{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/topic-aware-neural-response-generation","title":"Topic Aware Neural Response Generation","arxiv_id":"1606.08340","date":"2016-06-21","proceeding":null,"authors":["Chen Xing","Wei Wu","Yu Wu","Jie Liu","YaLou Huang","Ming Zhou","Wei-Ying Ma"],"abstract":"We consider incorporating topic information into the sequence-to-sequence\nframework to generate informative and interesting responses for chatbots. To\nthis end, we propose a topic aware sequence-to-sequence (TA-Seq2Seq) model. The\nmodel utilizes topics to simulate prior knowledge of human that guides them to\nform informative and interesting responses in conversation, and leverages the\ntopic information in generation by a joint attention mechanism and a biased\ngeneration probability. The joint attention mechanism summarizes the hidden\nvectors of an input message as context vectors by message attention,\nsynthesizes topic vectors by topic attention from the topic words of the\nmessage obtained from a pre-trained LDA model, and let these vectors jointly\naffect the generation of words in decoding. To increase the possibility of\ntopic words appearing in responses, the model modifies the generation\nprobability of topic words by adding an extra probability item to bias the\noverall distribution. Empirical study on both automatic evaluation metrics and\nhuman annotations shows that TA-Seq2Seq can generate more informative and\ninteresting responses, and significantly outperform the-state-of-the-art\nresponse generation models.","url_abs":"http://arxiv.org/abs/1606.08340v2","url_pdf":"http://arxiv.org/pdf/1606.08340v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"topic-aware-neural-response-generation","repo_url":"https://github.com/nouhadziri/THRED","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"response-generation","task_name":"Response Generation"}],"methods":[{"method_slug":"lda","method_name":"LDA"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1606.08340","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}