the performance of AMR parsing while endeavors to preserve the response of pre-trained models.Extensive experimental results on two English benchmark datasets show that both the single and joint pre-trained models significantly improve the performance (e.g., from 71.5 to 80.2 on AMR 2.0), which reaches the state of the art.The result is very encouraging since we achieve this with seq2seq models rather than complex models.We make our code and model available at https:// github.com/xdqkid/S2S-AMR-Parser.

As an important research issue in the natural language processing community, multi-label emotion detection has been drawing more and more attention in the last few years. However, almost all existing studies focus on one modality (e.g., textual modality). In this paper, we focus on multi-label emotion detection in a multi-modal scenario. In this scenario, we need to consider both the dependence among different labels (label dependence) and the dependence between each predicting label and different modalities (modality dependence). Particularly, we propose a multi-modal sequence-to-set approach to effectively model both kinds of dependence in multi-modal multi-label emotion detection. The detailed evaluation demonstrates the effectiveness of our approach.

Recently, speaker personality analysis has become an increasingly popular research task in human-computer interaction. Previous studies of user personality traits recognition normally focus on leveraging static information, i.e., tweets, images and social relationships in social platforms and websites. However, in this paper, we utilize three kinds of speaking dynamic information, i.e., textual, visual and acoustic temporal sequences, for a computer to interpret human personality traits from a face-to-face monologue. Specifically, we propose an explicit many2many (many-to-many) interactive approach to help AI efficiently recognize speaker personality traits. On the one hand, we encode the long feature sequence of human speaking for each modality with bidirectional LSTM network. On the other hand, we design a many2many attention mechanism explicitly to capture the interactions across multiple modalities for multiple interactive pairs. Empirical evaluation on 12 kinds of personality traits demonstrates the effectiveness of our proposed approach to multimodal speaker personality recognition.

Multi-modal utterance-level emotion detection has been a hot research topic in both multi-modal analysis and natural language processing communities. Different from traditional single-label multi-modal sentiment analysis, typical multi-modal emotion detection is naturally a multi-label problem where an utterance often contains multiple emotions. Existing studies normally focus on multi-modal fusion only and transform multi-label emotion classification into multiple binary classification problem independently. As a result, existing studies largely ignore two kinds of important dependency information: (1) Modality-to-label dependency, where different emotions can be inferred from different modalities, that is, different modalities contribute differently to each potential emotion. (2) Label-to-label dependency, where some emotions are more likely to coexist than those conflicting emotions. To simultaneously model above two kinds of dependency, we propose a unified approach, namely multi-modal emotion set generation network (MESGN) to generate an emotion set for an utterance. Specifically, we first employ a cross-modal transformer encoder to capture cross-modal interactions among different modalities, and a standard transformer encoder to capture temporal information for each modality-specific sequence given previous interactions. Then, we design a transformer-based discriminative decoding module equipped with modality-to-label attention to handle the modality-to-label dependency. In the meanwhile, we employ a reinforced decoding algorithm with self-critic learning to handle the label-to-label dependency. Finally, we validate the proposed MESGN architecture on a word-level aligned and unaligned multi-modal dataset. Detailed experimentation shows that our proposed MESGN architecture can effectively improve the performance of multi-modal multi-label emotion detection.

Through much exploration in the past decade, emotion analysis in conversations was mainly conducted in textual scenario. Nowadays, with the popularization of speech and video communication, academia and industry have become gradually aware of the need in multimodal scenario. Therefore, emotion detection in conversations becomes increasingly hot not only in natural language processing (NLP) community but also in multimodal analysis community. Although previous studies normally argue that the emotion of current utterance in a conversation is much influenced by the content of historical utterances, their speakers and emotions, they model the influence derived from the history to the current utterance at the same granularity (Intra-modal influence). Intuitively, the clues of emotion detection may not exist in the history of the same modality as current utterance, but in the history of other modalities (Inter-modal influence). Besides, previous studies normally model the information propagation as the conversation flow. Intuitively, bidirectional modeling of information propagation in conversations provides rich clues for emotion detection. Therefore, this paper proposes a bidirectional dynamic dual influence network for real-time emotion detection in conversations, which can simultaneously model both intra- and inter-modal influence with bidirectional information propagation for current utterance and its historical utterances. Detailed experiments demonstrate that our approach much advances the state-of-the-art.

Sentiment classification aims to detect polarity from a piece of text. The polarity is usually positive or negative, and the text genre is usually product review. The challenges of sentiment classification are that it is hard to capture semantic of reviews, and the labeled data is hard to annotate. Therefore, we propose neural co-training to learn the semantic representation of each review using the neural network model, and learn the information from unlabeled data using a co-training framework. In particular, we use the attention-based bi-directional Gated Recurrent Unit (Att-BiGRU) to model the semantic content of each review and regard different categories of the target product as different views. We then use a co-training framework to learn and predict the unlabeled reviews with different views. Experiment results with the Yelp dataset demonstrate the effectiveness of our approach.

In this article, we show that word translations can be explicitly incorporated into NMT effectively to avoid wrong translations. Specifically, we propose three cross-lingual encoders to explicitly incorporate word translations into NMT: (1) Factored encoder, which encodes a word and its translation in a vertical way; (2) Gated encoder, which uses a gated mechanism to selectively control the amount of word translations moving forward; and (3) Mixed encoder, which stitchingly learns a word and its translation annotations over sequences where words and their translations are alternatively mixed. Besides, we first use a simple word dictionary approach and then a word sense disambiguation (WSD) approach to effectively model the word context for better word translation. Experimentation on Chinese-to-English translation demonstrates that all proposed encoders are able to improve the translation accuracy for both traditional RNN-based NMT and recent self-attention-based NMT (hereafter referred to as Transformer ). Specifically, Mixed encoder yields the most significant improvement of 2.0 in BLEU on the RNN-based NMT, while Gated encoder improves 1.2 in BLEU on Transformer . This indicates the usefulness of an WSD approach in modeling word context for better word translation. This also indicates the effectiveness of our proposed cross-lingual encoders in explicitly modeling word translations to avoid wrong translations in NMT. Finally, we discuss in depth how word translations benefit different NMT frameworks from several perspectives.

Chinese zero pronoun (ZP) resolution plays a critical role in discourse analysis. Different from traditional mention-to-mention approaches, this article proposes a chain-to-chain approach to improve the performance of ZP resolution in three aspects. First, consecutive ZPs are clustered into coreferential chains, each working as one independent anaphor as a whole. In this way, those ZPs far away from their overt antecedents can be bridged via other consecutive ZPs in the same coreferential chains and thus better resolved. Second, common noun phrases (NPs) are automatically grouped into coreferential chains using traditional approaches, each working as one independent antecedent candidate as a whole. That is, those NPs occurring in the same coreferential chain are viewed as one antecedent candidate as a whole, and ZP resolution is made between ZP coreferential chains and common NP coreferential chains. In this way, the performance can be much improved due to the effective reduction of the search space by pruning singletons and negative instances. Third and finally, additional features from ZP and common NP coreferential chains are employed to better represent anaphors and their antecedent candidates, respectively. Comprehensive experiments on the OntoNotes V5.0 corpus show that our chain-to-chain approach significantly outperforms the state-of-the-art mention-to-mention approaches. To our knowledge, this is the first work to resolve zero pronouns in a chain-to-chain way.

Text summarization is one of the significant tasks of natural language processing, which automatically converts text into a summary. Some summarization systems, for short/long English, and short Chinese text, benefit from advances in the neural encoder-decoder model because of the availability of large datasets. However, the long Chinese text summarization research has been limited to datasets of a couple of hundred instances. This article aims to explore the long Chinese text summarization task. To begin with, we construct a first large-scale, long Chinese text summarization corpus, the Long Chinese Summarization of Police Inquiry Record Text (LCSPIRT). Based on this corpus, we propose a sequence-to-sequence (Seq2Seq) model that incorporates a global encoding process with an attention mechanism. Our model achieves a competitive result on the LCSPIRT corpus compared with several benchmark methods.

Language discrimination among similar languages, varieties, and dialects is a challenging natural language processing task. The traditional text-driven focus leads to poor results. In this article, we explore the effectiveness of speech-driven features toward language discrimination among Chinese dialects. First, we systematically explore the appropriateness of speech-driven MFCC features toward CNN-based language discrimination. Then, we design an end-to-end speech recognition model based on HMM-DNN to predict Chinese dialect words. We adopt attention mechanism to extract the discriminative words related to different Chinese dialects. Finally, through a CNN, we combine the word-level embedding and the MFCC-based features. Evaluation of two benchmark Chinese dialect corpora shows the appropriateness and effectiveness of the proposed speech-driven approach to fine-grained Chinese dialect discrimination compared to the state-of-the-art methods.

Stance detection aims to assign a stance label (i.e., favor or against) to a post towards a specific target. Recently, there is a growing interest in adopting neural models to detect stance of a document. However, most of these works focus on modeling the sequence of words to learn document representation, though other linguistic information, such as sentiment and arguments, are correlated with the stance of document, and may inspire us to explore the stance. In this article, we propose a hierarchical attention neural model to well study various linguistic information to better represent a document via hierarchical linguistic representations. In addition, we propose a hierarchical network with attention mechanism to weight the importance of various kinds of linguistic information, and learn the mutual attention between document and linguistic information. Detail evaluation on two benchmark datasets demonstrates the effectiveness of proposed hierarchical network with attention mechanism.

In the literature, existing studies on aspect sentiment classification (ASC) focus on individual non-interactive reviews.This paper extends the research to interactive reviews and proposes a new research task, namely Aspect Sentiment Classification towards Question-Answering (ASC-QA), for real-world applications.This new task aims to predict sentiment polarities for specific aspects from interactive QA style reviews.In particular, a high-quality annotated corpus is constructed for ASC-QA to facilitate corresponding research.On this basis, a Reinforced Bidirectional Attention Network (RBAN) approach is proposed to address two inherent challenges in ASC-QA, i.e., semantic matching between question and answer, and data noise.Experimental results demonstrate the great advantage of the proposed approach to ASC-QA against several state-of-the-art baselines.

In the literature, most of the previous studies on English implicit discourse relation recognition only use sentence-level representations, which cannot provide enough semantic information in Chinese due to its unique paratactic characteristics.In this paper, we propose a topic tensor network to recognize Chinese implicit discourse relations with both sentencelevel and topic-level representations.In particular, besides encoding arguments (discourse units) using a gated convolutional network to obtain sentence-level representations, we train a simplified topic model to infer the latent topic-level representations.Moreover, we feed the two pairs of representations to two factored tensor networks, respectively, to capture both the sentence-level interactions and topiclevel relevance using multi-slice tensors.Experimentation on CDTB, a Chinese discourse corpus, shows that our proposed model significantly outperforms several state-of-the-art baselines in both micro and macro F1-scores.

In this paper, we propose an Adversarial Attention Network for the task of multidimensional emotion regression, which automatically rates multiple emotion dimension scores for an input text.Especially, to determine which words are valuable for a particular emotion dimension, an attention layer is learnt to weight the words in an input sequence.Furthermore, adversarial training is employed between two attention layers to learn better word weights via a discriminator.In particular, a shared attention layer is incorporated to learn public word weights between two emotion dimensions.Empirical evaluation on the EMOBANK corpus shows that our approach achieves notable improvements in r-values on both EMOBANK Reader's and Writer's multi-dimensional emotion regression tasks in all domains over the state-of-the-art baselines.

Jun Gao, Wei Bi, Xiaojiang Liu, Junhui Li, Guodong Zhou, Shuming Shi. 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.

Longxiang Shen, Bowei Zou, Yu Hong, Guodong Zhou, Qiaoming Zhu, AiTi Aw. 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.

Xin Tan, Longyin Zhang, Deyi Xiong, Guodong Zhou. 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.

Jingjing Wang, Changlong Sun, Shoushan Li, Jiancheng Wang, Luo Si, Min Zhang, Xiaozhong Liu, Guodong Zhou. 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.

Xiabing Zhou, Zhongqing Wang, Shoushan Li, Guodong Zhou, Min 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.

Jie Zhu, Junhui Li, Muhua Zhu, Longhua Qian, Min Zhang, Guodong Zhou. 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.

In this paper, we propose a novel approach to multimodal sentiment analysis with focus on both textual and acoustic modalities. Especially, we utilize deep reinforcement learning to explore the clause-level structure in an utterance. On the basis, we perform multimodal interactions at clause-level to model hierarchical interactive representation for multimodal senitment analysis. Detailed evaluation on two benchmark datasets demonstrates the great effectiveness of our approach over several state-of-the-art baselines.

As an emerging research area in natural language processing, multi-modal human language analysis spans language, vision and audio modalities. Understanding multi-modal language requires not only the modeling of independent dynamics within each modality (intra-modal dynamics), but also more importantly interactive dynamics among different modalities (inter-modal dynamics). In this paper, we propose a hierarchical approach to multi-modal language analysis with two levels of attention mechanism, namely interaction-level, which captures the intra-modal and inter-modal dynamics across different modalities with multiple types of attention, and selection-level attention, which selects the effective representations for final prediction by calculating the importance of each vector obtained from interaction-level. Empirical evaluation demonstrates the effectiveness of our proposed approach to multi-modal sentiment classification, sentiment regression and emotion recognition.

Recently, emotion detection in conversations becomes a hot research topic in the Natural Language Processing community. In this paper, we focus on emotion detection in multi-speaker conversations instead of traditional two-speaker conversations in existing studies. Different from non-conversation text, emotion detection in conversation text has one specific challenge in modeling the context-sensitive dependence. Besides, emotion detection in multi-speaker conversations endorses another specific challenge in modeling the speaker-sensitive dependence. To address above two challenges, we propose a conversational graph-based convolutional neural network. On the one hand, our approach represents each utterance and each speaker as a node. On the other hand, the context-sensitive dependence is represented by an undirected edge between two utterances nodes from the same conversation and the speaker-sensitive dependence is represented by an undirected edge between an utterance node and its speaker node. In this way, the entire conversational corpus can be symbolized as a large heterogeneous graph and the emotion detection task can be recast as a classification problem of the utterance nodes in the graph. The experimental results on a multi-modal and multi-speaker conversation corpus demonstrate the great effectiveness of the proposed approach.

Computational modeling of human spoken language is an emerging research area in multimedia analysis spanning across the text and acoustic modalities. Multi-modal sentiment analysis is one of the most fundamental tasks in human spoken language understanding. In this paper, we propose a novel approach to selecting effective sentiment-relevant words for multi-modal sentiment analysis with focus on both the textual and acoustic modalities. Unlike the conventional soft attention mechanism, we employ a deep reinforcement learning mechanism to perform sentiment-relevant word selection and fully remove invalid words of each modality for multi-modal sentiment analysis. Specifically, we first align the raw text and audio at the word level and extract independent handcraft features for each modality to yield the textual and acoustic word sequence. Second, we establish two collaborative agents to deal with the textual and acoustic modalities in spoken language respectively. On this basis, we formulate the sentiment-relevant word selection process in a multi-modal setting as a multi-agent sequential decision problem and solve it with a multi-agent reinforcement learning approach. Detailed evaluations of multi-modal sentiment classification and emotion recognition on three benchmark datasets demonstrate the great effectiveness of our approach over several conventional competitive baselines.

BACKGROUND: Automatically understanding chemical-disease relations (CDRs) is crucial in various areas of biomedical research and health care. Supervised machine learning provides a feasible solution to automatically extract relations between biomedical entities from scientific literature, its success, however, heavily depends on large-scale biomedical corpora manually annotated with intensive labor and tremendous investment. RESULTS: We present an attention-based distant supervision paradigm for the BioCreative-V CDR extraction task. Training examples at both intra- and inter-sentence levels are generated automatically from the Comparative Toxicogenomics Database (CTD) without any human intervention. An attention-based neural network and a stacked auto-encoder network are applied respectively to induce learning models and extract relations at both levels. After merging the results of both levels, the document-level CDRs can be finally extracted. It achieves the precision/recall/F1-score of 60.3%/73.8%/66.4%, outperforming the state-of-the-art supervised learning systems without using any annotated corpus. CONCLUSION: Our experiments demonstrate that distant supervision is promising for extracting chemical disease relations from biomedical literature, and capturing both local and global attention features simultaneously is effective in attention-based distantly supervised learning.

Due to the ability of encoding and mapping semantic information into a highdimensional latent feature space, neural networks have been successfully used for detecting events to a certain extent.However, such a feature space can be easily contaminated by spurious features inherent in event detection.In this paper, we propose a self-regulated learning approach by utilizing a generative adversarial network to generate spurious features.On the basis, we employ a recurrent network to eliminate the fakes.Detailed experiments on the ACE 2005 and TAC-KBP 2015 corpora show that our proposed method is highly effective and adaptable.

Discourse parsing is a challenging task and plays a critical role in discourse analysis. This paper focus on the macro level discourse structure analysis, which has been less studied in the previous researches. We explore a macro discourse structure presentation schema to present the macro level discourse structure, and propose a corresponding corpus, named Macro Chinese Discourse Treebank. On these bases, we concentrate on two tasks of macro discourse structure analysis, including structure identification and nuclearity recognition. In order to reduce the error transmission between the associated tasks, we adopt a joint model of the two tasks, and an Integer Linear Programming approach is proposed to achieve global optimization with various kinds of constraints.

Event relation recognition is a challenging language processing task. It is required to determine the relation class of a pair of query events, such as causality, under the condition that there isn’t any reliable clue for use. We follow the traditional statistical approach in this paper, speculating the relation class of the target events based on the relation-class distributions on the similar events. There is minimal supervision used during the speculation process. In particular, we incorporate image processing into the acquisition of similar event instances, including the utilization of images for visually representing event scenes, and the use of the neural network based image matching for approximate calculation between events. We test our method on the ACE-R2 corpus and compared our model with the fully-supervised neural network models. Experimental results show that we achieve a comparable performance to CNN while slightly better than LSTM.

In view of the differences between the annotations of micro and macro discourse rela-tionships, this paper describes the relevant experiments on the construction of the Macro Chinese Discourse Treebank (MCDTB), a higher-level Chinese discourse corpus. Fol-lowing RST (Rhetorical Structure Theory), we annotate the macro discourse information, including discourse structure, nuclearity and relationship, and the additional discourse information, including topic sentences, lead and abstract, to make the macro discourse annotation more objective and accurate. Finally, we annotated 720 articles with a Kappa value greater than 0.6. Preliminary experiments on this corpus verify the computability of MCDTB.

Sentences in a well-formed text are connected to each other via various links to form the cohesive structure of the text. Current neural machine translation (NMT) systems translate a text in a conventional sentence-by-sentence fashion, ignoring such cross-sentence links and dependencies. This may lead to generate an incoherent target text for a coherent source text. In order to handle this issue, we propose a cache-based approach to modeling coherence for neural machine translation by capturing contextual information either from recently translated sentences or the entire document. Particularly, we explore two types of caches: a dynamic cache, which stores words from the best translation hypotheses of preceding sentences, and a topic cache, which maintains a set of target-side topical words that are semantically related to the document to be translated. On this basis, we build a new layer to score target words in these two caches with a cache-based neural model. Here the estimated probabilities from the cache-based neural model are combined with NMT probabilities into the final word prediction probabilities via a gating mechanism. Finally, the proposed cache-based neural model is trained jointly with NMT system in an end-to-end manner. Experiments and analysis presented in this paper demonstrate that the proposed cache-based model achieves substantial improvements over several state-of-the-art SMT and NMT baselines.

Stance detection aims to assign a stance label (for or against) to a post toward a specific target. Recently, there is a growing interest in using neural models to detect stance of documents. Most of these works model the sequence of words to learn document representation. However, much linguistic information, such as polarity and arguments of the document, is correlated with the stance of the document, and can inspire us to explore the stance. Hence, we present a neural model to fully employ various linguistic information to construct the document representation. In addition, since the influences of different linguistic information are different, we propose a hierarchical attention network to weigh the importance of various linguistic information, and learn the mutual attention between the document and the linguistic information. The experimental results on two datasets demonstrate the effectiveness of the proposed hierarchical attention neural model.

In realistic scenarios, a user profiling model (e.g., gender classification or age regression) learned from one social media might perform rather poorly when tested on another social media due to the different data distributions in the two media. In this paper, we address cross-media user profiling by bridging the knowledge between the source and target media with a uniform user embedding learning approach. In our approach, we first construct a cross-media user-word network to capture the relationship among users through the textual information and a modified cross-media user-user network to capture the relationship among users through the social information. Then, we learn user embedding by jointly learning the heterogeneous network composed of above two networks. Finally, we train a classification (or regression) model with the obtained user embeddings as input to perform user profiling. Empirical studies demonstrate the effectiveness of the proposed approach to two cross-media user profiling tasks, i.e., cross-media gender classification and cross-media age regression.

Question-Answer (QA) matching is a fundamental task in the Natural Language Processing community. In this paper, we first build a novel QA matching corpus with informal text which is collected from a product reviewing website. Then, we propose a novel QA matching approach, namely One vs. Many Matching, which aims to address the novel scenario where one question sentence often has an answer with multiple sentences. Furthermore, we improve our matching approach by employing both word-level and sentence-level attentions for solving the noisy problem in the informal text. Empirical studies demonstrate the effectiveness of the proposed approach to question-answer matching.

The task of nuclearity recognition in Chinese discourse remains challenging due to the demand for more deep semantic information. In this paper, we propose a novel text matching network (TMN) that encodes the discourse units and the paragraphs by combining Bi-LSTM and CNN to capture both global dependency information and local n-gram information. Moreover, it introduces three components of text matching, the Cosine, Bilinear and Single Layer Network, to incorporate various similarities and interactions among the discourse units. Experimental results on the Chinese Discourse TreeBank show that our proposed TMN model significantly outperforms various strong baselines in both micro-F1 and macro-F1.

Relation Classification aims to classify the semantic relationship between two marked entities in a given sentence. It plays a vital role in a variety of natural language processing applications. Most existing methods focus on exploiting mono-lingual data, e.g., in English, due to the lack of annotated data in other languages. In this paper, we come up with a feature adaptation approach for cross-lingual relation classification, which employs a generative adversarial network (GAN) to transfer feature representations from one language with rich annotated data to another language with scarce annotated data. Such a feature adaptation approach enables feature imitation via the competition between a relation classification network and a rival discriminator. Experimental results on the ACE 2005 multilingual training corpus, treating English as the source language and Chinese the target, demonstrate the effectiveness of our proposed approach, yielding an improvement of 5.7% over the state-of-the-art.

Chenlin Shen, Changlong Sun, Jingjing Wang, Yangyang Kang, Shoushan Li, Xiaozhong Liu, Luo Si, Min Zhang, Guodong Zhou. Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing. 2018.

We tackle discourse-level relation recognition, a problem of determining semantic relations between text spans.Implicit relation recognition is challenging due to the lack of explicit relational clues.The increasingly popular neural network techniques have been proven effective for semantic encoding, whereby widely employed to boost semantic relation discrimination.However, learning to predict semantic relations at a deep level heavily relies on a great deal of training data, but the scale of the publicly available data in this field is limited.In this paper, we follow Rutherford and Xue (2015) to expand the training data set using the corpus of explicitly-related arguments, by arbitrarily dropping the overtly presented discourse connectives.On the basis, we carry out an experiment of sampling, in which a simple active learning approach is used, so as to take the informative instances for data expansion.The goal is to verify whether the selective use of external data not only reduces the time consumption of retraining but also ensures a better system performance.Using the expanded training data, we retrain a convolutional neural network (CNN) based classifer which is a simplified version of Qin et al. (2016)'s stacking gated relation recognizer.Experimental results show that expanding the training set with small-scale carefully-selected external data yields substantial performance gain, with the improvements of about 4% for accuracy and 3.6% for F-score.This allows a weak classifier to achieve a comparable performance against the state-of-the-art systems.

Event factuality identification is an important semantic task in NLP. Traditional research heavily relies on annotated texts. This paper proposes a two-step framework, first extracting essential factors related with event factuality from raw texts as the input, and then identifying the factuality of events via a Generative Adversarial Network with Auxiliary Classification (AC-GAN). The use of AC-GAN allows the model to learn more syntactic information and address the imbalance among factuality values. Experimental results on FactBank show that our method significantly outperforms several state-of-the-art baselines, particularly on events with embedded sources, speculative and negative factuality values.

Aspect sentiment classification, a challenging task in sentiment analysis, has been attracting more and more attention in recent years. In this paper, we highlight the need for incorporating the importance degrees of both words and clauses inside a sentence and propose a hierarchical network with both word-level and clause-level attentions to aspect sentiment classification. Specifically, we first adopt sentence-level discourse segmentation to segment a sentence into several clauses. Then, we leverage multiple Bi-directional LSTM layers to encode all clauses and propose a word-level attention layer to capture the importance degrees of words in each clause. Third and finally, we leverage another Bi-directional LSTM layer to encode the outputs from the former layers and propose a clause-level attention layer to capture the importance degrees of all the clauses inside a sentence. Experimental results on the laptop and restaurant datasets from SemEval-2015 demonstrate the effectiveness of our proposed approach to aspect sentiment classification.

Previous studies normally formulate named entity recognition (NER) as a sequence labeling task and optimize the solution in the sentence level. In this article, we propose a document-level optimization approach to NER and apply it in a domain-specific document-level NER task. As a baseline, we apply a state-of-the-art approach, i.e., long-short-term memory (LSTM), to perform word classification. On this basis, we define a global objective function with the obtained word classification results and achieve global optimization via Integer Linear Programming (ILP). Specifically, in the ILP-based approach, we propose four kinds of constraints, i.e., label transition, entity length, label consistency, and domain-specific regulation constraints, to incorporate various entity recognition knowledge in the document level. Empirical studies demonstrate the effectiveness of the proposed approach to domain-specific document-level NER.

Even though a linguistics-free sequence to sequence model in neural machine translation (NMT) has certain capability of implicitly learning syntactic information of source sentences, this paper shows that source syntax can be explicitly incorporated into NMT effectively to provide further improvements.Specifically, we linearize parse trees of source sentences to obtain structural label sequences.On the basis, we propose three different sorts of encoders to incorporate source syntax into NMT: 1) Parallel RNN encoder that learns word and label annotation vectors parallelly; 2) Hierarchical RNN encoder that learns word and label annotation vectors in a two-level hierarchy; and 3) Mixed RNN encoder that stitchingly learns word and label annotation vectors over sequences where words and labels are mixed.Experimentation on Chinese-to-English translation demonstrates that all the three proposed syntactic encoders are able to improve translation accuracy.It is interesting to note that the simplest RNN encoder, i.e., Mixed RNN encoder yields the best performance with an significant improvement of 1.4 BLEU points.Moreover, an in-depth analysis from several perspectives is provided to reveal how source syntax benefits NMT.

User attribute classification aims to identify users’ attributes (e.g., gender, age and profession) by leveraging user generated content. However, conventional approaches to user attribute classification focus on single attribute classification involving only one user attribute, which completely ignores the relationship among various user attributes. In this paper, we confront a novel scenario in user attribute classification where relevant user attributes are jointly learned, attempting to make the relevant attribute classification tasks help each other. Specifically, we propose a joint learning approach, namely Aux-LSTM, which first learns a proper auxiliary representation between the related tasks and then leverages the auxiliary representation to integrate the learning process in both tasks. Empirical studies demonstrate the effectiveness of our proposed approach to joint learning on relevant user attributes.

A notably challenging problem in emotion analysis is recognizing the cause of an emotion. Although there have been a few studies on emotion cause detection, most of them work on news reports or a few of them focus on microblogs using a single-user structure (i.e., all texts in a microblog are written by the same user). In this article, we focus on emotion cause detection for Chinese microblogs using a multiple-user structure (i.e., texts in a microblog are successively written by several users). First, based on the fact that the causes of an emotion of a focused user may be provided by other users in a microblog with the multiple-user structure, we design an emotion cause annotation scheme which can deal with such a complicated case, and then provide an emotion cause corpus using the annotation scheme. Second, based on the analysis of the emotion cause corpus, we formalize two emotion cause detection tasks for microblogs (current-subtweet-based emotion cause detection and original-subtweet-based emotion cause detection). Furthermore, in order to examine the difficulty of the two emotion cause detection tasks and the contributions of texts written by different users in a microblog with the multiple-user structure, we choose two popular classification methods (SVM and LSTM) to do emotion cause detection. Our experiments show that the current-subtweet-based emotion cause detection is much more difficult than the original-subtweet-based emotion cause detection, and texts written by different users are very helpful for both emotion cause detection tasks. This study presents a pilot study of emotion cause detection which deals with Chinese microblogs using a complicated structure.

Discourse parsing is a challenging task and plays a critical role in discourse analysis. Since the release of the Rhetorical Structure Theory Discourse Treebank and the Penn Discourse Treebank, the research on English discourse parsing has attracted increasing attention and achieved considerable success in recent years. At the same time, some preliminary research on certain subtasks about discourse parsing for other languages, such as Chinese, has been conducted. In this article, we present an end-to-end Chinese discourse parser with the Connective-Driven Dependency Tree scheme, which consists of multiple components in a pipeline architecture, such as the elementary discourse unit (EDU) detector, discourse relation recognizer, discourse parse tree generator, and attribution labeler. In particular, the attribution labeler determines two attributions (i.e., sense and centering) for every nonterminal node (i.e., discourse relation) in the discourse parse trees. Systematically, our parser detects all EDUs in a free text, generates the discourse parse tree in a bottom-up way, and determines the sense and centering attributions for all nonterminal nodes by traversing the discourse parse tree. Comprehensive evaluation on the Connective-Driven Dependency Treebank corpus from both component-wise and error-cascading perspectives is conducted to illustrate how each component performs in isolation, and how the pipeline performs with error propagation. Finally, it shows that our end-to-end Chinese discourse parser achieves an overall F1 score of 20% with full automation.

Previous research on emotions analysis has placed much emphasis in monolingual instead of bilingual text. However, emotions on social media platforms are often found in bilingual or code-switching posts. Different from monolingual text, emotions in code-switching text can be expressed in both monolingual and bilingual forms. Moreover, more than one emotion can be expressed within a single post; yet they tend to be related in some ways which offers some implications. It is thus necessary to consider the correlation between different emotions. In this paper, a joint factor graph model is proposed to address this issue. In particular, attribute functions of the factor graph model are utilized to learn both monolingual and bilingual information from each post, factor functions are used to explore the relationship among different emotions, and a belief propagation algorithm is employed to learn and predict the model. Empirical studies demonstrate the importance of emotion analysis in code-switching text and the effectiveness of our proposed joint learning model.

In gender classification, labeled data is often limited while unlabeled data is ample. This motivates semi-supervised learning for gender classification to improve the performance by exploring the knowledge in both labeled and unlabeled data. In this paper, we propose a semi-supervised approach to gender classification by leveraging textual features and a specific kind of indirect links among the users which we call “same-interest” links. Specifically, we propose a factor graph, namely Textual and Social Factor Graph (TSFG), to model both the textual and the “same-interest” link information. Empirical studies demonstrate the effectiveness of the proposed approach to semi-supervised gender classification.

In the literature, various supervised learning approaches have been adopted to address the task of reader emotion classification. However, the classification performance greatly suffers when the size of the labeled data is limited. In this paper, we propose a two-view label propagation approach to semi-supervised reader emotion classification by exploiting two views, namely source text and response text in a label propagation algorithm. Specifically, our approach depends on two word-document bipartite graphs to model the relationship among the samples in the two views respectively. Besides, the two bipartite graphs are integrated by linking each source text sample with its corresponding response text sample via a length-sensitive transition probability. In this way, our two-view label propagation approach to semi-supervised reader emotion classification largely alleviates the reliance on the strong sufficiency and independence assumptions of the two views, as required in co-training. Empirical evaluation demonstrates the effectiveness of our two-view label propagation approach to semi-supervised reader emotion classification.

Previous studies on temporal relation extraction focus on mining sentence-level information or enforcing coherence on different temporal relation types among various event mentions in the same sentence or neighboring sentences, largely ignoring those discourse-level temporal relations in nonadjacent sentences. In this paper, we propose a discourse-level global inference model to mine those temporal relations between event mentions in document-level, especially in nonadjacent sentences. Moreover, we provide various kinds of discourse-level constraints, which derived from event semantics, to further improve our global inference model. Evaluation on a Chinese corpus justifies the effectiveness of our discourse-level global inference model over two strong baselines.

Emotions in code-switching text can be expressed in either monolingual or bilingual forms. However, relatively little research has emphasized on code-switching text. In this paper, we propose a Bilingual Attention Network (BAN) model to aggregate the monolingual and bilingual informative words to form vectors from the document representation, and integrate the attention vectors to predict the emotion. The experiments show that the effectiveness of the proposed model. Visualization of the attention layers illustrates that the model selects qualitatively informative words.

Textual information is of critical importance for automatic user classification in social media. However, most previous studies model textual features in a single perspective while the text in a user homepage typically possesses different styles of text, such as original message and comment from others. In this paper, we propose a novel approach, namely ensemble LSTM, to user classification by incorporating multiple textual perspectives. Specifically, our approach first learns a LSTM representation with a LSTM recurrent neural network and then presents a joint learning method to integrating all naturally-divided textual perspectives. Empirical studies on two basic user classification tasks, i.e., gender classification and age classification, demonstrate the effectiveness of the proposed approach to user classification with multiple textual perspectives.

Machine learning-based methods have obtained great progress on emotion classification. However, in most previous studies, the models are learned based on a single corpus which often suffers from insufficient labeled data. In this paper, we propose a corpus fusion approach to address emotion classification across two corpora which use different emotion taxonomies. The objective of this approach is to utilize the annotated data from one corpus to help the emotion classification on another corpus. An Integer Linear Programming (ILP) optimization is proposed to refine the classification results. Empirical studies show the effectiveness of the proposed approach to corpus fusion for emotion classification.

Semantic parsing maps a sentence in natural language into a structured meaning representation. Previous studies show that semantic parsing with synchronous context-free grammars (SCFGs) achieves favorable performance over most other alternatives. Motivated by the observation that the performance of semantic parsing with SCFGs is closely tied to the translation rules, this article explores to extend translation rules with high quality and increased coverage in three ways. First, we examine the difference between word alignments for semantic parsing and statistical machine translation (SMT) to better adapt word alignment in SMT to semantic parsing. Second, we introduce both structure and syntax informed nonterminals, better guiding the parsing in favor of well-formed structure, instead of using a uninformed nonterminal in SCFGs. Third, we address the unknown word translation issue via synthetic translation rules. Last but not least, we use a filtering approach to improve performance via predicting answer type. Evaluation on the standard GeoQuery benchmark dataset shows that our approach greatly outperforms the state of the art across various languages, including English, Chinese, Thai, German, and Greek.

Although several semi-supervised learning models have been proposed for English event extraction, there are few successful stories in Chinese due to its special characteristics. In this article, we propose a novel minimally supervised model for Chinese event extraction from multiple views. Besides the traditional pattern similarity view (PSV), a semantic relationship view (SRV) is introduced to capture the relevant event mentions from relevant documents. Moreover, a morphological structure view (MSV) is incorporated to both infer more positive patterns and help filter negative patterns via morphological structure similarity. An evaluation of the ACE 2005 Chinese corpus shows that our minimally supervised model significantly outperforms several strong baselines.

Event extraction in Chinese suffers greatly from the frequent missing of arguments. Statistical analysis of the automatic content extraction (ACE) 2005 Chinese corpus shows that nearly 55% of arguments do not occur in their corresponding event mentions. This problem greatly hinders the wide deployment of event extraction in real applications. This paper proposes a novel joint argument inference model to recover those missing arguments in event mentions from a semantic perspective. In particular, this model employs various types of argument consistencies to reveal intra-event argument semantics in multiple dimensions, such as argument-argument, argument-role and argument-trigger. Moreover, this model explores several linguistic-driven event relevance phenomena, such as Coreference, Sequence, and Parallel, to unveil inter-event argument semantics in various layers, such as sentence, discourse, and document. An evaluation of the ACE 2005 Chinese corpus justifies the effectiveness of our joint argument inference model compared to a state-of-the-art baseline.

Shoushan Li, Lei Huang, Rong Wang, Guodong Zhou. Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2015.

Bowei Zou, Qiaoming Zhu, Guodong Zhou. Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2015.

Semantic parsing maps a sentence in natural language into a structured meaning representation.Previous studies show that semantic parsing with synchronous contextfree grammars (SCFGs) achieves favorable performance over most other alternatives.Motivated by the observation that the performance of semantic parsing with SCFGs is closely tied to the translation rules, this paper explores extending translation rules with high quality and increased coverage in three ways.First, we introduce structure informed non-terminals, better guiding the parsing in favor of well formed structure, instead of using a uninformed non-terminal in SCFGs.Second, we examine the difference between word alignments for semantic parsing and statistical machine translation (SMT) to better adapt word alignment in SMT to semantic parsing.Finally, we address the unknown word translation issue via synthetic translation rules.Evaluation on the standard GeoQuery benchmark dataset shows that our approach achieves the state-of-the-art across various languages, including English, German and Greek.

Due to the commonality in natural language, negation focus plays a critical role in deep understanding of context.However, existing studies for negation focus identification major on supervised learning which is timeconsuming and expensive due to manual preparation of annotated corpus.To address this problem, we propose an unsupervised word-topic graph model to represent and measure the focus candidates from both lexical and topic perspectives.Moreover, we propose a document-sensitive biased Pag-eRank algorithm to optimize the ranking scores of focus candidates.Evaluation on the *SEM 2012 shared task corpus shows that our proposed method outperforms the state of the art on negation focus identification.

Document-level Machine Translation (MT) has been drawing more and more attention due to its potential of resolving sentence-level ambiguities and inconsistencies with the benefit of wide-range context. However, the lack of simple yet effective evaluation metrics largely impedes the development of such document-level MT systems. This paper proposes to improve traditional MT evaluation metrics by simplified lexical chain, modeling document-level phenomena from the perspectives of text cohesion. Experiments show the effectiveness of such method on evaluating document-level translation quality and its potential of integrating with traditional MT evaluation metrics to achieve higher correlation with human judgments.

Active learning (AL) has been proven effective to reduce human annotation efforts in NLP.However, previous studies on AL are limited to applications in a single language.This paper proposes a bilingual active learning paradigm for relation classification, where the unlabeled instances are first jointly chosen in terms of their prediction uncertainty scores in two languages and then manually labeled by an oracle.Instead of using a parallel corpus, labeled and unlabeled instances in one language are translated into ones in the other language and all instances in both languages are then fed into a bilingual active learning engine as pseudo parallel corpora.Experimental results on the ACE RDC 2005 Chinese and English corpora show that bilingual active learning for relation classification significantly outperforms monolingual active learning.

Negative expressions are common in natural language text and play a critical role in information extraction. However, the performances of current systems are far from satisfaction, largely due to its focus on intrasentence information and its failure to consider inter-sentence information. In this paper, we propose a graph model to enrich intrasentence features with inter-sentence features from both lexical and topic perspectives. Evaluation on the *SEM 2012 shared task corpus indicates the usefulness of contextual discourse information in negation focus identification and justifies the effectiveness of our graph model in capturing such global information.

Discourse parsing is a challenging task and plays a critical role in discourse analysis.In this paper, we focus on labeling full argument spans of discourse connectives in the Penn Discourse Treebank (PDTB).Previous studies cast this task as a linear tagging or subtree extraction problem.In this paper, we propose a novel constituent-based approach to argument labeling, which integrates the advantages of both linear tagging and subtree extraction.In particular, the proposed approach unifies intra-and intersentence cases by treating the immediately preceding sentence as a special constituent.Besides, a joint inference mechanism is introduced to incorporate global information across arguments into our constituent-based approach via integer linear programming.Evaluation on PDT-B shows significant performance improvements of our constituent-based approach over the best state-of-the-art system.It also shows the effectiveness of our joint inference mechanism in modeling global information across arguments.

In this paper, we propose a Connectivedriven Dependency Tree (CDT) scheme to represent the discourse rhetorical structure in Chinese language, with elementary discourse units as leaf nodes and connectives as non-leaf nodes, largely motivated by the Penn Discourse Treebank and the Rhetorical Structure Theory.In particular, connectives are employed to directly represent the hierarchy of the tree structure and the rhetorical relation of a discourse, while the nuclei of discourse units are globally determined with reference to the dependency theory.Guided by the CDT scheme, we manually annotate a Chinese Discourse Treebank (CDTB) of 500 documents.Preliminary evaluation justifies the appropriateness of the CDT scheme to Chinese discourse analysis and the usefulness of our manually annotated CDTB corpus.

Sentiment and emotion classification have been popularly but separately studied in natural language processing. In this paper, we address joint learning on sentiment and emotion classification where both the labeled data for sentiment and emotion classification are available. The objective of this joint-learning is to benefit the two tasks from each other for improving their performances. Specifically, an extra data set that is annotated with both sentiment and emotion labels are employed to estimate the transformation probability between the two kinds of labels. Furthermore, the transformation probability is leveraged to transfer the classification labels to benefit the two tasks from each other. Empirical studies demonstrate the effectiveness of our approach for the novel joint learning task.

Personal profile information on social media like LinkedIn.comand Facebook.com is at the core of many interesting applications, such as talent recommendation and contextual advertising.However, personal profiles usually lack organization confronted with the large amount of available information.Therefore, it is always a challenge for people to find desired information from them.In this paper, we address the task of personal profile summarization by leveraging both personal profile textual information and social networks.Here, using social networks is motivated by the intuition that, people with similar academic, business or social connections (e.g.co-major, co-university, and cocorporation) tend to have similar experience and summaries.To achieve the learning process, we propose a collective factor graph (CoFG) model to incorporate all these resources of knowledge to summarize personal profiles with local textual attribute functions and social connection factors.Extensive evaluation on a large-scale dataset from LinkedIn.comdemonstrates the effectiveness of the proposed approach.

Scope detection is a key task in information extraction.This paper proposes a new approach for tree kernel-based scope detection by using the structured syntactic parse information.In addition, we have explored the way of selecting compatible features for different part-of-speech cues.Experiments on the BioScope corpus show that both constituent and dependency structured syntactic parse features have the advantage in capturing the potential relationships between cues and their scopes.Compared with the state of the art scope detection systems, our system achieves substantial improvement.

In this paper, we present a simplified shallow semantic parsing approach to extracting opinion targets. This is done by formulating opinion target extraction (OTE) as a shallow semantic parsing problem with the opinion expression as the predicate and the corresponding targets as its arguments. In principle, our parsing approach to OTE differs from the state-of-the-art sequence labeling one in two aspects. First, we model OTE from parse tree level, where abundant structured syntactic information is available for use, instead of word sequence level, where only lexical information is available. Second, we focus on determining whether a constituent, rather than a word, is an opinion target or not, via a simplified shallow semantic parsing framework. Evaluation on two datasets shows that structured syntactic information plays a critical role in capturing the domination relationship between an opinion expression and its targets. It also shows that our parsing approach much outperforms the state-of-the-art sequence labeling one.

Motivated by the critical importance of connectives in recognizing discourse relations, we present an unsupervised cross-argument inference mechanism to implicit discourse relation recognition. The basic idea is to infer the implicit discourse relation of an argument pair from a large number of comparable argument pairs, which are automatically retrieved from the web in an unsupervised way. In this way, the inference proceeds from explicit relations to implicit ones via connective as bridge. This kind of pair-to-pair inference is based on the assumption that two argument pairs with high content similarity (i.e. comparable argument pairs) should have similar discourse relationship. Evaluation on PDTB proves the effectiveness of our inference mechanism in implicit relation recognition to the four level-1 relations. It also shows that our mechanism significantly outperforms other alternatives.

Semi-supervised sentiment classification aims to train a classifier with a small number of labeled data (called seed data) and a large amount of unlabeled data. a big advantage of this approach is its saving of annotation effort by using the unlabeled data which is usually freely available. In this paper, we propose an approach to further minimize the annotation effort of semi-supervised sentiment classification by actively selecting the seed data. Specifically, a novel selection strategy is proposed to simultaneously select good words and documents for manual annotation by considering both of their annotation costs and informativeness. Experimental results demonstrate the effectiveness of our approach.

This paper focuses on exploring the features of product reviews that satisfy users, by which to improve the automatic helpfulness voting for the reviews on commercial websites. Compared to the previous work, which single-mindedly adopts the textual features to assess the review helpfulness, we propose that user preferences are more explicit clues to infer the opinions of users on the review helpfulness. By using the user-preference based features, we firstly implement a binary helpfulness based review classification system to divide helpful reviews and useless, and on the basis, we secondly build a Ranking SVM based automatic helpfulness voting system (AHV) which rank reviews based on their helpfulness. Experiments used a large scale dataset containing over 34,266 reviews on 1289 products to test the systems, which achieves promising performances with accuracy of up to 0.72 and [email protected] of 0.25, and at least 9% accuracy improvement compared to the textual-feature based helpfulness assessment.

This article presents a unified framework for extracting standard and update summaries from a set of documents. In particular, a topic modeling approach is employed for salience determination and a dynamic modeling approach is proposed for redundancy control. In the topic modeling approach for salience determination, we represent various kinds of text units, such as word, sentence, document, documents, and summary, using a single vector space model via their corresponding probability distributions over the inherent topics of given documents or a related corpus. Therefore, we are able to calculate the similarity between any two text units via their topic probability distributions. In the dynamic modeling approach for redundancy control, we consider the similarity between the summary and the given documents, and the similarity between the sentence and the summary, besides the similarity between the sentence and the given documents, for standard summarization while for update summarization, we also consider the similarity between the sentence and the history documents or summary. Evaluation on TAC 2008 and 2009 in English language shows encouraging results, especially the dynamic modeling approach in removing the redundancy in the given documents. Finally, we extend the framework to Chinese multi-document summarization and experiments show the effectiveness of our framework.

Sentiment classification has undergone significant development in recent years. However, most existing studies assume the balance between negative and positive samples, which may not be true in reality. In this paper, we investigate imbalanced sentiment classification instead. In particular, a novel clustering-based stratified under-sampling framework and a centroid-directed smoothing strategy are proposed to address the imbalanced class and feature distribution problems respectively. Evaluation across different datasets shows the effectiveness of both the under-sampling framework and the smoothing strategy in handling the imbalanced problems in real sentiment classification applications.

Event anaphora resolution plays a critical role in discourse analysis. This paper proposes a tree kernel-based framework for event pronoun resolution. In particular, a new tree expansion scheme is introduced to automatically determine a proper parse tree structure for event pronoun resolution by considering various kinds of competitive information related with the anaphor and the antecedent candidate. Evaluation on the OntoNotes English corpus shows the appropriateness of the tree kernel-based framework and the effectiveness of competitive information for event pronoun resolution.

This article explores unified semantic role labeling (SRL) for both verbal and nominal predicates in the Chinese language. This is done by considering SRL for both verbal and nominal predicates in a unified framework. First, we systematically examine various kinds of features for verbal SRL and nominal SRL, respectively, besides those widely used ones. Then we further improve the performance of nominal SRL with various kinds of verbal evidence, that is, merging the training instances from verbal predicates and integrating various kinds of features derived from SRL for verbal predicates. Finally, we address the issue of automatic predicate recognition, which is essential for nominal SRL. Evaluation on Chinese PropBank and Chinese NomBank shows that our unified approach significantly improves the performance, in particular that of nominal SRL. To the best of our knowledge, this is the first reported work of unified verbal and nominal SRL on Chinese PropBank and NomBank.

This article proposes a new approach to dynamically determine the tree span for tree kernel-based semantic relation extraction between named entities. The basic idea is to employ constituent dependency information in keeping the necessary nodes and their head children along the path connecting the two entities in the syntactic parse tree, while removing the noisy information from the tree, eventually leading to a dynamic syntactic parse tree. This article also explores various entity features and their possible combinations via a unified syntactic and semantic tree framework, which integrates both structural syntactic parse information and entity-related semantic information. Evaluation on the ACE RDC 2004 English and 2005 Chinese benchmark corpora shows that our dynamic syntactic parse tree much outperforms all previous tree spans, indicating its effectiveness in well representing the structural nature of relation instances while removing redundant information. Moreover, the unified parse and semantic tree significantly outperforms the single syntactic parse tree, largely due to the remarkable contributions from entity-related semantic features such as its type, subtype, mention-level as well as their bi-gram combinations. Finally, the best performance so far in semantic relation extraction is achieved via a composite kernel, which combines this tree kernel with a linear, state-of-the-art, feature-based kernel.

Convolution tree kernel has shown promising results in semantic role labeling (SRL). However, this kernel does not consider much linguistic knowledge in kernel design and only performs hard matching between subtrees. To overcome these constraints, this paper proposes a grammar-driven convolution tree kernel for SRL by introducing more linguistic knowledge. Compared with the standard convolution tree kernel, the proposed grammar-driven kernel has two advantages: 1) grammar-driven approximate substructure matching, and 2) grammar-driven approximate tree node matching. The two approximate matching mechanisms enable the proposed kernel to better explore linguistically motivated structured knowledge. Experiments on the CoNLL-2005 SRL shared task and the PropBank I corpus show that the proposed kernel outperforms the standard convolution tree kernel significantly. Moreover, we present a composite kernel to integrate a feature-based polynomial kernel and the proposed grammar-driven convolution tree kernel for SRL. Experimental results show that our composite kernel-based method significantly outperforms the previously best-reported ones.

This paper proposes a novel composite kernel for relation extraction. The composite kernel consists of two individual kernels: an entity kernel that allows for entity-related features and a convolution parse tree kernel that models syntactic information of relation examples. The motivation of our method is to fully utilize the nice properties of kernel methods to explore diverse knowledge for relation extraction. Our study illustrates that the composite kernel can effectively capture both flat and structured features without the need for extensive feature engineering, and can also easily scale to include more features. Evaluation on the ACE corpus shows that our method outperforms the previous best-reported methods and significantly out-performs previous two dependency tree kernels for relation extraction.

This paper proposes a novel hierarchical learning strategy to deal with the data sparseness problem in relation extraction by modeling the commonality among related classes. For each class in the hierarchy either manually predefined or automatically clustered, a linear discriminative function is determined in a top-down way using a perceptron algorithm with the lower-level weight vector derived from the upper-level weight vector. As the upper-level class normally has much more positive training examples than the lower-level class, the corresponding linear discriminative function can be determined more reliably. The upper-level discriminative function then can effectively guide the discriminative function learning in the lower-level, which otherwise might suffer from limited training data. Evaluation on the ACE RDC 2003 corpus shows that the hierarchical strategy much improves the performance by 5.6 and 5.1 in F-measure on least- and medium- frequent relations respectively. It also shows that our system outperforms the previous best-reported system by 2.7 in F-measure on the 24 subtypes using the same feature set.

This paper proposes a novel document re-ranking approach in information retrieval, which is done by a label propagation-based semi-supervised learning algorithm to utilize the intrinsic structure underlying in the large document data. Since no labeled relevant or irrelevant documents are generally available in IR, our approach tries to extract some pseudo labeled documents from the ranking list of the initial retrieval. For pseudo relevant documents, we determine a cluster of documents from the top ones via cluster validation-based k-means clustering; for pseudo irrelevant ones, we pick a set of documents from the bottom ones. Then the ranking of the documents can be conducted via label propagation. Evaluation on benchmark corpora shows that the approach can achieve significant improvement over standard baselines and performs better than other related approaches.

Extracting semantic relationships between entities is challenging. This paper investigates the incorporation of diverse lexical, syntactic and semantic knowledge in feature-based relation extraction using SVM. Our study illustrates that the base phrase chunking information is very effective for relation extraction and contributes to most of the performance improvement from syntactic aspect while additional information from full parsing gives limited further enhancement. This suggests that most of useful information in full parse trees for relation extraction is shallow and can be captured by chunking. We also demonstrate how semantic information such as WordNet and Name List, can be used in feature-based relation extraction to further improve the performance. Evaluation on the ACE corpus shows that effective incorporation of diverse features enables our system outperform previously best-reported systems on the 24 ACE relation subtypes and significantly outperforms tree kernel-based systems by over 20 in F-measure on the 5 ACE relation types.

This paper proposes an ensemble of classifiers for biomedical name recognition in which three classifiers, one Support Vector Machine and two discriminative Hidden Markov Models, are combined effectively using a simple majority voting strategy. In addition, we incorporate three post-processing modules, including an abbreviation resolution module, a protein/gene name refinement module and a simple dictionary matching module, into the system to further improve the performance. Evaluation shows that our system achieves the best performance from among 10 systems with a balanced F-measure of 82.58 on the closed evaluation of the BioCreative protein/gene name recognition task (Task 1A).

Named entity recognition identifies and classifies entity names in a text document into some predefined categories. It resolves the “who”, “where” and “how much” problems in information extraction and leads to the resolution of the “what” and “how” problems in further processing. This paper presents a Hidden Markov Model (HMM) and proposes a HMM-based named entity recognizer implemented as the system PowerNE. Through the HMM and an effective constraint relaxation algorithm to deal with the data sparseness problem, PowerNE is able to effectively apply and integrate various internal and external evidences of entity names. Currently, four evidences are included: (1) a simple deterministic internal feature of the words, such as capitalization and digitalization; (2) an internal semantic feature of the important triggers; (3) an internal gazetteer feature, which determines the appearance of the current word string in the provided gazetteer list; and (4) an external macro context feature, which deals with the name alias phenomena. In this way, the named entity recognition problem is resolved effectively. PowerNE has been benchmarked with the Message Understanding Conferences (MUC) data. The evaluation shows that, using the formal training and test data of the MUC-6 and MUC-7 English named entity tasks, and it achieves the F-measures of 96.6 and 94.1, respectively. Compared with the best reported machine learning system, it achieves a 1.7 higher F-measure with one quarter of the training data on MUC-6, and a 3.6 higher F-measure with one ninth of the training data on MUC-7. In addition, it performs slightly better than the best reported handcrafted rule-based systems on MUC-6 and MUC-7.

In this paper, we propose a multi-criteriabased active learning approach and effectively apply it to named entity recognition. Active learning targets to minimize the human annotation efforts by selecting examples for labeling. To maximize the contribution of the selected examples, we consider the multiple criteria: informativeness, representativeness and diversity and propose measures to quantify them. More comprehensively, we incorporate all the criteria using two selection strategies, both of which result in less labeling cost than single-criterion-based method. The results of the named entity recognition in both MUC-6 and GENIA show that the labeling cost can be reduced by at least 80 % without degrading the performance. 1

Coreferential information of a candidate, such as the properties of its antecedents, is important for pronoun resolution because it reflects the salience of the candidate in the local discourse. Such information, however, is usually ignored in previous learning-based systems. In this paper we present a trainable model which incorporates coreferential information of candidates into pronoun resolution. Preliminary experiments show that our model will boost the resolution performance given the right antecedents of the candidates. We further discuss how to apply our model in real resolution where the antecedents of the candidate are found by a separate noun phrase resolution module. The experimental results show that our model still achieves better performance than the baseline.

MOTIVATION: With an overwhelming amount of textual information in molecular biology and biomedicine, there is a need for effective and efficient literature mining and knowledge discovery that can help biologists to gather and make use of the knowledge encoded in text documents. In order to make organized and structured information available, automatically recognizing biomedical entity names becomes critical and is important for information retrieval, information extraction and automated knowledge acquisition. RESULTS: In this paper, we present a named entity recognition system in the biomedical domain, called PowerBioNE. In order to deal with the special phenomena of naming conventions in the biomedical domain, we propose various evidential features: (1) word formation pattern; (2) morphological pattern, such as prefix and suffix; (3) part-of-speech; (4) head noun trigger; (5) special verb trigger and (6) name alias feature. All the features are integrated effectively and efficiently through a hidden Markov model (HMM) and a HMM-based named entity recognizer. In addition, a k-Nearest Neighbor (k-NN) algorithm is proposed to resolve the data sparseness problem in our system. Finally, we present a pattern-based post-processing to automatically extract rules from the training data to deal with the cascaded entity name phenomenon. From our best knowledge, PowerBioNE is the first system which deals with the cascaded entity name phenomenon. Evaluation shows that our system achieves the F-measure of 66.6 and 62.2 on the 23 classes of GENIA V3.0 and V1.1, respectively. In particular, our system achieves the F-measure of 75.8 on the "protein" class of GENIA V3.0. For comparison, our system outperforms the best published result by 7.8 on GENIA V1.1, without help of any dictionaries. It also shows that our HMM and the k-NN algorithm outperform other models, such as back-off HMM, linear interpolated HMM, support vector machines, C4.5, C4.5 rules and RIPPER, by effectively capturing the local context dependency and resolving the data sparseness problem. Moreover, evaluation on GENIA V3.0 shows that the post-processing for the cascaded entity name phenomenon improves the F-measure by 3.9. Finally, error analysis shows that about half of the errors are caused by the strict annotation scheme and the annotation inconsistency in the GENIA corpus. This suggests that our system achieves an acceptable F-measure of 83.6 on the 23 classes of GENIA V3.0 and in particular 86.2 on the "protein" class, without help of any dictionaries. We think that a F-measure of 90 on the 23 classes of GENIA V3.0 and in particular 92 on the "protein" class, can be achieved through refining of the annotation scheme in the GENIA corpus, such as flexible annotation scheme and annotation consistency, and inclusion of a reasonable biomedical dictionary. AVAILABILITY: A demo system is available at http://textmining.i2r.a-star.edu.sg/NLS/demo.htm. Technology license is available upon the bilateral agreement.

In this paper we propose a competition learning approach to coreference resolution. Traditionally, supervised machine learning approaches adopt the single-candidate model. Nevertheless the preference relationship between the antecedent candidates cannot be determined accurately in this model. By contrast, our approach adopts a twin-candidate learning model. Such a model can present the competition criterion for antecedent candidates reliably, and ensure that the most preferred candidate is selected. Furthermore, our approach applies a candidate filter to reduce the computational cost and data noises during training and resolution. The experimental results on MUC-6 and MUC-7 data set show that our approach can outperform those based on the single-candidate model.

This paper proposes a Hidden Markov Model (HMM) and an HMM-based chunk tagger, from which a named entity (NE) recognition (NER) system is built to recognize and classify names, times and numerical quantities. Through the HMM, our system is able to apply and integrate four types of internal and external evidences: 1) simple deterministic internal feature of the words, such as capitalization and digitalization; 2) internal semantic feature of important triggers; 3) internal gazetteer feature; 4) external macro context feature. In this way, the NER problem can be resolved effectively. Evaluation of our system on MUC-6 and MUC-7 English NE tasks achieves F-measures of 96.6% and 94.1% respectively. It shows that the performance is significantly better than reported by any other machine-learning system. Moreover, the performance is even consistently better than those based on handcrafted rules.

This paper proposes a new error-driven HMM-based text chunk tagger with context-dependent lexicon. Compared with standard HMM-based tagger, this tagger uses a new Hidden Markov Modelling approach which incorporates more contextual information into a lexical entry. Moreover, an error-driven learning approach is adopted to decrease the memory requirement by keeping only positive lexical entries and makes it possible to further incorporate more context-dependent lexical entries. Experiments show that this technique achieves overall precision and recall rates of 93.40% and 93.95% for all chunk types, 93.60% and 94.64% for noun phrases, and 94.64% and 94.75% for verb phrases when trained on PENN WSJ TreeBank section 00-19 and tested on section 20-24, while 25-fold validation experiments of PENN WSJ TreeBank show overall precision and recall rates of 96.40% and 96.47% for all chunk types, 96.49% and 96.99% for noun phrases, and 97.13% and 97.36% for verb phrases.