{"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/modeling-multi-turn-conversation-with-deep","title":"Modeling Multi-turn Conversation with Deep Utterance Aggregation","arxiv_id":"1806.09102","date":"2018-06-24","proceeding":"COLING 2018 8","authors":["Zhuosheng Zhang","Jiangtong Li","Pengfei Zhu","Hai Zhao","Gongshen Liu"],"abstract":"Multi-turn conversation understanding is a major challenge for building\nintelligent dialogue systems. This work focuses on retrieval-based response\nmatching for multi-turn conversation whose related work simply concatenates the\nconversation utterances, ignoring the interactions among previous utterances\nfor context modeling. In this paper, we formulate previous utterances into\ncontext using a proposed deep utterance aggregation model to form a\nfine-grained context representation. In detail, a self-matching attention is\nfirst introduced to route the vital information in each utterance. Then the\nmodel matches a response with each refined utterance and the final matching\nscore is obtained after attentive turns aggregation. Experimental results show\nour model outperforms the state-of-the-art methods on three multi-turn\nconversation benchmarks, including a newly introduced e-commerce dialogue\ncorpus.","url_abs":"http://arxiv.org/abs/1806.09102v2","url_pdf":"http://arxiv.org/pdf/1806.09102v2.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":"modeling-multi-turn-conversation-with-deep","repo_url":"https://github.com/cooelf/DeepUtteranceAggregation","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"conversational-response-selection","task_name":"Conversational Response Selection"},{"task_slug":"retrieval","task_name":"Retrieval"}],"methods":[],"datasets_introduced":[{"slug":"e-commerce-1","name":"E-commerce","full_name":"E-commerce"}],"methods_introduced":[],"results":[{"leaderboard":"/sota/conversational-response-selection-on-douban-1","task":"Conversational Response Selection","dataset":"Douban","model":"DUA","rank_in_archive_order":14,"of":16,"metrics":{"MAP":"0.551","MRR":"0.599","P@1":"0.421","R10@1":"0.243","R10@2":"0.421","R10@5":"0.780"},"uses_additional_data":false},{"leaderboard":"/sota/conversational-response-selection-on-e","task":"Conversational Response Selection","dataset":"E-commerce","model":"DUA","rank_in_archive_order":14,"of":15,"metrics":{"R10@1":"0.501","R10@2":"0.700","R10@5":"0.921"},"uses_additional_data":false},{"leaderboard":"/sota/conversational-response-selection-on-ubuntu-1","task":"Conversational Response Selection","dataset":"Ubuntu Dialogue (v1, Ranking)","model":"DUA","rank_in_archive_order":21,"of":25,"metrics":{"R10@1":"0.752","R10@2":"0.868","R10@5":"0.962"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.09102","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}