关键词: 问答系统, 政府网站, 语义理解能力, 智能问答, 测评问题, 智能客服

Abstract: [Purpose/Significance] Intelligent question and answer（Q&A）system is becoming an important facility for public services delivery. The purpose of this study is to promote the construction of government intelligent Q&A systems by evaluation, and reduce the administrative burden of citizens. [Design/Methodology] Firstly, based on literature research, theoretical analysis and case study, an evaluation indicator system for government intelligent Q&A system was constructed, and the indicator weight was determined by expert survey and AHP method. Secondly, a testing question set for problem solving quality was developed based on the hot consultation issues. Finally, 30 provincial government Q&A systems in China were evaluated with the indicator system and the matched questions. [Findings/Conclusion] The evaluation index system of government intelligent question and answering system includes three dimensions, namely, the quality of problem solving, the quality of service interaction, and the quality of infrastructure construction, as well as 18 refined indicators, such as the rate of high-frequency question answering, the rate of synonymous question answering, and the number of interactions. The testing question set includes five categories of questions: high-frequency, synonymous, error, ellipsis, and English ones. The practical evaluation proves the usability and effectiveness of the indicator system and testing questions set. [Originality/Value] This study constructs an evaluation index system for government intelligent Q&A systems from the perspective of reducing administrative burden, compiles testing question sets using text mining technique, and proves their usability through an actual evaluation.

Key words: Q&A system, Government website, Capability of semantics understanding, Intelligent Q&A, Testing questions, Intelligent customer service