孙 众， 吕恺悦， 施智平， 骆力明
【关 键 词 】：
教学结构； 教学事件； 课堂教学分析； 人工智能； 人机协同
规则化与可计算是人工智能技术支持课堂教学分析的前提。以S-T行为及FIAS言语交互为代表的分析方法，用时间取样的研究思路量化课堂教学，虽然对规则化、客观分析作出了显著贡献，但一直受困于费时低效、未能有效建立数量结构与意义理解的联系，以及不能较好地促进教师专业成长等问题。建立以教学事件为基本分析维度，综合双主教学结构等教育理论，以及计算机视觉和自然语言理解等技术，从教学事件识别与教学阶段划分、教学法结构序列、时间取样的行为和言语交互分析、基于证据的教学解读和人机协同的教学改进等阶段，构建课堂教学分析TESTII（Teaching Events， SPS， Time Coding， Interpretation， Improvement）框架，成为人工智能支持课堂教学质量提升和变革课堂教学结构的解决方案和发展走向。
Regularization and computability are the prerequisites for AI technology to support classroom teaching analysis. According to the analysis method represented by S-T behavior and FIAS verbal interaction, time sampling is used to quantify classroom teaching. Although this method has made significant contributions to regularization and objective analysis, it fails to establish a link between quantitative structure and meaning understanding, and fails to promote teachers' professional development, for it has been plagued by such problems as time consumption and inefficiency. This paper takes the teaching events as the basic dimension of analysis, integrates the educational theories such as dual master teaching structure, computer vision and natural language understanding, and constructs a TESTII (Teaching Events, SPS, Time Coding, Interpretation, Improvement) framework for classroom teaching analysis from the stages of teaching event identification and teaching stage division, sequence of pedagogy structure, behavior and speech interaction analysis of time sampling, evidence-based teaching interpretation and teaching improvement of human-machine cooperation. This framework has become a solution and development trend of artificial intelligence to support the improvement of classroom teaching quality and the reform of classroom teaching structure.