A Controllable Model of Grounded Response Generation

Current end-to-end neural conversation models inherently lack the flexibility to impose semantic control in the response generation process. This control is essential to ensure that users' semantic intents are satisfied and to impose a degree of specificity on generated outputs. Attempts to boost informativeness alone come at the expense of factual accuracy, as attested by GPT-2's propensity to "hallucinate" facts. While this may be mitigated by access to background knowledge, there is scant guarantee of relevance and informativeness in generated responses. We propose a framework that we call controllable grounded response generation (CGRG), in which lexical control phrases are either provided by an user or automatically extracted by a content planner from dialogue context and grounding knowledge. Quantitative and qualitative results show that, using this framework, a GPT-2 based model trained on a conversation-like Reddit dataset outperforms strong generation baselines.

[1]  Sungjin Lee,et al.  Jointly Optimizing Diversity and Relevance in Neural Response Generation , 2019, NAACL.

[2]  Huda Khayrallah,et al.  Improved Lexically Constrained Decoding for Translation and Monolingual Rewriting , 2019, NAACL.

[3]  Ani Nenkova,et al.  The Feasibility of Embedding Based Automatic Evaluation for Single Document Summarization , 2019, EMNLP.

[4]  Alan W. Black,et al.  A Dataset for Document Grounded Conversations , 2018, EMNLP.

[5]  Xiaodong Liu,et al.  Conversing by Reading: Contentful Neural Conversation with On-demand Machine Reading , 2019, ACL.

[6]  Maxine Eskénazi,et al.  Unsupervised Discrete Sentence Representation Learning for Interpretable Neural Dialog Generation , 2018, ACL.

[7]  Yejin Choi,et al.  The Curious Case of Neural Text Degeneration , 2019, ICLR.

[8]  Jason Weston,et al.  What makes a good conversation? How controllable attributes affect human judgments , 2019, NAACL.

[9]  Mitesh M. Khapra,et al.  Towards Exploiting Background Knowledge for Building Conversation Systems , 2018, EMNLP.

[10]  Zheng-Yu Niu,et al.  Knowledge Aware Conversation Generation with Explainable Reasoning over Augmented Graphs , 2019, EMNLP.

[11]  Yoav Goldberg,et al.  Controlling Linguistic Style Aspects in Neural Language Generation , 2017, ArXiv.

[12]  Eric P. Xing,et al.  Target-Guided Open-Domain Conversation , 2019, ACL.

[13]  Jason Weston,et al.  Wizard of Wikipedia: Knowledge-Powered Conversational agents , 2018, ICLR.

[14]  Lukasz Kaiser,et al.  Attention is All you Need , 2017, NIPS.

[15]  Hang Li,et al.  Neural Responding Machine for Short-Text Conversation , 2015, ACL.

[16]  Jianfeng Gao,et al.  DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation , 2020, ACL.

[17]  Jianfeng Gao,et al.  A Diversity-Promoting Objective Function for Neural Conversation Models , 2015, NAACL.

[18]  Alexander M. Rush,et al.  Challenges in Data-to-Document Generation , 2017, EMNLP.

[19]  Quoc V. Le,et al.  A Neural Conversational Model , 2015, ArXiv.

[20]  George R. Doddington,et al.  Automatic Evaluation of Machine Translation Quality Using N-gram Co-Occurrence Statistics , 2002 .

[21]  Qun Liu,et al.  Lexically Constrained Decoding for Sequence Generation Using Grid Beam Search , 2017, ACL.

[22]  Lei Li,et al.  CGMH: Constrained Sentence Generation by Metropolis-Hastings Sampling , 2018, AAAI.

[23]  Dilek Z. Hakkani-Tür,et al.  DeepCopy: Grounded Response Generation with Hierarchical Pointer Networks , 2019, SIGdial.

[24]  Yang Feng,et al.  Knowledge Diffusion for Neural Dialogue Generation , 2018, ACL.

[25]  Ali Farhadi,et al.  Defending Against Neural Fake News , 2019, NeurIPS.

[26]  Mirella Lapata,et al.  Learning to Generate Product Reviews from Attributes , 2017, EACL.

[27]  Lu Wang,et al.  Sentence-Level Content Planning and Style Specification for Neural Text Generation , 2019, EMNLP.

[28]  Lihong Li,et al.  Neural Approaches to Conversational AI , 2019, Found. Trends Inf. Retr..

[29]  Lav R. Varshney,et al.  CTRL: A Conditional Transformer Language Model for Controllable Generation , 2019, ArXiv.

[30]  Matt Post,et al.  ParaBank: Monolingual Bitext Generation and Sentential Paraphrasing via Lexically-constrained Neural Machine Translation , 2019, AAAI.

[31]  Jianfeng Gao,et al.  A Persona-Based Neural Conversation Model , 2016, ACL.

[32]  Ming-Wei Chang,et al.  A Knowledge-Grounded Neural Conversation Model , 2017, AAAI.

[33]  Erik Cambria,et al.  Augmenting End-to-End Dialogue Systems With Commonsense Knowledge , 2018, AAAI.

[34]  Jianfeng Gao,et al.  A Neural Network Approach to Context-Sensitive Generation of Conversational Responses , 2015, NAACL.

[35]  Yu Zhang,et al.  Flexible End-to-End Dialogue System for Knowledge Grounded Conversation , 2017, ArXiv.

[36]  Ilya Sutskever,et al.  Language Models are Unsupervised Multitask Learners , 2019 .

[37]  Sungjin Lee,et al.  Structuring Latent Spaces for Stylized Response Generation , 2019, EMNLP.

[38]  Salim Roukos,et al.  Bleu: a Method for Automatic Evaluation of Machine Translation , 2002, ACL.

[39]  Eric P. Xing,et al.  Toward Controlled Generation of Text , 2017, ICML.