pith. sign in

arxiv: 2606.12902 · v1 · pith:MGYYA2BZnew · submitted 2026-06-11 · 💻 cs.CL

PRISM: Prosody-Integrated Multi-Agent Reasoning Framework for Empathetic Spoken Dialogue

Wen Zhang , Xiaocui Yang , Zhuoyue Gao , Shi Feng , Daling Wang , Yifei Zhang This is my paper

Pith reviewed 2026-06-27 06:51 UTC · model grok-4.3

classification 💻 cs.CL
keywords empathetic spoken dialoguemulti-agent frameworkprosody-to-language translationspeech perceptionresponse generationspeech synthesisLLM reasoning
0
0 comments X

The pith

PRISM decouples speech perception, response generation and synthesis into agents plus a prosody-to-language step to raise empathy and prosodic fit in spoken dialogue.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents PRISM as a way to build empathetic spoken dialogue systems that keep both semantic content and emotional tone aligned. It splits the work across separate agents for listening to speech, deciding on a reply, and turning the reply into audio, then adds a translation step that turns prosody details into ordinary language so large language models can reason about emotion more reliably. The framework also lets the agents call external knowledge sources when needed. If the approach works, spoken systems would produce replies whose wording and delivery both convey the right feeling instead of losing acoustic cues in transcription or giving up control in fully end-to-end models.

Core claim

PRISM decouples speech perception, response generation, and speech synthesis into coordinated components, introduces a prosody-to-language translation mechanism to stabilize large language model reasoning, and enables on-demand invocation of external knowledge tools, yielding consistent improvements in empathy, prosodic appropriateness, and text response generation quality across objective and subjective metrics.

What carries the argument

The prosody-to-language translation mechanism that converts acoustic cues into textual descriptions inside the multi-agent coordination loop.

If this is right

  • Higher empathy scores on both automatic and human evaluations.
  • More appropriate prosody in the final spoken output.
  • Improved quality of the generated text responses.
  • More stable emotional reasoning by the language model component.
  • On-demand use of external knowledge improves reply relevance.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The modular split could make it easier to swap in new tools or update one stage without retraining everything.
  • Similar translation steps might help other tasks where non-text signals need to reach a language model.
  • The design might reduce training cost compared with large end-to-end speech models by reusing existing components.

Load-bearing premise

Dividing the system into separate agents for perception, generation and synthesis plus translating prosody into language will reliably stabilize reasoning and produce net gains without new coordination failures or error sources.

What would settle it

A controlled comparison in which a single integrated model or a standard cascade pipeline without the translation step matches or exceeds PRISM on all empathy, prosody and response-quality metrics.

Figures

Figures reproduced from arXiv: 2606.12902 by Daling Wang, Shi Feng, Wen Zhang, Xiaocui Yang, Yifei Zhang, Zhuoyue Gao.

Figure 1
Figure 1. Figure 1: Overview of the PRISM Framework. 2.1. Perceiver For a given input speech x, Perceiver outputs a structured state s = {T, a}, where T is the transcription text, a represents a set of paralinguistic attributes designed to capture the speaker’s emotional and expressive state. The paralinguistic attributes a include four categories of cues: (i) affective cues, including the recognized emotion category and its … view at source ↗
Figure 3
Figure 3. Figure 3: LLM-based evaluation results. Only wins and losses are shown; ties are omitted. 3.4.2. Ablation Studies We conducted ablation studies on two key aspects: autonomous knowledge invocation and prosody description. Specifically, we evaluated the results under the following conditions: enforcing knowledge usage at every turn (Always Kno), excluding knowl￾edge entirely (w/o Kno) and removing prosody descriptions… view at source ↗
Figure 2
Figure 2. Figure 2: Human evaluation results. Human Evaluation. We recruited three researchers spe￾cializing in empathetic dialogue systems as annotators. A to￾tal of 100 dialogue samples were randomly selected for evalu￾ation. Each sample was independently rated by all annotators. The evaluation is conducted from two aspects, text quality and speech quality, covering the following six dimensions: Empa￾thy, Informativity, Flu… view at source ↗
read the original abstract

Empathetic spoken dialogue systems require not only semantically appropriate responses but also emotionally aligned prosodic expression. However, cascade pipelines often discard acoustic cues during speech-to-text conversion, while end-to-end speech models lack interpretable control over emotion and knowledge integration. To address these challenges, we propose PRISM, a multi-agent framework for empathetic spoken dialogue that decouples speech perception, response generation, and speech synthesis into coordinated components. PRISM introduces a prosody-to-language translation mechanism to stabilize large language model reasoning and enables on-demand invocation of external knowledge tools for empathetic dialogue generation. Experimental results demonstrate that PRISM achieves consistent improvements in empathy, prosodic appropriateness, and text response generation quality across objective and subjective metrics. Our code is available at: https://github.com/Bxzfrm/PRISM.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The manuscript proposes PRISM, a multi-agent framework for empathetic spoken dialogue that decouples speech perception, response generation, and speech synthesis into coordinated agents. It introduces a prosody-to-language translation mechanism to stabilize LLM reasoning and supports on-demand invocation of external knowledge tools. The authors report that PRISM achieves consistent improvements in empathy, prosodic appropriateness, and text response generation quality across objective and subjective metrics, with code released at a public GitHub repository.

Significance. If the experimental claims hold under rigorous validation, the work could contribute to the field by offering an interpretable multi-agent alternative to cascade and end-to-end spoken dialogue systems, particularly through explicit prosody integration and knowledge tool use. The open code release is a positive factor for reproducibility.

major comments (1)
  1. [Abstract] Abstract: the central claim that PRISM 'achieves consistent improvements' across metrics is stated without any description of experimental design, baselines, metric definitions, statistical tests, datasets, or ablation studies. This absence makes the data-to-claim link impossible to evaluate and is load-bearing for the paper's primary assertion.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their review and the opportunity to respond. We address the single major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that PRISM 'achieves consistent improvements' across metrics is stated without any description of experimental design, baselines, metric definitions, statistical tests, datasets, or ablation studies. This absence makes the data-to-claim link impossible to evaluate and is load-bearing for the paper's primary assertion.

    Authors: The abstract is intentionally concise to summarize the core contribution and key findings within standard length constraints. The experimental design (including datasets, baselines such as cascade and end-to-end spoken dialogue systems, metric definitions for empathy and prosodic fit, statistical significance testing, and ablation studies) is fully detailed in Sections 4 (Experimental Setup) and 5 (Results and Analysis) of the manuscript, with all claims directly supported by those sections. This follows conventional academic structure where abstracts state outcomes and the body provides the rigorous link between data and claims. We are willing to add a single sentence to the abstract briefly referencing the evaluation protocol if the editor deems it necessary for clarity. revision: partial

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The manuscript describes an architectural framework (multi-agent decoupling of perception/generation/synthesis plus a prosody-to-language translation step) and reports empirical improvements on objective/subjective metrics. No equations, derivations, fitted parameters, or self-citation chains appear in the provided text. All central claims are framed as experimental outcomes rather than quantities defined from the inputs by construction. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no free parameters, axioms, or invented entities are described.

pith-pipeline@v0.9.1-grok · 5677 in / 1010 out tokens · 17339 ms · 2026-06-27T06:51:49.384584+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

45 extracted references · 3 canonical work pages

  1. [1]

    ASR-text dialogue model-TTS

    Introduction In recent years, advances in large language models (LLMs) and speech technologies have driven human-machine dialogue sys- tems from text-based interaction toward more natural spoken interaction. Compared with text dialogue, spoken dialogue con- veys not only linguistic content but also prosodic and emotional cues [1, 2]. These paralinguistic ...

  2. [2]

    The speaker spoke quickly with an angry tone

    Method We propose a prosody-aware multi-agent empathetic spoken di- alogue framework consisting of four components: Perceiver, Manager, Responder, and V ocalizer, as illustrated in Figure 1. arXiv:2606.12902v1 [cs.CL] 11 Jun 2026 VocalizerPerceiver Manager Responder Emotion category Verification Expressive intensity Interaction strategy Speech Input Whisp...

    Pith/arXiv arXiv 2026

  3. [3]

    Dataset TOOL-ED[20] is a tool-augmented extension of the ED [21] dataset, designed to study empathetic dialogue generation with external knowledge integration

    Experiments 3.1. Dataset TOOL-ED[20] is a tool-augmented extension of the ED [21] dataset, designed to study empathetic dialogue generation with external knowledge integration. Each sample consists of a short dialogue context and its corresponding empathetic response, along with annotations indicating whether external knowledge should be invoked.AvaMERG[2...

    arXiv 2041

  4. [4]

    Conclusion We present PRISM, a multi-agent framework for empathetic spoken dialogue that integrates prosody-to-language translation and adaptive knowledge invocation. By decoupling perception, reasoning, and synthesis, PRISM enables interpretable emo- tion modeling and controllable speech generation, achieving improved empathy and prosodic alignment over ...

  5. [5]

    62272092, 62172086), and the Fun- damental Research Funds for the Central Universities under Grants (N25XQD004)

    Acknowledgments The work was supported by the National Natural Science Foun- dation of China (Nos. 62272092, 62172086), and the Fun- damental Research Funds for the Central Universities under Grants (N25XQD004). Thanks to the KinaMind society for their inspiring environment and unwavering support

  6. [6]

    All technical contributions, experimental de- sign, and analyses were conducted by the authors

    Generative AI Use Disclosure Generative AI tools were used only for proofreading and lan- guage polishing. All technical contributions, experimental de- sign, and analyses were conducted by the authors

  7. [7]

    The geneva minimalistic acoustic parameter set (gemaps) for voice research and affective computing,

    F. Eyben, K. R. Scherer, B. W. Schuller, J. Sundberg, E. Andr ´e, C. Busso, L. Y . Devillers, J. Epps, P. Laukka, S. S. Narayanan et al., “The geneva minimalistic acoustic parameter set (gemaps) for voice research and affective computing,”IEEE transactions on affective computing, vol. 7, no. 2, pp. 190–202, 2015

    2015

  8. [8]

    The role of prosody in spoken question answering,

    J. Chi, M. de Seyssel, and N. Schluter, “The role of prosody in spoken question answering,” inFindings of the Association for Computational Linguistics: NAACL 2025, 2025, pp. 8468–8479

    2025

  9. [9]

    Emotional chatting machine: emotional conversation generation with inter- nal and external memory,

    H. Zhou, M. Huang, T. Zhang, X. Zhu, and B. Liu, “Emotional chatting machine: emotional conversation generation with inter- nal and external memory,” inProceedings of the Thirty-Second AAAI Conference on Artificial Intelligence and Thirtieth Innova- tive Applications of Artificial Intelligence Conference and Eighth AAAI Symposium on Educational Advances i...

    2018

  10. [10]

    Speech emotion recognition: two decades in a nutshell, benchmarks, and ongoing trends,

    B. W. Schuller, “Speech emotion recognition: two decades in a nutshell, benchmarks, and ongoing trends,”Commun. ACM, vol. 61, no. 5, p. 90–99, Apr. 2018. [Online]. Available: https://doi.org/10.1145/3129340

    work page doi:10.1145/3129340 2018

  11. [11]

    AnnaAgent: Dynamic evolution agent system with multi-session memory f or realistic seeker simulation,

    M. Wang, P. Wang, L. Wu, X. Yang, D. Wang, S. Feng, Y . Chen, B. Wang, and Y . Zhang, “AnnaAgent: Dynamic evolution agent system with multi-session memory f or realistic seeker simulation,” inFindings of the Association for Computational Linguistics: ACL 2025, W. Che, J. Nabende, E. Shutova, and M. T. Pilehvar, Eds. Vienna, Austria: Association for Comput...

    2025

  12. [12]

    Audiogpt: Understanding and generating speech, music, sound, and talking head,

    R. Huang, M. Li, D. Yang, J. Shi, X. Chang, Z. Ye, Y . Wu, Z. Hong, J. Huang, J. Liuet al., “Audiogpt: Understanding and generating speech, music, sound, and talking head,” inProceed- ings of the AAAI Conference on Artificial Intelligence, vol. 38, no. 21, 2024, pp. 23 802–23 804

    2024

  13. [13]

    Towards end-to-end spoken language understanding,

    D. Serdyuk, Y . Wang, C. Fuegen, A. Kumar, B. Liu, and Y . Ben- gio, “Towards end-to-end spoken language understanding,” in 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2018, pp. 5754–5758

    2018

  14. [14]

    Wavchat: A survey of spoken dialogue models,

    S. Ji, Y . Chen, M. Fang, J. Zuo, J. Lu, H. Wang, Z. Jiang, L. Zhou, S. Liu, X. Chenget al., “Wavchat: A survey of spoken dialogue models,”arXiv preprint arXiv:2411.13577, 2024

    arXiv 2024

  15. [15]

    Eyben, M

    F. Eyben, M. W ¨ollmer, and B. Schuller, “Opensmile: the munich versatile and fast open-source audio feature extractor,” inProceedings of the 18th ACM International Conference on Multimedia, ser. MM ’10. New York, NY , USA: Association for Computing Machinery, 2010, p. 1459–1462. [Online]. Available: https://doi.org/10.1145/1873951.1874246

    work page doi:10.1145/1873951.1874246 2010

  16. [16]

    Audiolm: A language modeling approach to audio generation,

    Z. Borsos, R. Marinier, D. Vincent, E. Kharitonov, O. Pietquin, M. Sharifi, D. Roblek, O. Teboul, D. Grangier, M. Tagliasacchi, and N. Zeghidour, “Audiolm: A language modeling approach to audio generation,”IEEE/ACM Trans. Audio, Speech and Lang. Proc., vol. 31, p. 2523–2533, Jun. 2023. [Online]. Available: https://doi.org/10.1109/TASLP.2023.3288409

    work page doi:10.1109/taslp.2023.3288409 2023

  17. [17]

    SpeechGPT: Empowering large language models with intrinsic cross-modal conversational abilities,

    D. Zhang, S. Li, X. Zhang, J. Zhan, P. Wang, Y . Zhou, and X. Qiu, “SpeechGPT: Empowering large language models with intrinsic cross-modal conversational abilities,” inFindings of the Association for Computational Linguistics: EMNLP 2023, H. Bouamor, J. Pino, and K. Bali, Eds. Singapore: Association for Computational Linguistics, Dec. 2023, pp. 15 757–15 ...

    2023

  18. [18]

    Neural codec language models are zero-shot text to speech synthesizers,

    S. Chen, C. Wang, Y . Wu, Z. Zhang, L. Zhou, S. Liu, Z. Chen, Y . Liu, H. Wang, J. Li, L. He, S. Zhao, and F. Wei, “Neural codec language models are zero-shot text to speech synthesizers,”IEEE Transactions on Audio, Speech and Language Processing, vol. 33, pp. 705–718, 2025

    2025

  19. [19]

    Audiopalm: A large language model that can speak and listen,

    P. K. Rubenstein, C. Asawaroengchai, D. D. Nguyen, A. Bapna, Z. Borsos, F. de Chaumont Quitry, P. Chen, D. E. Badawy, W. Han, E. Kharitonov, H. Muckenhirn, D. R. Padfield, J. Qin, D. Rozenberg, T. N. Sainath, J. Schalkwyk, M. Sharifi, M. D. Tadmor, Ramanovich, M. Tagliasacchi, A. Tudor, M. Velimirovi’c, D. Vincent, J. Yu, Y . Wang, V . Zayats, N. Zeghidou...

    Pith/arXiv arXiv 2023

  20. [20]

    Retrieval-augmented generation for knowledge- intensive nlp tasks,

    P. Lewis, E. Perez, A. Piktus, F. Petroni, V . Karpukhin, N. Goyal, H. K ¨uttler, M. Lewis, W.-t. Yih, T. Rockt ¨aschel, S. Riedel, and D. Kiela, “Retrieval-augmented generation for knowledge- intensive nlp tasks,” inProceedings of the 34th International Con- ference on Neural Information Processing Systems, ser. NIPS ’20. Red Hook, NY , USA: Curran Assoc...

    2020

  21. [21]

    React: Synergizing reasoning and acting in language models,

    S. Yao, J. Zhao, D. Yu, N. Du, I. Shafran, K. Narasimhan, and Y . Cao, “React: Synergizing reasoning and acting in language models,”ArXiv, vol. abs/2210.03629, 2022. [Online]. Available: https://api.semanticscholar.org/CorpusID:252762395

    Pith/arXiv arXiv 2022

  22. [22]

    Robust speech recognition via large-scale weak su- pervision,

    A. Radford, J. W. Kim, T. Xu, G. Brockman, C. McLeavey, and I. Sutskever, “Robust speech recognition via large-scale weak su- pervision,” inProceedings of the 40th International Conference on Machine Learning, ser. ICML’23. JMLR.org, 2023

    2023

  23. [23]

    emotion2vec: Self-supervised pre-training for speech emotion representation,

    Z. Ma, Z. Zheng, J. Ye, J. Li, Z. Gao, S. Zhang, and X. Chen, “emotion2vec: Self-supervised pre-training for speech emotion representation,” inFindings of the Association for Computational Linguistics: ACL 2024, L.-W. Ku, A. Martins, and V . Srikumar, Eds. Bangkok, Thailand: Association for Computational Linguistics, Aug. 2024, pp. 15 747–15 760. [Online]...

    2024

  24. [24]

    Langgpt: Rethinking structured reusable prompt design framework for llms from the programming language,

    M. Wang, Y . Liu, X. Liang, S. Li, Y . Huang, X. Zhang, S. Shen, C. Guan, D. Wang, S. hi Feng, H. Zhang, Y . Zhang, M. Zheng, and C. Zhang, “Langgpt: Rethinking structured reusable prompt design framework for llms from the programming language,”

  25. [25]

    Available: https://arxiv.org/abs/2402.16929

    [Online]. Available: https://arxiv.org/abs/2402.16929

    arXiv

  26. [26]

    Styletts 2: towards human-level text-to-speech through style dif- fusion and adversarial training with large speech language mod- els,

    Y . A. Li, C. Han, V . S. Raghavan, G. Mischler, and N. Mesgarani, “Styletts 2: towards human-level text-to-speech through style dif- fusion and adversarial training with large speech language mod- els,” inProceedings of the 37th International Conference on Neu- ral Information Processing Systems, ser. NIPS ’23. Red Hook, NY , USA: Curran Associates Inc., 2023

    2023

  27. [27]

    TOOL-ED: Enhancing empathetic response generation with the tool calling capability of LLM,

    H. Cao, Y . Zhang, S. Feng, X. Yang, D. Wang, and Y . Zhang, “TOOL-ED: Enhancing empathetic response generation with the tool calling capability of LLM,” inProceedings of the 31st International Conference on Computational Linguistics, O. Rambow, L. Wanner, M. Apidianaki, H. Al-Khalifa, B. D. Eugenio, and S. Schockaert, Eds. Abu Dhabi, UAE: Association for...

    2025

  28. [28]

    Towards empathetic open-domain conversation models: A new benchmark and dataset,

    H. Rashkin, E. M. Smith, M. Li, and Y .-L. Boureau, “Towards empathetic open-domain conversation models: A new benchmark and dataset,” inProceedings of the 57th Annual Meeting of the Association for Computational Linguistics, A. Korhonen, D. Traum, and L. M`arquez, Eds. Florence, Italy: Association for Computational Linguistics, Jul. 2019, pp. 5370–5381. ...

    2019

  29. [29]

    Towards multimodal empathetic response generation: A rich text-speech-vision avatar-based benchmark,

    H. Zhang, Z. Meng, M. Luo, H. Han, L. Liao, E. Cambria, and H. Fei, “Towards multimodal empathetic response generation: A rich text-speech-vision avatar-based benchmark,” inProceedings of the ACM on Web Conference 2025, ser. WWW ’25. New York, NY , USA: Association for Computing Machinery, 2025, p. 2872–2881. [Online]. Available: https://doi.org/10.1145/ ...

    arXiv 2025

  30. [30]

    SALMONN: Towards generic hearing abilities for large language models,

    C. Tang, W. Yu, G. Sun, X. Chen, T. Tan, W. Li, L. Lu, Z. MA, and C. Zhang, “SALMONN: Towards generic hearing abilities for large language models,” inThe Twelfth International Conference on Learning Representations, 2024. [Online]. Available: https://openreview.net/forum?id=14rn7HpKVk

    2024

  31. [31]

    Osum-echat: Enhancing end-to-end empathetic spoken chatbot via understanding-driven spoken dialogue,

    X. Geng, Q. Shao, H. Xue, S. Wang, H. Xie, Z. Guo, Y . Zhao, G. Li, W. Tian, C. Wang, Z. Zhao, K. Xia, Z. Zhang, Z. Lin, T. Zuo, M. Shao, Y . Cao, G. Ma, L. Li, Y . Dai, D. Gao, D. Guo, and L. Xie, “Osum-echat: Enhancing end-to-end empathetic spoken chatbot via understanding-driven spoken dialogue,” 2025. [Online]. Available: https://arxiv.org/abs/2508.09600

    arXiv 2025

  32. [32]

    Qwen2.5-omni technical report,

    J. Xu, Z. Guo, J. He, H. Hu, T. He, S. Bai, K. Chen, J. Wang, Y . Fan, K. Dang, B. Zhang, X. Wang, Y . Chu, and J. Lin, “Qwen2.5-omni technical report,” 2025. [Online]. Available: https://arxiv.org/abs/2503.20215

    Pith/arXiv arXiv 2025

  33. [33]

    LLaMA-omni 2: LLM-based real-time spoken chatbot with autoregressive streaming speech synthesis,

    Q. Fang, Y . Zhou, S. Guo, S. Zhang, and Y . Feng, “LLaMA-omni 2: LLM-based real-time spoken chatbot with autoregressive streaming speech synthesis,” inProceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), W. Che, J. Nabende, E. Shutova, and M. T. Pilehvar, Eds. Vienna, Austria: Association for Co...

    2025

  34. [34]

    OpenS2S: Advancing fully open-source end-to-end empathetic large speech language model,

    C. Wang, T. Peng, W. Yang, Y . Bai, G. Wang, J. Lin, L. Jia, L. Wu, J. Wang, C. Zong, and J. Zhang, “OpenS2S: Advancing fully open-source end-to-end empathetic large speech language model,” inProceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, I. Habernal, P. Schulam, and J. Tiedemann, Eds. Suzhou...

    2025

  35. [35]

    Qwen2.5 technical report,

    Qwenet al., “Qwen2.5 technical report,” 2025. [Online]. Available: https://arxiv.org/abs/2412.15115

    Pith/arXiv arXiv 2025

  36. [36]

    The llama 3 herd of models,

    A. Grattafiori, A. Dubey, A. Jauhriet al., “The llama 3 herd of models,” 2024. [Online]. Available: https://arxiv.org/abs/2407. 21783

    2024

  37. [37]

    LlamaFactory: Unified efficient fine-tuning of 100+ language models,

    Y . Zheng, R. Zhang, J. Zhang, Y . Ye, and Z. Luo, “LlamaFactory: Unified efficient fine-tuning of 100+ language models,” inProceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations), Y . Cao, Y . Feng, and D. Xiong, Eds. Bangkok, Thailand: Association for Computational Linguistics, Aug. 2024,...

    2024

  38. [38]

    COMET: Commonsense transformers for automatic knowledge graph construction,

    A. Bosselut, H. Rashkin, M. Sap, C. Malaviya, A. Celikyilmaz, and Y . Choi, “COMET: Commonsense transformers for automatic knowledge graph construction,” inProceedings of the 57th Annual Meeting of the Association for Computational Linguistics, A. Korhonen, D. Traum, and L. M `arquez, Eds. Florence, Italy: Association for Computational Linguistics, Jul. 2...

    2019

  39. [39]

    Available: https://aclanthology.org/P19-1470/

    [Online]. Available: https://aclanthology.org/P19-1470/

  40. [40]

    Bleu: a method for automatic evaluation of machine translation,

    K. Papineni, S. Roukos, T. Ward, and W.-J. Zhu, “Bleu: a method for automatic evaluation of machine translation,” in Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics, P. Isabelle, E. Charniak, and D. Lin, Eds. Philadelphia, Pennsylvania, USA: Association for Computational Linguistics, Jul. 2002, pp. 311–318. [Online]...

    2002

  41. [41]

    Bertscore: Evaluating text generation with bert,

    T. Zhang, V . Kishore, F. Wu, K. Q. Weinberger, and Y . Artzi, “Bertscore: Evaluating text generation with bert,”arXiv preprint arXiv:1904.09675, 2019

    Pith/arXiv arXiv 1904

  42. [42]

    ROUGE: A package for automatic evaluation of summaries,

    C.-Y . Lin, “ROUGE: A package for automatic evaluation of summaries,” inText Summarization Branches Out. Barcelona, Spain: Association for Computational Linguistics, Jul. 2004, pp. 74–81. [Online]. Available: https://aclanthology.org/W04-1013/

    2004

  43. [43]

    A diversity- promoting objective function for neural conversation models,

    J. Li, M. Galley, C. Brockett, J. Gao, and B. Dolan, “A diversity- promoting objective function for neural conversation models,” inProceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, K. Knight, A. Nenkova, and O. Rambow, Eds. San Diego, California: Association for ...

    2016

  44. [44]

    A technique for the measurement of attitudes

    R. Likert, “A technique for the measurement of attitudes.” Archives of psychology, 1932

    1932

  45. [45]

    Cem: Commonsense-aware empathetic response generation,

    S. Sabour, C. Zheng, and M. Huang, “Cem: Commonsense-aware empathetic response generation,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 36, no. 10, 2022, pp. 11 229–11 237

    2022