Question answering in conversations: Query refinement using contextual and semantic information
作者：
Highlights：
• This paper introduces a query refinement method applied to questions asked by users during a meeting.
• To answer the questions, our method leverages the local context with external semantic resources, either WordNet or word embeddings.
• The method first represents the local context by extracting keywords from the transcript of the conversation.
• The proposed method then expands the queries with keywords that best represent the topic of the query.
• We compare our query expansion approach, showing its superiority when manual or automatic transcripts are used.
摘要
•This paper introduces a query refinement method applied to questions asked by users during a meeting.•To answer the questions, our method leverages the local context with external semantic resources, either WordNet or word embeddings.•The method first represents the local context by extracting keywords from the transcript of the conversation.•The proposed method then expands the queries with keywords that best represent the topic of the query.•We compare our query expansion approach, showing its superiority when manual or automatic transcripts are used.
论文关键词：Query refinement,Query expansion,Context modeling,Speech-based information retrieval,Evaluation of information retrieval
论文评审过程：Received 25 January 2016, Revised 5 June 2016, Accepted 7 June 2016, Available online 14 June 2016, Version of Record 19 November 2016.
论文官网地址：https://doi.org/10.1016/j.datak.2016.06.003
