Minlie Huang

Xiao Liu, Xuanyu Lei, Shengyuan Wang et al. · tsinghua

Alignment has become a critical step for instruction-tuned Large Language Models (LLMs) to become helpful assistants. However, the effective evaluation of alignment for emerging Chinese LLMs is still largely unexplored. To fill in this gap, we introduce AlignBench, a comprehensive multi-dimensional benchmark for evaluating LLMs' alignment in Chinese. We design a human-in-the-loop data curation pipeline, containing eight main categories, 683 real-scenario rooted queries and corresponding human verified references. To ensure the correctness of references, each knowledge-intensive query is accompanied with evidences collected from reliable web sources (including URLs and quotations) by our annotators. For automatic evaluation, our benchmark employs a rule-calibrated multi-dimensional LLM-as-Judge~\cite{zheng2023judging} approach with Chain-of-Thought to generate explanations and final ratings, ensuring high reliability and interpretability. All evaluation code, data, and LLM generations are available at \url{https://github.com/THUDM/AlignBench}. Since its release, AlignBench has been adopted by top (Chinese) LLMs for evaluating their alignment capabilities in Chinese, including ChatGLM, Qwen, DeepSeek, Yi, Baichuan, and Abab.

Jian Guan, Jesse Dodge, David Wadden et al. · allen-ai, cmu

Recent progress in natural language processing (NLP) owes much to remarkable advances in large language models (LLMs). Nevertheless, LLMs frequently "hallucinate," resulting in non-factual outputs. Our carefully-designed human evaluation substantiates the serious hallucination issue, revealing that even GPT-3.5 produces factual outputs less than 25% of the time. This underscores the importance of fact verifiers in order to measure and incentivize progress. Our systematic investigation affirms that LLMs can be repurposed as effective fact verifiers with strong correlations with human judgments. Surprisingly, FLAN-T5-11B, the least factual generator in our study, performs the best as a fact verifier, even outperforming more capable LLMs like GPT3.5 and ChatGPT. Delving deeper, we analyze the reliance of these LLMs on high-quality evidence, as well as their deficiencies in robustness and generalization ability. Our study presents insights for developing trustworthy generation models.

Jiayi Gui, Yiming Liu, Jiale Cheng et al.

Large Language Models (LLMs) have demonstrated notable capabilities across various tasks, showcasing complex problem-solving abilities. Understanding and executing complex rules, along with multi-step planning, are fundamental to logical reasoning and critical for practical LLM agents and decision-making systems. However, evaluating LLMs as effective rule-based executors and planners remains underexplored. In this paper, we introduce LogicGame, a novel benchmark designed to evaluate the comprehensive rule understanding, execution, and planning capabilities of LLMs. Unlike traditional benchmarks, LogicGame provides diverse games that contain a series of rules with an initial state, requiring models to comprehend and apply predefined regulations to solve problems. We create simulated scenarios in which models execute or plan operations to achieve specific outcomes. These game scenarios are specifically designed to distinguish logical reasoning from mere knowledge by relying exclusively on predefined rules. This separation allows for a pure assessment of rule-based reasoning capabilities. The evaluation considers not only final outcomes but also intermediate steps, providing a comprehensive assessment of model performance. Moreover, these intermediate steps are deterministic and can be automatically verified. LogicGame defines game scenarios with varying difficulty levels, from simple rule applications to complex reasoning chains, in order to offer a precise evaluation of model performance on rule understanding and multi-step execution. Utilizing LogicGame, we test various LLMs and identify notable shortcomings in their rule-based logical reasoning abilities.

Zhuocheng Gong, Ang Lv, Jian Guan et al.

Is it always necessary to compute tokens from shallow to deep layers in Transformers? The continued success of vanilla Transformers and their variants suggests an undoubted "yes". In this work, however, we attempt to break the depth-ordered convention by proposing a novel architecture dubbed mixture-of-modules (MoM), which is motivated by an intuition that any layer, regardless of its position, can be used to compute a token as long as it possesses the needed processing capabilities. The construction of MoM starts from a finite set of modules defined by multi-head attention and feed-forward networks, each distinguished by its unique parameterization. Two routers then iteratively select attention modules and feed-forward modules from the set to process a token. The selection dynamically expands the computation graph in the forward pass of the token, culminating in an assembly of modules. We show that MoM provides not only a unified framework for Transformers and their numerous variants but also a flexible and learnable approach for reducing redundancy in Transformer parameterization. We pre-train various MoMs using OpenWebText. Empirical results demonstrate that MoMs, of different parameter counts, consistently outperform vanilla transformers on both GLUE and XSUM benchmarks. More interestingly, with a fixed parameter budget, MoM-large enables an over 38% increase in depth for computation graphs compared to GPT-2-large, resulting in absolute gains of 1.4 on GLUE and 1 on XSUM. On the other hand, MoM-large also enables an over 60% reduction in depth while involving more modules per layer, yielding a 16% reduction in TFLOPs and a 43% decrease in memory usage compared to GPT-2-large, while maintaining comparable performance.

Qingjie Zhang, Di Wang, Haoting Qian et al.

Intrinsic self-correction was proposed to improve LLMs' responses via feedback prompts solely based on their inherent capability. However, recent works show that LLMs' intrinsic self-correction fails without oracle labels as feedback prompts. In this paper, we aim to interpret LLMs' intrinsic self-correction for different tasks, especially for those failure cases. By including one simple task and three complex tasks with state-of-the-art (SOTA) LLMs like ChatGPT families (o1, 4o, 3.5-turbo) and Llama families (2-7B, 3-8B, and 3.1-8B), we design three interpretation methods to reveal the dark side of LLMs' intrinsic self-correction. We identify intrinsic self-correction can (1) cause LLMs to waver both intermedia and final answers and lead to prompt bias on simple factual questions; (2) introduce human-like cognitive bias on complex tasks. In light of our findings, we also provide two simple yet effective strategies for alleviation: question repeating and supervised fine-tuning with a few samples. We open-source our work at https://x-isc.info/.

Jiale Cheng, Ruiliang Lyu, Xiaotao Gu et al. · tsinghua

Video generation models have achieved remarkable progress in text-to-video tasks. These models are typically trained on text-video pairs with highly detailed and carefully crafted descriptions, while real-world user inputs during inference are often concise, vague, or poorly structured. This gap makes prompt optimization crucial for generating high-quality videos. Current methods often rely on large language models (LLMs) to refine prompts through in-context learning, but suffer from several limitations: they may distort user intent, omit critical details, or introduce safety risks. Moreover, they optimize prompts without considering the impact on the final video quality, which can lead to suboptimal results. To address these issues, we introduce VPO, a principled framework that optimizes prompts based on three core principles: harmlessness, accuracy, and helpfulness. The generated prompts faithfully preserve user intents and, more importantly, enhance the safety and quality of generated videos. To achieve this, VPO employs a two-stage optimization approach. First, we construct and refine a supervised fine-tuning (SFT) dataset based on principles of safety and alignment. Second, we introduce both text-level and video-level feedback to further optimize the SFT model with preference learning. Our extensive experiments demonstrate that VPO significantly improves safety, alignment, and video quality compared to baseline methods. Moreover, VPO shows strong generalization across video generation models. Furthermore, we demonstrate that VPO could outperform and be combined with RLHF methods on video generation models, underscoring the effectiveness of VPO in aligning video generation models. Our code and data are publicly available at https://github.com/thu-coai/VPO.

Shu-Xun Yang, Cunxiang Wang, Yidong Wang et al.

Evaluating mathematical capabilities is critical for assessing the overall performance of large language models (LLMs). However, existing evaluation methods often focus solely on final answers, resulting in highly inaccurate and uninterpretable evaluation outcomes, as well as their failure to assess proof or open-ended problems. To address these issues, we propose a novel mathematical process evaluation agent based on Tree-of-Error, called StepMathAgent. This agent incorporates four internal core operations: logical step segmentation, step scoring, score aggregation and error tree generation, along with four external extension modules: difficulty calibration, simplicity evaluation, completeness validation and format assessment. Furthermore, we introduce StepMathBench, a benchmark comprising 1,000 step-divided process evaluation instances, derived from 200 high-quality math problems grouped by problem type, subject category and difficulty level. Experiments on StepMathBench show that our proposed StepMathAgent outperforms all state-of-the-art methods, demonstrating human-aligned evaluation preferences and broad applicability to various scenarios. Our data and code are available at https://github.com/SHU-XUN/StepMathAgent.

Jiale Cheng, Yida Lu, Xiaotao Gu et al.

Although Large Language Models (LLMs) are becoming increasingly powerful, they still exhibit significant but subtle weaknesses, such as mistakes in instruction-following or coding tasks. As these unexpected errors could lead to severe consequences in practical deployments, it is crucial to investigate the limitations within LLMs systematically. Traditional benchmarking approaches cannot thoroughly pinpoint specific model deficiencies, while manual inspections are costly and not scalable. In this paper, we introduce a unified framework, AutoDetect, to automatically expose weaknesses in LLMs across various tasks. Inspired by the educational assessment process that measures students' learning outcomes, AutoDetect consists of three LLM-powered agents: Examiner, Questioner, and Assessor. The collaboration among these three agents is designed to realize comprehensive and in-depth weakness identification. Our framework demonstrates significant success in uncovering flaws, with an identification success rate exceeding 30% in prominent models such as ChatGPT and Claude. More importantly, these identified weaknesses can guide specific model improvements, proving more effective than untargeted data augmentation methods like Self-Instruct. Our approach has led to substantial enhancements in popular LLMs, including the Llama series and Mistral-7b, boosting their performance by over 10% across several benchmarks. Code and data are publicly available at https://github.com/thu-coai/AutoDetect.

Kaidi Xu, Zhouxing Shi, Huan Zhang et al.

Linear relaxation based perturbation analysis (LiRPA) for neural networks, which computes provable linear bounds of output neurons given a certain amount of input perturbation, has become a core component in robustness verification and certified defense. The majority of LiRPA-based methods focus on simple feed-forward networks and need particular manual derivations and implementations when extended to other architectures. In this paper, we develop an automatic framework to enable perturbation analysis on any neural network structures, by generalizing existing LiRPA algorithms such as CROWN to operate on general computational graphs. The flexibility, differentiability and ease of use of our framework allow us to obtain state-of-the-art results on LiRPA based certified defense on fairly complicated networks like DenseNet, ResNeXt and Transformer that are not supported by prior works. Our framework also enables loss fusion, a technique that significantly reduces the computational complexity of LiRPA for certified defense. For the first time, we demonstrate LiRPA based certified defense on Tiny ImageNet and Downscaled ImageNet where previous approaches cannot scale to due to the relatively large number of classes. Our work also yields an open-source library for the community to apply LiRPA to areas beyond certified defense without much LiRPA expertise, e.g., we create a neural network with a probably flat optimization landscape by applying LiRPA to network parameters. Our opensource library is available at https://github.com/KaidiXu/auto_LiRPA.

Fei Huang, Dazhen Wan, Zhihong Shao et al.

In text generation evaluation, many practical issues, such as inconsistent experimental settings and metric implementations, are often ignored but lead to unfair evaluation and untenable conclusions. We present CoTK, an open-source toolkit aiming to support fast development and fair evaluation of text generation. In model development, CoTK helps handle the cumbersome issues, such as data processing, metric implementation, and reproduction. It standardizes the development steps and reduces human errors which may lead to inconsistent experimental settings. In model evaluation, CoTK provides implementation for many commonly used metrics and benchmark models across different experimental settings. As a unique feature, CoTK can signify when and which metric cannot be fairly compared. We demonstrate that it is convenient to use CoTK for model development and evaluation, particularly across different experimental settings.

Guanqun Bi, Zhuang Chen, Zhoufu Liu et al.

Automating structured clinical interviews could revolutionize mental healthcare accessibility, yet existing large language models (LLMs) approaches fail to align with psychiatric diagnostic protocols. We present MAGI, the first framework that transforms the gold-standard Mini International Neuropsychiatric Interview (MINI) into automatic computational workflows through coordinated multi-agent collaboration. MAGI dynamically navigates clinical logic via four specialized agents: 1) an interview tree guided navigation agent adhering to the MINI's branching structure, 2) an adaptive question agent blending diagnostic probing, explaining, and empathy, 3) a judgment agent validating whether the response from participants meet the node, and 4) a diagnosis Agent generating Psychometric Chain-of- Thought (PsyCoT) traces that explicitly map symptoms to clinical criteria. Experimental results on 1,002 real-world participants covering depression, generalized anxiety, social anxiety and suicide shows that MAGI advances LLM- assisted mental health assessment by combining clinical rigor, conversational adaptability, and explainable reasoning.

Jinfeng Zhou, Yuxuan Chen, Jianing Yin et al.

Cognitive Restructuring (CR) is a psychotherapeutic process aimed at identifying and restructuring an individual's negative thoughts, arising from mental health challenges, into more helpful and positive ones via multi-turn dialogues. Clinician shortage and stigma urge the development of human-LLM interactive psychotherapy for CR. Yet, existing efforts implement CR via simple text rewriting, fixed-pattern dialogues, or a one-shot CR workflow, failing to align with the psychotherapeutic process for effective CR. To address this gap, we propose CRDial, a novel framework for CR, which creates multi-turn dialogues with specifically designed identification and restructuring stages of negative thoughts, integrates sentence-level supportive conversation strategies, and adopts a multi-channel loop mechanism to enable iterative CR. With CRDial, we distill Crisp, a large-scale and high-quality bilingual dialogue dataset, from LLM. We then train Crispers, Crisp-based conversational LLMs for CR, at 7B and 14B scales. Extensive human studies show the superiority of Crispers in pointwise, pairwise, and intervention evaluations.

Qingjie Zhang, Yujia Fu, Yang Wang et al.

Large Reasoning Models (LRMs) have shown impressive performance on complex reasoning tasks such as mathematics, yet they also display misbehaviors that expose their limitations. In particular, when faced with hard questions, LRMs often engage in unproductive reasoning until context limit, producing wrong answers while wasting substantial computation. This phenomenon reflects a fundamental issue: current answering paradigms overlook the relationship between questions and LRMs' capability boundaries. In this paper, we investigate whether LRMs possess self-awareness of capability boundaries. We begin by an observation that LRMs may know what they cannot solve through expressed reasoning confidence. For black-box models, we find that reasoning expressions reveal boundary signals, with accelerated growing confidence trajectory for solvable problems but convergent uncertainty trajectory for unsolvable ones. For white-box models, we show that hidden states of the last input token encode boundary information, with solvable and unsolvable problems linearly separable even before reasoning begins. Building on these findings, we propose two simple yet effective optimization strategies: reasoning expression monitoring and hidden states monitoring. Experiments demonstrate that these boundary-aware strategies enable LRMs to avoid unproductive reasoning without sacrificing accuracy, significantly improving reliability and efficiency by cutting token usage up to 62.7 - 93.6%.

Yue Zhang, Zhiliang Tian, Shicheng Zhou et al.

Legal Judgment Prediction (LJP) is a pivotal task in legal AI. Existing semantic-enhanced LJP models integrate judicial precedents and legal knowledge for high performance. But they neglect legal reasoning logic, a critical component of legal judgments requiring rigorous logical analysis. Although some approaches utilize legal reasoning logic for high-quality predictions, their logic rigidity hinders adaptation to case-specific logical frameworks, particularly in complex cases that are lengthy and detailed. This paper proposes a rule-enhanced legal judgment prediction framework based on first-order logic (FOL) formalism and comparative learning (CL) to develop an adaptive adjustment mechanism for legal judgment logic and further enhance performance in LJP. Inspired by the process of human exam preparation, our method follows a three-stage approach: first, we initialize judgment rules using the FOL formalism to capture complex reasoning logic accurately; next, we propose a Confusion-aware Contrastive Learning (CACL) to dynamically optimize the judgment rules through a quiz consisting of confusable cases; finally, we utilize the optimized judgment rules to predict legal judgments. Experimental results on two public datasets show superior performance across all metrics. The code is publicly available{https://anonymous.4open.science/r/RLJP-FDF1}.

Shen Gao, Hao Li, Chengrui Huang et al.

Large language model agents have demonstrated remarkable advancements across various complex tasks. Recent works focus on optimizing the agent team or employing self-reflection to iteratively solve complex tasks. Since these agents are all based on the same LLM, only conducting self-evaluation or removing underperforming agents does not substantively enhance the capability of the agents. We argue that a comprehensive evaluation and accumulating experience from evaluation feedback is an effective approach to improving system performance. In this paper, we propose Reusable Experience Accumulation with 360$^\circ$ Assessment (360$^\circ$REA), a hierarchical multi-agent framework inspired by corporate organizational practices. The framework employs a novel 360$^\circ$ performance assessment method for multi-perspective performance evaluation with fine-grained assessment. To enhance the capability of agents in addressing complex tasks, we introduce dual-level experience pool for agents to accumulate experience through fine-grained assessment. Extensive experiments on complex task datasets demonstrate the effectiveness of 360$^\circ$REA.

Yi Feng, Jiaqi Wang, Wenxuan Zhang et al. · tencent-ai

Recent progress in large language models (LLMs) has opened new possibilities for mental health support, yet current approaches lack realism in simulating specialized psychotherapy and fail to capture therapeutic progression over time. Narrative therapy, which helps individuals transform problematic life stories into empowering alternatives, remains underutilized due to limited access and social stigma. We address these limitations through a comprehensive framework with two core components. First, INT (Interactive Narrative Therapist) simulates expert narrative therapists by planning therapeutic stages, guiding reflection levels, and generating contextually appropriate expert-like responses. Second, IMA (Innovative Moment Assessment) provides a therapy-centric evaluation method that quantifies effectiveness by tracking "Innovative Moments" (IMs), critical narrative shifts in client speech signaling therapy progress. Experimental results on 260 simulated clients and 230 human participants reveal that INT consistently outperforms standard LLMs in therapeutic quality and depth. We further demonstrate the effectiveness of INT in synthesizing high-quality support conversations to facilitate social applications.

Jiaxin Wen, Ruiqi Zhong, Pei Ke et al.

When using language models (LMs) to solve complex problems, humans might struggle to understand the LM-generated solutions and repair the flawed ones. To assist humans in repairing them, we propose to automatically decompose complex solutions into multiple simpler pieces that correspond to specific subtasks. We introduce a novel objective for learning task decomposition, termed assistive value (AssistV), which measures the feasibility and speed for humans to repair the decomposed solution. We collect a dataset of human repair experiences on different decomposed solutions. Utilizing the collected data as in-context examples, we then learn to critique, refine, and rank decomposed solutions to improve AssistV. We validate our method under competitive programming problems: under 177 hours of human study, our method enables non-experts to solve 33.3\% more problems, speeds them up by 3.3x, and empowers them to match unassisted experts.

Rongsheng Zhang, Xiaoxi Mao, Le Li et al.

Recently, a variety of neural models have been proposed for lyrics generation. However, most previous work completes the generation process in a single pass with little human intervention. We believe that lyrics creation is a creative process with human intelligence centered. AI should play a role as an assistant in the lyrics creation process, where human interactions are crucial for high-quality creation. This paper demonstrates \textit{Youling}, an AI-assisted lyrics creation system, designed to collaborate with music creators. In the lyrics generation process, \textit{Youling} supports traditional one pass full-text generation mode as well as an interactive generation mode, which allows users to select the satisfactory sentences from generated candidates conditioned on preceding context. The system also provides a revision module which enables users to revise undesired sentences or words of lyrics repeatedly. Besides, \textit{Youling} allows users to use multifaceted attributes to control the content and format of generated lyrics. The demo video of the system is available at https://youtu.be/DFeNpHk0pm4.

Rongsheng Zhang, Yinhe Zheng, Xiaoxi Mao et al.

Unsupervised domain adaptation (UDA) with pre-trained language models (PrLM) has achieved promising results since these pre-trained models embed generic knowledge learned from various domains. However, fine-tuning all the parameters of the PrLM on a small domain-specific corpus distort the learned generic knowledge, and it is also expensive to deployment a whole fine-tuned PrLM for each domain. This paper explores an adapter-based fine-tuning approach for unsupervised domain adaptation. Specifically, several trainable adapter modules are inserted in a PrLM, and the embedded generic knowledge is preserved by fixing the parameters of the original PrLM at fine-tuning. A domain-fusion scheme is introduced to train these adapters using a mix-domain corpus to better capture transferable features. Elaborated experiments on two benchmark datasets are carried out, and the results demonstrate that our approach is effective with different tasks, dataset sizes, and domain similarities.

Fei Mi, Wanhao Zhou, Fengyu Cai et al.

As the labeling cost for different modules in task-oriented dialog (ToD) systems is expensive, a major challenge is to train different modules with the least amount of labeled data. Recently, large-scale pre-trained language models, have shown promising results for few-shot learning in ToD. In this paper, we devise a self-training approach to utilize the abundant unlabeled dialog data to further improve state-of-the-art pre-trained models in few-shot learning scenarios for ToD systems. Specifically, we propose a self-training approach that iteratively labels the most confident unlabeled data to train a stronger Student model. Moreover, a new text augmentation technique (GradAug) is proposed to better train the Student by replacing non-crucial tokens using a masked language model. We conduct extensive experiments and present analyses on four downstream tasks in ToD, including intent classification, dialog state tracking, dialog act prediction, and response selection. Empirical results demonstrate that the proposed self-training approach consistently improves state-of-the-art pre-trained models (BERT, ToD-BERT) when only a small number of labeled data are available.

Yinhe Zheng, Yida Wang, Pei Ke et al.

This paper propose to combine pretrained language models with the modular dialogue paradigm for open-domain dialogue modeling. Our method, semantic-enhanced finetuning, instantiates conversation understanding, planning, and response generation as a language model finetuning task. At inference, we disentangle semantic and token variations by specifying sampling methods and constraints for each module separately. For training and evaluation, we present X-Weibo, a Chinese multi-turn open-domain dialogue dataset with automatic annotation for emotions, DAs, and topical words. Experiments show that semantic-enhanced finetuning outperforms strong baselines on non-semantic and semantic metrics, improves the human-evaluated relevance, coherence, and informativeness, and exhibits considerable controllability over semantic variables.

Zhihong Shao, Zitao Liu, Jiyong Zhang et al.

Adversarial examples are vital to expose the vulnerability of machine learning models. Despite the success of the most popular substitution-based methods which substitutes some characters or words in the original examples, only substitution is insufficient to uncover all robustness issues of models. In this paper, we present AdvExpander, a method that crafts new adversarial examples by expanding text, which is complementary to previous substitution-based methods. We first utilize linguistic rules to determine which constituents to expand and what types of modifiers to expand with. We then expand each constituent by inserting an adversarial modifier searched from a CVAE-based generative model which is pre-trained on a large scale corpus. To search adversarial modifiers, we directly search adversarial latent codes in the latent space without tuning the pre-trained parameters. To ensure that our adversarial examples are label-preserving for text matching, we also constrain the modifications with a heuristic rule. Experiments on three classification tasks verify the effectiveness of AdvExpander and the validity of our adversarial examples. AdvExpander crafts a new type of adversarial examples by text expansion, thereby promising to reveal new robustness issues.

Chencheng Xu, Qiao Liu, Minlie Huang et al.

A major challenge in the pharmaceutical industry is to design novel molecules with specific desired properties, especially when the property evaluation is costly. Here, we propose MNCE-RL, a graph convolutional policy network for molecular optimization with molecular neighborhood-controlled embedding grammars through reinforcement learning. We extend the original neighborhood-controlled embedding grammars to make them applicable to molecular graph generation and design an efficient algorithm to infer grammatical production rules from given molecules. The use of grammars guarantees the validity of the generated molecular structures. By transforming molecular graphs to parse trees with the inferred grammars, the molecular structure generation task is modeled as a Markov decision process where a policy gradient strategy is utilized. In a series of experiments, we demonstrate that our approach achieves state-of-the-art performance in a diverse range of molecular optimization tasks and exhibits significant superiority in optimizing molecular properties with a limited number of property evaluations.

Fei Mi, Liangwei Chen, Mengjie Zhao et al.

Natural language generation (NLG) is an essential component of task-oriented dialog systems. Despite the recent success of neural approaches for NLG, they are typically developed in an offline manner for particular domains. To better fit real-life applications where new data come in a stream, we study NLG in a "continual learning" setting to expand its knowledge to new domains or functionalities incrementally. The major challenge towards this goal is catastrophic forgetting, meaning that a continually trained model tends to forget the knowledge it has learned before. To this end, we propose a method called ARPER (Adaptively Regularized Prioritized Exemplar Replay) by replaying prioritized historical exemplars, together with an adaptive regularization technique based on ElasticWeight Consolidation. Extensive experiments to continually learn new domains and intents are conducted on MultiWoZ-2.0 to benchmark ARPER with a wide range of techniques. Empirical results demonstrate that ARPER significantly outperforms other methods by effectively mitigating the detrimental catastrophic forgetting issue.

Yinhe Zheng, Zikai Chen, Rongsheng Zhang et al.

Generating stylized responses is essential to build intelligent and engaging dialogue systems. However, this task is far from well-explored due to the difficulties of rendering a particular style in coherent responses, especially when the target style is embedded only in unpaired texts that cannot be directly used to train the dialogue model. This paper proposes a stylized dialogue generation method that can capture stylistic features embedded in unpaired texts. Specifically, our method can produce dialogue responses that are both coherent to the given context and conform to the target style. In this study, an inverse dialogue model is first introduced to predict possible posts for the input responses, and then this inverse model is used to generate stylized pseudo dialogue pairs based on these stylized unpaired texts. Further, these pseudo pairs are employed to train the stylized dialogue model with a joint training process, and a style routing approach is proposed to intensify stylistic features in the decoder. Automatic and manual evaluations on two datasets demonstrate that our method outperforms competitive baselines in producing coherent and style-intensive dialogue responses.

Mantong Zhou, Zhouxing Shi, Minlie Huang et al.

Open-domain question answering (QA) aims to find the answer to a question from a large collection of documents.Though many models for single-document machine comprehension have achieved strong performance, there is still much room for improving open-domain QA systems since document retrieval and answer reranking are still unsatisfactory. Golden documents that contain the correct answers may not be correctly scored by the retrieval component, and the correct answers that have been extracted may be wrongly ranked after other candidate answers by the reranking component. One of the reasons is derived from the independent principle in which each candidate document (or answer) is scored independently without considering its relationship to other documents (or answers). In this work, we propose a knowledge-aided open-domain QA (KAQA) method which targets at improving relevant document retrieval and candidate answer reranking by considering the relationship between a question and the documents (termed as question-document graph), and the relationship between candidate documents (termed as document-document graph). The graphs are built using knowledge triples from external knowledge resources. During document retrieval, a candidate document is scored by considering its relationship to the question and other documents. During answer reranking, a candidate answer is reranked using not only its own context but also the clues from other documents. The experimental results show that our proposed method improves document retrieval and answer reranking, and thereby enhances the overall performance of open-domain question answering.

Ryuichi Takanobu, Qi Zhu, Jinchao Li et al.

There is a growing interest in developing goal-oriented dialog systems which serve users in accomplishing complex tasks through multi-turn conversations. Although many methods are devised to evaluate and improve the performance of individual dialog components, there is a lack of comprehensive empirical study on how different components contribute to the overall performance of a dialog system. In this paper, we perform a system-wise evaluation and present an empirical analysis on different types of dialog systems which are composed of different modules in different settings. Our results show that (1) a pipeline dialog system trained using fine-grained supervision signals at different component levels often obtains better performance than the systems that use joint or end-to-end models trained on coarse-grained labels, (2) component-wise, single-turn evaluation results are not always consistent with the overall performance of a dialog system, and (3) despite the discrepancy between simulators and human users, simulated evaluation is still a valid alternative to the costly human evaluation especially in the early stage of development.

Zheng Zhang, Ryuichi Takanobu, Qi Zhu et al.

Due to the significance and value in human-computer interaction and natural language processing, task-oriented dialog systems are attracting more and more attention in both academic and industrial communities. In this paper, we survey recent advances and challenges in task-oriented dialog systems. We also discuss three critical topics for task-oriented dialog systems: (1) improving data efficiency to facilitate dialog modeling in low-resource settings, (2) modeling multi-turn dynamics for dialog policy learning to achieve better task-completion performance, and (3) integrating domain ontology knowledge into the dialog model. Besides, we review the recent progresses in dialog evaluation and some widely-used corpora. We believe that this survey, though incomplete, can shed a light on future research in task-oriented dialog systems.

Lizi Liao, Ryuichi Takanobu, Yunshan Ma et al.

When traveling to a foreign country, we are often in dire need of an intelligent conversational agent to provide instant and informative responses to our various queries. However, to build such a travel agent is non-trivial. First of all, travel naturally involves several sub-tasks such as hotel reservation, restaurant recommendation and taxi booking etc, which invokes the need for global topic control. Secondly, the agent should consider various constraints like price or distance given by the user to recommend an appropriate venue. In this paper, we present a Deep Conversational Recommender (DCR) and apply to travel. It augments the sequence-to-sequence (seq2seq) models with a neural latent topic component to better guide response generation and make the training easier. To consider the various constraints for venue recommendation, we leverage a graph convolutional network (GCN) based approach to capture the relationships between different venues and the match between venue and dialog context. For response generation, we combine the topic-based component with the idea of pointer networks, which allows us to effectively incorporate recommendation results. We perform extensive evaluation on a multi-turn task-oriented dialog dataset in travel domain and the results show that our method achieves superior performance as compared to a wide range of baselines.

Jun Feng, Minlie Huang, Yijie Zhang et al.

In this paper we address a task of relation mention extraction from noisy data: extracting representative phrases for a particular relation from noisy sentences that are collected via distant supervision. Despite its significance and value in many downstream applications, this task is less studied on noisy data. The major challenges exists in 1) the lack of annotation on mention phrases, and more severely, 2) handling noisy sentences which do not express a relation at all. To address the two challenges, we formulate the task as a semi-Markov decision process and propose a novel hierarchical reinforcement learning model. Our model consists of a top-level sentence selector to remove noisy sentences, a low-level mention extractor to extract relation mentions, and a reward estimator to provide signals to guide data denoising and mention extraction without explicit annotations. Experimental results show that our model is effective to extract relation mentions from noisy data.

Yilin Niu, Chao Qiao, Hang Li et al.

Text similarity calculation is a fundamental problem in natural language processing and related fields. In recent years, deep neural networks have been developed to perform the task and high performances have been achieved. The neural networks are usually trained with labeled data in supervised learning, and creation of labeled data is usually very costly. In this short paper, we address unsupervised learning for text similarity calculation. We propose a new method called Word Embedding based Edit Distance (WED), which incorporates word embedding into edit distance. Experiments on three benchmark datasets show WED outperforms state-of-the-art unsupervised methods including edit distance, TF-IDF based cosine, word embedding based cosine, Jaccard index, etc.

Tom Young, Erik Cambria, Iti Chaturvedi et al.

Building dialog agents that can converse naturally with humans is a challenging yet intriguing problem of artificial intelligence. In open-domain human-computer conversation, where the conversational agent is expected to respond to human responses in an interesting and engaging way, commonsense knowledge has to be integrated into the model effectively. In this paper, we investigate the impact of providing commonsense knowledge about the concepts covered in the dialog. Our model represents the first attempt to integrating a large commonsense knowledge base into end-to-end conversational models. In the retrieval-based scenario, we propose the Tri-LSTM model to jointly take into account message and commonsense for selecting an appropriate response. Our experiments suggest that the knowledge-augmented models are superior to their knowledge-free counterparts in automatic evaluation.

Qiao Qian, Minlie Huang, Haizhou Zhao et al.

Endowing a chatbot with personality or an identity is quite challenging but critical to deliver more realistic and natural conversations. In this paper, we address the issue of generating responses that are coherent to a pre-specified agent profile. We design a model consisting of three modules: a profile detector to decide whether a post should be responded using the profile and which key should be addressed, a bidirectional decoder to generate responses forward and backward starting from a selected profile value, and a position detector that predicts a word position from which decoding should start given a selected profile value. We show that general conversation data from social media can be used to generate profile-coherent responses. Manual and automatic evaluation shows that our model can deliver more coherent, natural, and diversified responses.

Minlie Huang, Yujie Cao, Chao Dong

Sentiment analysis on social media data such as tweets and weibo has become a very important and challenging task. Due to the intrinsic properties of such data, tweets are short, noisy, and of divergent topics, and sentiment classification on these data requires to modeling various contexts such as the retweet/reply history of a tweet, and the social context about authors and relationships. While few prior study has approached the issue of modeling contexts in tweet, this paper proposes to use a hierarchical LSTM to model rich contexts in tweet, particularly long-range context. Experimental results show that contexts can help us to perform sentiment classification remarkably better.

Jun Feng, Mantong Zhou, Yu Hao et al.

Knowledge graph embedding refers to projecting entities and relations in knowledge graph into continuous vector spaces. State-of-the-art methods, such as TransE, TransH, and TransR build embeddings by treating relation as translation from head entity to tail entity. However, previous models can not deal with reflexive/one-to-many/many-to-one/many-to-many relations properly, or lack of scalability and efficiency. Thus, we propose a novel method, flexible translation, named TransF, to address the above issues. TransF regards relation as translation between head entity vector and tail entity vector with flexible magnitude. To evaluate the proposed model, we conduct link prediction and triple classification on benchmark datasets. Experimental results show that our method remarkably improve the performance compared with several state-of-the-art baselines.

Biao Liu, Minlie Huang

Prior knowledge has been shown very useful to address many natural language processing tasks. Many approaches have been proposed to formalise a variety of knowledge, however, whether the proposed approach is robust or sensitive to the knowledge supplied to the model has rarely been discussed. In this paper, we propose three regularization terms on top of generalized expectation criteria, and conduct extensive experiments to justify the robustness of the proposed methods. Experimental results demonstrate that our proposed methods obtain remarkable improvements and are much more robust than baselines.