Preprocessing of normative documents for interactive question answering
作者：
Highlights：
• Current question answering systems do not perform well on normative documents.
• Our method exploits the specific structure of the normative documents.
• The performance is significantly better, even without retraining a language model.
• Interactive disambiguation of user’s questions further increases the precision.
• A novel dataset for normative documents is presented.
摘要
•Current question answering systems do not perform well on normative documents.•Our method exploits the specific structure of the normative documents.•The performance is significantly better, even without retraining a language model.•Interactive disambiguation of user’s questions further increases the precision.•A novel dataset for normative documents is presented.
论文关键词：Interactive question answering,Conversational agent,Normative documents,Information retrieval,Document graph,Document structure
论文评审过程：Received 24 January 2021, Revised 23 November 2021, Accepted 25 November 2021, Available online 2 December 2021, Version of Record 8 December 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.116314
