情报科学 ›› 2024, Vol. 42 ›› Issue (8): 54-64.

• 理论研究 • 上一篇    下一篇

基于KANO模型的政务智能问答系统功能优化研究

  

  • 出版日期:2024-08-01 发布日期:2024-11-05

  • Online:2024-08-01 Published:2024-11-05

摘要: 【目的/意义】研究政务智能问答系统的功能开发优先级,为系统建设的资源配置提供参考。【方法/过程】运 用KANO模型与Better-Worse系数,采用问卷调查法,对政务智能问答系统的用户进行调研,识别系统各项功能的 重要程度。【结果/结论】在政务智能问答系统的会话代理、理性代理和具身代理三类功能的16个子功能中,应优先 开发的功能依次为:1基本问答,2多渠道咨询,3相关问题推荐,4同义理解,5人工转接,6热点服务,7语言人性化,8 语境感知,9情感反馈,10问答纠偏。【创新/局限】指出了政务智能问答系统应当优先开发的功能:针对会话代理功 能,重点提升语义理解能力和对用户提问的自动规范能力;针对理性代理功能,重点增强相关问题和热点问题推荐 功能,同时谨慎控制地区导航功能和敏感词汇识别功能;针对具身代理功能,应该加强情感回应功能,积极塑造正 面形象。

Abstract: 【Purpose/significance】Study the priority of function development for government intelligent Q & A system, and provide a reference for resource allocation in system construction【. Method/process】Use the KANO model and Better-Worse coefficient, a ques⁃ tionnaire survey was conducted to investigate the users of the government intelligent Q & A system and identify the importance of vari⁃ ous possible functions of the system.【Result/conclusion】Among the 16 sub-functions of dialogical agent, rational agent, and embod⁃ ied agent, the priority order of development is: question and answering, multi-channel consultation, recommendation of related ques⁃ tions, synonymous understanding, manual transfer, hotspot service, humanization of language, contextual awareness, emotional feed⁃ back, question and answer correction【. Innovation/limitation】This article creatively points out the functions that should be prioritized for the development of government intelligent Q & A systems: In terms of the dialogical agent, priority should be given to strengthening semantic understanding ability and automatic standardization of user question; in terms of the rational agent, priority should be given to enhancing the recommendation of related and hot issues, and cautiously control the function of regional navigation and the function of sensitive vocabulary recognition; in terms of the embodied agency, priority should be given to strengthening emotional response and shaping a positive government image.