{"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/complex-sequential-question-answering-towards","title":"Complex Sequential Question Answering: Towards Learning to Converse Over Linked Question Answer Pairs with a Knowledge Graph","arxiv_id":"1801.10314","date":"2018-01-31","proceeding":null,"authors":["Amrita Saha","Vardaan Pahuja","Mitesh M. Khapra","Karthik Sankaranarayanan","Sarath Chandar"],"abstract":"While conversing with chatbots, humans typically tend to ask many questions,\na significant portion of which can be answered by referring to large-scale\nknowledge graphs (KG). While Question Answering (QA) and dialog systems have\nbeen studied independently, there is a need to study them closely to evaluate\nsuch real-world scenarios faced by bots involving both these tasks. Towards\nthis end, we introduce the task of Complex Sequential QA which combines the two\ntasks of (i) answering factual questions through complex inferencing over a\nrealistic-sized KG of millions of entities, and (ii) learning to converse\nthrough a series of coherently linked QA pairs. Through a labor intensive\nsemi-automatic process, involving in-house and crowdsourced workers, we created\na dataset containing around 200K dialogs with a total of 1.6M turns. Further,\nunlike existing large scale QA datasets which contain simple questions that can\nbe answered from a single tuple, the questions in our dialogs require a larger\nsubgraph of the KG. Specifically, our dataset has questions which require\nlogical, quantitative, and comparative reasoning as well as their combinations.\nThis calls for models which can: (i) parse complex natural language questions,\n(ii) use conversation context to resolve coreferences and ellipsis in\nutterances, (iii) ask for clarifications for ambiguous queries, and finally\n(iv) retrieve relevant subgraphs of the KG to answer such questions. However,\nour experiments with a combination of state of the art dialog and QA models\nshow that they clearly do not achieve the above objectives and are inadequate\nfor dealing with such complex real world settings. We believe that this new\ndataset coupled with the limitations of existing models as reported in this\npaper should encourage further research in Complex Sequential QA.","url_abs":"http://arxiv.org/abs/1801.10314v2","url_pdf":"http://arxiv.org/pdf/1801.10314v2.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":"complex-sequential-question-answering-towards","repo_url":"https://github.com/mali-git/CSQA_Implementation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"knowledge-graphs","task_name":"Knowledge Graphs"},{"task_slug":"question-answering","task_name":"Question Answering"}],"methods":[],"datasets_introduced":[{"slug":"csqa","name":"CSQA","full_name":""}],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1801.10314","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}