在帮助学生掌握学科知识的同时，培养他们协作解决问题能力，以应对社会实践与未来挑战是当前教育的重点。解决问题过程中的学生的讨论或会话内容含有丰富的信息，能反映学生的问题解决能力与合作能力。然而，如何对讨论或会话内容进行有效分析，并以此对学生的问题解决及相关能力进行客观评价，是目前教学研究与实践面临的挑战。自然语言处理技术为会话文本的自动化分析提供了有效的技术手段，但常规方法难以满足面向能力测评的会话数据分析。. 本课题旨在研究面向问题解决能力测评的会话数据(包括语音和文本)分析方法与技术，具体包括建立一套面向问题解决能力测评的中文会话数据自动化分类与编码方法，开发一套可供一线教师分析学生讨论数据及评估学生问题解决能力的自动化分析测评软件包，并将此技术用于实际教学情境（例如STEM课程的“船模设计分析”单元），通过对比专家的人工测评结果，来检验该方法或技术的信效度水平。

It is important to help students to develop collaborative problem-solving abilities to cope with real-world practice and dynamic challenges. Discussion plays a vital role in collaborative problem-solving. Meanwhile, the content of discussions may reflect students’ abilities in problem-solving and collaboration. Nevertheless, analyzing discussion or discourse data with the purpose to assess students’ abilities is a challenge. Although natural language processing technology provides an effective technical means for automated analysis of discussion content, conventional methods and technology are inadequate for measuring students problem-solving abilities through discourse analysis..This project aims to explore the method and technology of discourse data (including audio and text) analysis with the purpose to assess student problem-solving abilities based on their discussion data. It includes: (1) the establishment of a set of methods for coding and classification of discourse data in Chinese; (2) the development of a computer-based system/tool to automate the discourse analysis, which can be used by teachers to analyze students’ discussion content and assess their problem-solving abilities; and (3) empirical studies of applying the developed approach to authentic problem-solving and collaborative learning contexts (e.g., ship model design and analysis in a STEM course) to test the validity and reliability of the approach by comparing the assessment results from domain experts with those from the system.