中国电化教育

在人工智能助推教师队伍建设背景下，借助画像技术实现对教师能力的精准测评并助力其个性化发展，愈发重要。然而，面向教师画像的能力测评精准性仍有待提高，其可视化呈现方式的解释性也仍显不足。研究根据教师能力的构成特征及多模态数据测评文献分析，构建了全面细粒度的面向教师画像的能力测评框架，并从数据多源采集及其准确处理的角度，提出了有效场景信息识别、数据采集高支持度保障、跨场景数据动态融合、非线性特征提取算法优化等四方面的测评精准性提升策略。结合自主研发的师范生教学基本技能智能实训系统，研究分析了教师画像的数据可视化特征，探讨了能力外显的行为、社会交互、知识、情感等四类数据属性的可视化方式。

In the context of artificial intelligence promoting the construction of teachers' team, it is increasing important to realize the accurate evaluation and personalized development of teachers' ability supported by teachers' profile. However, the accuracy of ability evaluation of teachers' profile still needs to be improved, and the explanation of its visual presentation is also insufficient. According to the characteristic of the composition of teachers' ability, and based on the analysis of literature of multi-modal data evaluation, this study constructs a comprehensive and fine-grained ability evaluation framework for teachers' profile, which provides a basis for accurate evaluation. From the perspective of multi-source data acquisition and its accurate processing, the researcher also puts forward four strategies to improve the accuracy of evaluation, including effective scene information identification, high support for data acquisition, dynamic fusion of cross-scene data, and optimization of nonlinear feature extraction algorithm. Combined with the self-developed intelligent training system for normal students' basic teaching skills, this study analyzes the data visualization characteristics of teachers' profile, and discusses the visualization methods of four types of data attributes, including behavior, social interaction, knowledge and emotion.