论文检索

Tongyoung Kim, Jeongeun Lee, SooJin Yoon, SungHwan Kim, Dongha Lee

Conversational Recommender Systems (CRSs) aim to engage users in dialogue to provide tailored recommendations. While traditional CRSs focus on eliciting preferences and retrieving items, real-world e-commerce interactions involve more complex decision-making, where users consider multiple factors beyond simple attributes. To capture this complexity, we introduce Conversational Sales (CSALES), a novel task that integrates preference elicitation, recommendation, and persuasion within a unified conversational framework. To support realistic and systematic evaluation, we present CSUSER, an evaluation protocol with LLM-based user simulator grounded in real-world behavioral data by modeling fine-grained user profiles for personalized interaction. We also propose CSI, a conversational sales agent that proactively infers contextual user profiles and strategically selects actions through conversation. Comprehensive experiments show that CSI significantly improves both recommendation success and persuasive effectiveness across diverse user profiles.

Yue Zhao, Xiaoyu Wang, Dan Wang, Zhonglin Jiang, Qingqing Gu, Teng Chen, Ningyuan Xi, Jinxian Qu, Yong Chen, Luo Ji

World models have been widely utilized in robotics, gaming, and autonomous driving. However, their applications to natural language tasks are relatively limited. In this paper, we construct the dialogue world model, which could predict future utterances and user beliefs, including emotion, sentiment, and intention. In this paper, we propose a framework called DreamCUB, which shows that this user belief modeling and the entire dialogue world model can be established by LLM post-training. By defining a POMDP, we apply model-based reinforcement learning to the dialogue system and solve it by maximizing the information bottleneck. Experiments show that the pretrained dialogue world model can achieve state-of-the-art performances on emotion classification and sentiment identification, while dialogue quality is also enhanced by joint training of policy, critic and dialogue world model. Further analysis reveals that DreamCUB holds a reasonable exploration-exploitation balance and also transfers well to out-of-domain scenarios such as empathetic dialogues.

Pengchao Feng, Ziyang Ma, Wenxi Chen, Yao Li, Sheng Wang, Kai Yu, Xie Chen

End-to-end speech-to-speech (S2S) dialogue systems have recently garnered increasing research attention for their lower latency and more natural integration of nonverbal cues such as emotion and speaker identity. However, these systems face key challenges, particularly in incorporating external knowledge, a capability commonly addressed by Retrieval-Augmented Generation (RAG) in text-based large language models (LLMs). The core difficulty lies in the modality gap between input speech and retrieved textual knowledge, which hinders effective integration of information. To address this issue, we propose a novel end-to-end RAG framework that directly retrieves relevant textual knowledge from speech queries. Experimental results demonstrate that our method significantly improves the performance of end-to-end S2S dialogue systems while achieving higher retrieval efficiency. Although the overall performance still lags behind the SOTA cascaded models, our framework offers a promising direction for enhancing knowledge integration in end-to-end S2S systems. Our code and dataset are released.

Jason Lucas, John Chen, Ali Al-Lawati, Mahjabin Nahar, Mahnoosh Mehrabani

Large Language Models (LLMs) have introduced paradigm-shifting approaches in natural language processing. Yet, their transformative in-context learning (ICL) capabilities remain underutilized, especially in customer service dialogue summarization—a domain plagued by generative hallucinations, detail omission, and inconsistencies. We present Chain-of-Interactions (CoI), a novel single-instance, multi-step framework that orchestrates information extraction, self-correction, and evaluation through sequential interactive generation chains. By strategically leveraging LLMs’ ICL capabilities through precisely engineered prompts, CoI dramatically enhances abstractive task-oriented dialogue summarization (ATODS) quality and usefulness. Our comprehensive evaluation on real-world and benchmark human-agent interaction datasets demonstrates CoI’s effectiveness through rigorous testing across 11 models and 7 prompting approaches, with 9 standard automatic evaluation metrics, 3 LLM-based evaluations, and human studies involving 480 evaluators across 9 quality dimensions. Results reveal CoI’s decisive superiority, outperforming all single-step approaches and achieving 6× better entity preservation, 49% higher quality scores, and 322% improvement in accuracy compared to state-of-the-art multi-step Chain-of-Density (CoD). This research addresses critical gaps in task-oriented dialogue summarization for customer service applications and establishes new standards for harnessing LLMs’ reasoning capabilities in practical, industry-relevant contexts.

Yuansheng Ni, Ping Nie, Kai Zou, Xiang Yue, Wenhu Chen

Large language models (LLMs) often struggle with visualization tasks like plotting diagrams, charts, where success depends on both code correctness and visual semantics. Existing instruction-tuning datasets lack execution-grounded supervision and offer limited support for iterative code correction, resulting in fragile and unreliable plot generation. We present **VisCode-200K**, a large-scale instruction tuning dataset for Python-based visualization and self-correction. It contains over 200K examples from two sources: (1) validated plotting code from open-source repositories, paired with natural language instructions and rendered plots; and (2) 45K multi-turn correction dialogues from Code-Feedback, enabling models to revise faulty code using runtime feedback. We fine-tune Qwen2.5-Coder-Instruct on VisCode-200K to create **VisCoder**, and evaluate it on PandasPlotBench. VisCoder significantly outperforms strong open-source baselines and approaches the performance of proprietary models like GPT-4o-mini. We further adopt a self-debug evaluation protocol to assess iterative repair, demonstrating the benefits of feedback-driven learning for executable, visually accurate code generation.

Xinping Lei, Tong Zhou, Yubo Chen, Kang Liu, Jun Zhao

Large Language Models (LLMs) hold significant promise for accelerating academic ideation but face critical challenges in grounding ideas and mitigating confirmation bias during refinement. To address these limitations, we propose MotivGraph-SoIQ, a novel framework that enhances LLM ideation by integrating a Motivational Knowledge Graph (MotivGraph), which provides essential grounding from research literature, with a Q-Driven Socratic Ideator. The Ideator, a dual-agent system utilizing Socratic questioning, facilitates a rigorous refinement process that mitigates confirmation bias and significantly improves idea quality across dimensions of novelty, experimental feasibility, and motivation. Our experimental results demonstrate MotivGraph-SoIQ’s effectiveness. Comparative studies show significant quantitative improvements over SOTA methods across LLM-based scoring, ELO ranking, and human evaluation. Ablation studies further validate the crucial contributions of both the MotivGraph for enhancing idea novelty and practicality, and the Socratic dialogue with the teacher agent for substantial quality improvement. This work underscores the potential of combining structured knowledge with interactive, critique-based refinement for robust LLM ideation.

Xiantao Zhang

This paper introduces AuraDial, a large-scale, human-centric dialogue dataset for Chinese AI psychological counseling, comprising over 300,000 single-turn dialogues and 90,000 multi-turn dialogue sessions. A key distinction of AuraDial is its instruction set, primarily derived from real-world user queries, better reflecting genuine expression patterns compared to synthetic or template-based alternatives. Furthermore, we propose an innovative rephrasing-based data generation methodology designed to foster more human-like and empathetic responses, addressing a common shortcoming in AI-generated dialogue. Experimental results demonstrate that models fine-tuned on AuraDial significantly outperform those trained on other public datasets in generating empathetic and relevant replies. AuraDial offers a novel, valuable resource to the Chinese NLP community for advancing AI in psychological counseling. The dataset is publicly available at [https://huggingface.co/datasets/Mxode/AuraDial](https://huggingface.co/datasets/Mxode/AuraDial).

Weiting Tan, Jiachen Lian, Hirofumi Inaguma, Paden Tomasello, Philipp Koehn, Xutai Ma

We present an Audio-Visual Language Model (AVLM) for expressive speech generation by integrating full-face visual cues into a pre-trained expressive speech model. We explore multiple visual encoders and multimodal fusion strategies during pre-training to identify the most effective integration approach. Subsequent fine-tuning on emotion recognition and expressive dialogue tasks yields substantial gains over speech-only baselines (e.g., +5 F1 in emotion recognition). AVLM highlights the value of expressive visual information in guiding speech generation and offers a foundation for end-to-end multimodal conversational systems.

Shiki Sato, Jun Baba, Asahi Hentona, Shinji Iwata, Akifumi Yoshimoto, Koichiro Yoshino

Chat-oriented dialogue systems that deliver tangible benefits, such as sharing news or frailty prevention for seniors, require proactive acquisition of specific user information via chats on user-favored topics. This study proposes the Proactive Information Acquisition (PIA) task to support the development of these systems. In this task, a system needs to acquire a user’s answers to predefined questions without making the user feel abrupt while engaging in a chat on a predefined topic. We created and analyzed a dataset of 650 PIA chats, identifying key challenges and effective strategies for recent LLMs. Our system, designed from these insights, surpassed the performance of LLMs prompted solely with task instructions. Finally, we demonstrate that automatic evaluation of this task is reasonably accurate, suggesting its potential as a framework to efficiently develop techniques for systems dealing with complex dialogue goals, extending beyond the scope of PIA alone. Our dataset is available at: https://github.com/CyberAgentAILab/PIA

Jiseung Hong, Grace Byun, Seungone Kim, Kai Shu

Large Language Models (LLMs) are expected to provide helpful and harmless responses, yet they often exhibit sycophancy—conforming to user beliefs regardless of factual accuracy or ethical soundness. Prior research on sycophancy has primarily focused on single-turn factual correctness, overlooking the dynamics of real-world interactions. In this work, we introduce SYCON Bench (SYcophantic CONformity benchmark), a novel evaluation suite that assesses sycophantic behavior in multi-turn, free-form conversational settings. Our benchmark measures how quickly a model conforms to the user (Turn of Flip) and how frequently it shifts its stance under sustained user pressure (Number of Flip). Applying SYCON Bench to 17 LLMs across three real-world scenarios, we find that sycophancy remains a prevalent failure mode. Our analysis shows that alignment tuning amplifies sycophantic behavior, whereas model scaling and reasoning optimization strengthen the model’s ability to resist undesirable user views. Reasoning models generally outperform instruction-tuned models but often fail when they over-index on logical exposition instead of directly addressing the user’s underlying beliefs. Finally, we evaluate four additional prompting strategies and demonstrate that adopting a third-person perspective reduces sycophancy by up to 63.8% in debate scenario.

Yuanchen Shi, Jiawang Hao, Fang Kong

Emotional support conversation systems aim to help users alleviate distress through empathetic dialogue. However, existing ESC datasets often use coarse-grained problem categories, limiting models’ ability to address users’ complex, overlapping challenges. To address this, we propose a generalizable fine-grained problem enhancement method that systematically augments problem types, user scenarios, and profiles, enabling the construction of richer and more diverse ESC corpora. As a demonstration, we construct EmoCare, a large-scale ESC dataset with 2.6K dialogues and 42.8K utterances, expanding problem type coverage from 13 to 45 fine-grained categories. Building on this data augmentation process, we introduce FPEMF, a flexible framework for empathetic dialogue generation, which comprises two modules: fine-grained problem enhancement and multi-dimensional feedback, which can be seamlessly integrated with various backbone models. The multi-dimensional feedback module evaluates responses from four perspectives: emotional understanding, strategy effectiveness, contextual consistency, and topic relevance, guiding models to generate more supportive replies. Experiments show that FPEMF consistently improves both automatic and human evaluation metrics across different models.

Qianli Wang, Tatiana Anikina, Nils Feldhus, Simon Ostermann, Fedor Splitt, Jiaao Li, Yoana Tsoneva, Sebastian Möller, Vera Schmitt

Conversational explainable artificial intelligence (ConvXAI) systems based on large language models (LLMs) have garnered considerable attention for their ability to enhance user comprehension through dialogue-based explanations. Current ConvXAI systems often are based on intent recognition to accurately identify the user’s desired intention and map it to an explainability method. While such methods offer great precision and reliability in discerning users’ underlying intentions for English, a significant challenge in the scarcity of training data persists, which impedes multilingual generalization. Besides, the support for free-form custom inputs, which are user-defined data distinct from pre-configured dataset instances, remains largely limited. To bridge these gaps, we first introduce MultiCoXQL, a multilingual extension of the CoXQL dataset spanning five typologically diverse languages, including one low-resource language. Subsequently, we propose a new parsing approach aimed at enhancing multilingual parsing performance, and evaluate three LLMs on MultiCoXQL using various parsing strategies. Furthermore, we present Compass, a new multilingual dataset designed for custom input extraction in ConvXAI systems, encompassing 11 intents across the same five languages as MultiCoXQL. We conduct monolingual, cross-lingual, and multilingual evaluations on Compass, employing three LLMs of varying sizes alongside BERT-type models.

Cheng-Han Chiang, Xiaofei Wang, Chung-Ching Lin, Kevin Lin, Linjie Li, Radu Kopetz, Yao Qian, Zhendong Wang, Zhengyuan Yang, Hung-yi Lee 等

Audio-aware large language models (ALLMs) can understand the textual and non-textual information in the audio input. In this paper, we explore using ALLMs as an automatic judge to assess the speaking styles of speeches. We use ALLM judges to evaluate the speeches generated by SLMs on two tasks: voice style instruction following and role-playing. The speaking style we consider includes emotion, volume, speaking pace, word emphasis, pitch control, and non-verbal elements. We use four spoken language models (SLMs) to complete the two tasks and use humans and ALLMs to judge the SLMs’ responses. We compare two ALLM judges, GPT-4o-audio and Gemini-2.5-pro, with human evaluation results and show that the agreement between Gemini and human judges is comparable to the agreement between human evaluators. These promising results show that ALLMs can be used as a judge to evaluate SLMs. Our results also reveal that current SLMs, even GPT-4o-audio, still have room for improvement in controlling the speaking style and generating natural dialogues.

Jing Ye, Rui Wang, Yuchuan Wu, Victor Ma, Feiteng Fang, Fei Huang, Yongbin Li

Reinforcement Learning Fine-Tuning (RLFT) has achieved notable success in tasks with objectively verifiable answers (e.g., code generation, mathematical reasoning), yet struggles with open-ended subjective tasks like role-playing dialogue. Traditional reward modeling approaches, which rely on independent sample-wise scoring, face dual challenges: subjective evaluation criteria and unstable reward signals. Motivated by the insight that human evaluation inherently combines explicit criteria with implicit comparative judgments, we propose Comparative Policy Optimization (CPO). CPO redefines the reward evaluation paradigm by shifting from sample-wise scoring to comparative group-wise scoring. Building on the same principle, we introduce the CharacterArena evaluation framework, which comprises two stages: (1) Contextualized Multi-turn Role-playing Simulation, and (2) Trajectory-level Comparative Evaluation. By operationalizing subjective scoring via objective trajectory comparisons, CharacterArena minimizes contextual bias and enables more robust and fair performance evaluation. Empirical results on CharacterEval, CharacterBench, and CharacterArena confirm that CPO effectively mitigates reward ambiguity and leads to substantial improvements in dialogue quality.

Pier Felice Balestrucci, Ondrej Dusek, Luca Anselma, Alessandro Mazzei

We investigate how large language models (LLMs) can produce personalized dialogue responses, specifically focusing on whether they reflect linguistic styles pertaining to different generations: Baby Boomers, Generation X, Generation Y, and Generation Z. We create P-MultiWoZ, a personalized, generation-specific version of MultiWOZ 2.2, by prompting LLMs, and validate its alignment with the original dataset through automatic and human evaluations. To validate the appropriateness of generational linguistic traits, we introduce GeMoSC, a corpus of generation-annotated movie dialogues. Linguistic analysis and perplexity test suggest that P-MultiWoZ reflects patterns consistent with GeMoSC. Finally, a human evaluation reveals that annotators were able to mostly correctly identify the generation behind P-MultiWoZ dialogues, based only on a single query-reply pair.

Xintong Wang, Yixiao Liu, Jingheng Pan, Liang Ding, Longyue Wang, Chris Biemann

Detoxifying offensive language while preserving the speaker’s original intent is a challenging yet critical goal for improving the quality of online interactions. Although large language models (LLMs) show promise in rewriting toxic content, they often default to overly polite rewrites, distorting the emotional tone and communicative intent. This problem is especially acute in Chinese, where toxicity often arises implicitly through emojis, homophones, or discourse context. We present ToxiRewriteCN, the first Chinese detoxification dataset explicitly designed to preserve sentiment polarity. The dataset comprises 1,556 carefully annotated triplets, each containing a toxic sentence, a sentiment-aligned non-toxic rewrite, and labeled toxic spans. It covers five real-world scenarios: standard expressions, emoji-induced and homophonic toxicity, as well as single-turn and multi-turn dialogues. We evaluate 17 LLMs, including commercial and open-source models with variant architectures, across four dimensions: detoxification accuracy, fluency, content preservation, and sentiment polarity. Results show that while commercial and MoE models perform best overall, all models struggle to balance safety with emotional fidelity in more subtle or context-heavy settings such as emoji, homophone, and dialogue-based inputs. We release ToxiRewriteCN to support future research on controllable, sentiment-aware detoxification for Chinese.

Margaret Hughes, Brandon Roy, Elinor Poole-Dayan, Deb Roy, Jad Kabbara

Growing literature explores toxicity and polarization in discourse, with comparatively less work on characterizing what makes dialogue prosocial and constructive. We explore conversational discourse and investigate a method for characterizing its quality built upon the notion of “responsivity”—whether one person’s conversational turn is responding to a preceding turn. We develop and evaluate methods for quantifying responsivity—first through semantic similarity of speaker turns, and second by leveraging state-of-the-art large language models (LLMs) to identify the relation between two speaker turns. We evaluate both methods against a ground truth set of human-annotated conversations. Furthermore, selecting the better performing LLM-based approach, we characterize the nature of the response—whether it responded to that preceding turn in a substantive way or not. We view these responsivity links as a fundamental aspect of dialogue but note that conversations can exhibit significantly different responsivity structures. Accordingly, we then develop conversation-level derived metrics to address various aspects of conversational discourse. We use these derived metrics to explore other conversations and show that they support meaningful characterizations and differentiations across a diverse collection of conversations.

Hongyu Sun, Yusuke Sakai, Haruki Sakajo, Shintaro Ozaki, Kazuki Hayashi, Hidetaka Kamigaito, Taro Watanabe

Continuous instruction following closely mirrors real-world tasks by requiring models to solve sequences of interdependent steps, yet existing multi-step instruction datasets suffer from three key limitations: (1) lack of logical coherence across turns, (2) narrow topical breadth and depth, and (3) reliance on rigid templates or heavy manual effort. We introduce LoCt-Pipeline, a novel pipeline that leverages modern LLMs’ reasoning capabilities to assemble rich, topic-related single-instruction data into multi-turn dialogues, producing chains that are logically coherent, progressively deepen in content, and span diverse domains without fixed templates or extensive human annotation. We employed this pipeline to construct LoCt-Instruct for assessing models’ problem-solving abilities. The generated chains serve as a testbed for benchmarking a variety of models, including reasoning-oriented architectures, instruction-tuned variants, and state-of-the-art closed-source LLMs on their capacity to follow and correctly respond to each step. Our results reveal a substantial performance gap between current LLMs and human solvers. These findings highlight the need for more robust continuous instruction following. We publicly release the dataset and end-to-end pipeline.

Minki Hong, Jangho Choi, Jihie Kim

Social norms govern culturally appropriate behavior in communication, enabling dialogue systems to produce responses that are not only coherent but also socially acceptable. We present NormGenesis, a multicultural framework for generating and annotating socially grounded dialogues across English, Chinese, and Korean. To model the dynamics of social interaction beyond static norm classification, we propose a novel dialogue type, Violation-to-Resolution (V2R), which models the progression of conversations following norm violations through recognition and socially appropriate repair. To improve pragmatic consistency in underrepresented languages, we implement an exemplar-based iterative refinement early in the dialogue synthesis process. This design introduces alignment with linguistic, emotional, and sociocultural expectations before full dialogue generation begins. Using this framework, we construct a dataset of 10,800 multi-turn dialogues annotated at the turn level for norm adherence, speaker intent, and emotional response. Human and LLM-based evaluations demonstrate that NormGenesis significantly outperforms existing datasets in refinement quality, dialogue naturalness, and generalization performance. We show that models trained on our V2R-augmented data exhibit improved pragmatic competence in ethically sensitive contexts. Our work establishes a new benchmark for culturally adaptive dialogue modeling and provides a scalable methodology for norm-aware generation across linguistically and culturally diverse languages.

Jiazheng Liu, Sipeng Zheng, Börje F. Karlsson, Zongqing Lu

Multimodal large language models (MLLMs), built on large-scale pre-trained vision towers and language models, have shown great capabilities in multimodal understanding. However, most existing MLLMs are trained on single-turn vision question-answering tasks, which do not accurately reflect real-world human conversations. In this paper, we introduce MMDiag, a new large-scale multi-turn multimodal dialogue dataset. This dataset is collaboratively generated through deliberately designed rules and GPT assistance, featuring complex dialogues with contextual dependencies that force models to track, ground, and recall information across multiple turns and disparate visual regions. MMDiag serves as a strong benchmark for multi-turn multimodal dialogue learning and brings more challenges to the grounding and reasoning capabilities of MLLMs. Further, inspired by human vision processing we present DiagNote, equipped with multimodal grounding and reasoning capabilities. DiagNote adopts a novel dual-module architecture that explicitly separates reasoning from grounding: a reasoning module (Deliberate) performs step-by-step Chain-of-Thought, while a grounding module (Gaze) provides precise visual focus by predicting bounding box annotations. These modules interact iteratively, enabling DiagNote to dynamically refine its understanding. We empirically demonstrate the advantages of DiagNote in both grounding and jointly processing and reasoning with vision and language information over existing MLLMs.