{"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/attentive-memory-networks-efficient-machine","title":"Attentive Memory Networks: Efficient Machine Reading for Conversational Search","arxiv_id":"1712.07229","date":"2017-12-19","proceeding":null,"authors":["Tom Kenter","Maarten de Rijke"],"abstract":"Recent advances in conversational systems have changed the search paradigm.\nTraditionally, a user poses a query to a search engine that returns an answer\nbased on its index, possibly leveraging external knowledge bases and\nconditioning the response on earlier interactions in the search session. In a\nnatural conversation, there is an additional source of information to take into\naccount: utterances produced earlier in a conversation can also be referred to\nand a conversational IR system has to keep track of information conveyed by the\nuser during the conversation, even if it is implicit.\n  We argue that the process of building a representation of the conversation\ncan be framed as a machine reading task, where an automated system is presented\nwith a number of statements about which it should answer questions. The\nquestions should be answered solely by referring to the statements provided,\nwithout consulting external knowledge. The time is right for the information\nretrieval community to embrace this task, both as a stand-alone task and\nintegrated in a broader conversational search setting.\n  In this paper, we focus on machine reading as a stand-alone task and present\nthe Attentive Memory Network (AMN), an end-to-end trainable machine reading\nalgorithm. Its key contribution is in efficiency, achieved by having an\nhierarchical input encoder, iterating over the input only once. Speed is an\nimportant requirement in the setting of conversational search, as gaps between\nconversational turns have a detrimental effect on naturalness. On 20 datasets\ncommonly used for evaluating machine reading algorithms we show that the AMN\nachieves performance comparable to the state-of-the-art models, while using\nconsiderably fewer computations.","url_abs":"http://arxiv.org/abs/1712.07229v1","url_pdf":"http://arxiv.org/pdf/1712.07229v1.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":"attentive-memory-networks-efficient-machine","repo_url":"https://bitbucket.org/TomKenter/attentive-memory-networks-code","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"conversational-search","task_name":"Conversational Search"},{"task_slug":"information-retrieval","task_name":"Information Retrieval"},{"task_slug":"reading-comprehension","task_name":"Reading Comprehension"},{"task_slug":"retrieval","task_name":"Retrieval"}],"methods":[{"method_slug":"memory-network","method_name":"Memory Network"},{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1712.07229","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}