对话场景下的情感引导问题生成模型

doi:10.11896/jsjkx.231000002

摘要/Abstract

摘要： 人机对话系统已在多种智能服务场景中得到广泛应用。当前的人机对话系统可以感知对话者的情感,并根据上下文给出具备特定情感的响应。但是,具备特定情感的响应难以确保能够有效地引导人们产生特定的情感,例如,一个具备“高兴”情感的响应并不能保证人们产生高兴的情感。在一些场景中,人机对话系统需要引导用户达到某种特定的情感状态,以利于对话的持续开展或提升交互效率,如对话心理陪护或在线智能教学。当前的人机对话系统仅针对“积极/消极”等粗粒度情感引导进行了探索,难以应对细粒度情感引导任务。同时,针对对话的心理研究指出,“问题”会显著影响对话方情感的走向。基于上述背景,提出了一种对话场景下的情感引导问题生成模型。该模型基于GPT预训练模型,将需要引导对话方产生的情感作为情感知识引入模型的响应生成过程之中,同时引入了上下文情感感知机制和常识知识融合机制,并采用多任务学习的方法增强了模型的情感感知能力和对话响应生成能力。鉴于这是首次提出面向细粒度情感引导的问题生成任务,因此构建了情感引导数据集用于训练和实验,并且提出了基于提示学习的自动评价方法。最终,自动评价和人工评价的结果表明,所提模型能有效地生成问题,以引导对话方产生特定的情感。

关键词: 情感引导, 问题生成, 情感对话, 提示学习, 多任务学习

Abstract: Human-machine dialog systems have been widely used in intelligent services.Existing human-machine dialog systems can perceive the interlocutor’s emotional state and give a response with an appropriate emotion based on context.However,it is difficult to ensure that a response with a specific emotion can elicit the same emotion from people.For example,a response with a “joy” emotion does not guarantee that people will experience a “joy” emotion.In some scenarios,human-machine dialogue systems need to guide users to reach a specific emotional state to facilitate the continuous development of a conversation or improve inter-action efficiency,such as dialogue psychological escort or online intelligent teaching.Current human-computer dialogue systems focus on coarse-grained emotion eliciting,such as “positive/negative”,and therefore are difficult to handle fine-grained emotion eliciting.On the other side,research on dialogue psychology indicates that “questions” in a conversation can significantly affect the emotions of interlocutors.Based on the above background,a question-generation model for emotional elicitation in dialogue scenarios is proposed.This model is based on the GPT pre-trained model and incorporates the knowledge of the emotion to be elicited into the response generation.The model also introduces a contextual emotional perception mechanism and a common sense knowledge fusion mechanism and uses multi-task learning to enhance the emotion perception ability and conversation response generation ability.Given that it is the first time to propose a question generation task for fine-grained emotion eliciting,an emotional eliciting dataset has been constructed for training and experiments.An automatic evaluation method based on prompt lear-ning has been designed.Finally,automatic evaluation and human evaluation demonstrate that the proposed model can generate questions that can effectively elicit target emotions.

Key words: Emotion eliciting, Question generation, Emotional dialogue, Prompt learning, Multi-task learning

中图分类号:

TP181

胥备, 许鹏. 对话场景下的情感引导问题生成模型[J]. 计算机科学, 2024, 51(11): 265-272. https://doi.org/10.11896/jsjkx.231000002

XU Bei, XU Peng. Emotion Elicited Question Generation Model in Dialogue Scenarios[J]. Computer Science, 2024, 51(11): 265-272. https://doi.org/10.11896/jsjkx.231000002

参考文献

[1] TURNING A M.Computing machinery and intelligence[M].Springer Netherlands,2009:23-65.
[2] PICARD R W.Affective Computing[J].User Modeling andUser-Adapted Interaction,2002,12(1):85-89.
[3] CHRISTENSEN H,GRIFFITHS K M,KORTEN A E,et al.A comparison of changes in anxiety and depression symptoms of spontaneous users and trial participants of a cognitive behavior therapy website[J].Journal of Medical Internet Research,2004,6(4):e46-e46.
[4] PRENDINGER H,MORI J,ISHIZUKA M.Using human phy-siology to evaluate subtle expressivity of a virtual quizmaster in a mathematical game[J].International journal of human-computer studies,2005,62(2):231-245.
[5] GHOSH S,CHOLLET M,LAKSANA E,et al.Affect-LM:ANeural Language Model for Customizable Affective Text Gene-ration[C]//Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics(Volume 1:Long Papers).2017:634-642.
[6] ZHOU H,HUANG M L,ZHANG T,et al.Emotional chatting machine:emotional conversation generation with internal and external memory[C]//Proceedings of the Thirty-Second AAAI Conference on Artificial Intelligence and Thirtieth Innovative Applications of Artificial Intelligence Conference and Eighth AAAI Symposium on Educational Advances in Artificial Intelligence.2018:730-738.
[7] SABOUR S.,ZHENG C J,HUANG M L.Cem:Commonsense-aware empathetic response generation[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2022,36(10):11229-11237.
[8] HWANG J D,BHAGAVATULA C,LEBRAS R,et al.Comet-atomic 2020:on symbolic and neural commonsense knowledge graphs[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2021,35(7):6384-6392.
[9] ZHOU J,ZHENG C,WANG B,et al.CASE:Aligning Coarse-to-Fine Cognition and Affection for Empathetic Response Ge-neration[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics(Volume 1:Long Papers).2023:8223-8237.
[10] LUBIS N,SAKTI S,YOSHINO K,et al.Eliciting positive emotion through affect-sensitive dialogue response generation:a neural network approach[C]//Proceedings of the Thirty-Second AAAI Conference on Artificial Intelligence and Thirtieth Innovative Applications of Artificial Intelligence Conference and Eighth AAAI Symposium on Educational Advances in Artificial Intelligence.2018:5293-5300.
[11] WANG S,XU X,WU W,et al.Towards Multi-Turn Empathetic Dialogs with Positive Emotion Elicitation[J].arXiv:2204.10509,2022.
[12] JIANG H,ZHU Y,ZHANG X,et al.Emotion Eliciting Ma-chine:Emotion Eliciting Conversation Generation based on Dual Generator[J].arXiv:2105.08251,2021.
[13] GONG Z,MIN Q,ZHANG Y.Eliciting Rich Positive Emotions in Dialogue Generation[C]//Proceedings of the First Workshop on Social Influence in Conversations(SICon 2023).2023:1-8.
[14] ZHOU J F,CHEN Z,WANG B,et al.Facilitating Multi-turnEmotional Support Conversation with Positive Emotion Elicitation:A Reinforcement Learning Approach[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics(Volume 1:Long Papers).2023:1714-1729.
[15] HUANG K,YEOMANS M,BROOKS A W,et al.It doesn’thurt to ask:Question-asking increases liking[J].Journal of Personality and Social Psychology,2017,113(3):430-452.
[16] MCEVOY P,PLANT R.Dementia care:using empathic curiosity to establish the common ground that is necessary for mea-ningful communication[J].Journal of Psychiatric and Mental Health Nursing,2014,21(6):477-482.
[17] SVIKHNUSHINA E,VOINEA I,WELIVITA A,et al.A taxo-nomy of empathetic questions in social dialogs[C]//Proceedings of the 60th Annual Meeting of the Association for ComputationalLinguistics(Volume 1:Long Papers).2022:2952-2973.
[18] WEIZENBAUM J.ELIZA—a computer program for the study of natural language communication between man and machine[J].Communications of the ACM,1966,9(1):36-45.
[19] COLBY K M,WEBER S,HILF F D.Artificial paranoia[J].Artificial Intelligence,1971,2(1):1-25.
[20] THORAT S A,JADHAV V.A review on implementation issuesof rule-based chatbot systems[C]//Proceedings of the International Conference on Innovative Computing & Communications(ICICC).2020.
[21] YAN R,LI J,YU Z.Deep learning for dialogue systems:Chit-chat and beyond[J].Foundations and Trends© in Information Retrieval,2022,15(5):417-589.
[22] LIU S,CHEN H,REN Z,et al.Knowledge diffusion for neural dialogue generation[C]//Proceedings of the 56th Annual Mee-ting of the Association for Computational Linguistics(Volume 1:Long Papers).2018:1489-1498.
[23] ZHOU H,YOUNG T,HUANG M L,et al.Commonsenseknowledge aware conversation generation with graph attention[C]//IJCAI.2018:4623-4629.
[24] SONG H,ZHANG W N,HU J,et al.Generating persona consistent dialogues by exploiting natural language inference[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2020:8878-8885.
[25] LI J Y,SUN X.A Syntactically Constrained Bidirectional-Asynchronous Approach for Emotional Conversation Generation[C]//Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing.2018:678-683.
[26] HUANG C Y,ZAIANE O R,TRABELSI A,et al.Automaticdialogue generation with expressed emotions[C]//Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics:Human Language Technologies,Volume 2(Short Papers).2018:49-54.
[27] MA Z Q,YANG R,DU B,et al.A control unit for emotionalconversation generation[J].IEEE Access,2020,8:43168-43176.
[28] GAO Y,BING L,CHEN W,et al.Difficulty controllable generation of reading comprehension questions[J].arXiv:1807.03586,2018.
[29] FEI Z C,ZHANG Q,GUI T,et al.CQG:A simple and effective controlled generation framework for multi-hop question generation[C]//Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics(Volume 1:Long Papers).2022:6896-6906.
[30] WANG Y S,LIU C,HUANG M L,et al.Learning to Ask Questions in Open-domain Conversational Systems with Typed Decoders[C]//Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics(Volume 1:Long Papers).2018:2193-2203.
[31] GAO Y F,LI P,KING I,et al.Interconnected Question Generation with Coreference Alignment and Conversation Flow Mode-ling[C]//Proceedings of the 57th Annual Meeting of the Asso-ciation for Computational Linguistics.2019:4853-4862.
[32] LI Y R,SU H,SHEN X,et al.DailyDialog:A Manually Labelled Multi-turn Dialogue Dataset[C]//Proceedings of the Eighth International Joint Conference on Natural Language Processing(Volume 1:Long Papers).2017:986-995.
[33] RASHKIN H,SMITH E M,LI M,et al.Towards Empathetic Open-domain Conversation Models:A New Benchmark and Dataset[C]//Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics.2019:5370-5381.
[34] DANESCU-NICULESCU-MIZIL C,LEE L.Chameleons inImagined Conversations:A New Approach to Understanding Coordination of Linguistic Style in Dialogs[C]//Proceedings of the 2nd Workshop on Cognitive Modeling and Computational Linguistics.2011:76-87.
[35] EKMAN P.Emotion in the human face:guidelines for research and an integration of findings[M].Pergamon Press,1972.
[36] DEVLIN J,CHANG M W,LEE K,et al.BERT:Pre-training of Deep Bidirectional Transformers for Language Understanding[C]//Proceedings of NAACL-HLT.2019:4171-4186.
[37] LIU Y,OTT M,GOYAL N,et al.Roberta:A robustly opti-mized bert pretraining approach[J].arXiv:1907.11692,2019.
[38] ZHANG Y Z,SUN S,GALLEY M,et al.DIALOGPT:Large-Scale Generative Pre-training for Conversational.Response Ge-neration[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics:System Demonstrations.2020:270-278.
[39] WANG Y D,KE P,ZHENG Y,et al.A large-scale chineseshort-text conversation dataset[C]//Natural Language Proces-sing and Chinese Computing:9th CCF International Conference.Springer International Publishing,2020:91-103.
[40] RADFORD A,WU J,CHILD R,et al.Language models are unsupervised multitask learners[J].OpenAI blog,2019,1(8):9.
[41] LIU X D,HE P,CHEN W,et al.Multi-Task Deep Neural Networks for Natural Language Understanding[C]//Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics.2019:4487-4496.
[42] XU W R,GU X,CHEN G.Generating emotional controllableresponse based on multi-task and dual attention framework[J].IEEE Access,2019,7:93734-93741.
[43] LIU P F,YUAN W Z,FU J L,et al.Pre-train,prompt,and predict:A systematic survey of prompting methods in natural language processing[J].ACM Computing Surveys,2023,55(9):1-35.
[44] BÄCKMAN L,DIXON R A.Psychological compensation:a theo-retical framework[J].Psychological Bulletin,1992,112(2):259-283.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed