张 琳1 姜 强1 赵 蔚1 赵 艳2
（1.东北师范大学 信息科学与技术学院，吉林 长春，130117；2.长春师范大学 教育学院 教育学院，吉林 长春，130032）
【摘 要】如何科学有效地评价数字教育资源质量以促进其优化建设和有效供给，成为当前教育领域亟待解决的关键问题。文章采集国家高等教育智慧教育平台中音乐艺术、文史哲法、教育教学、医学保健、计算机和经济管理类资源学习者的在线评论数据，采用主题聚类方法确定数字教育资源质量评价维度，然后应用融合主题特征向量的意见挖掘模型得到各维度的质量评分。研究结果表明：影响数字教育资源质量最重要的维度是内容组织和语言表达，其次分别是知识讲解、教学材料、学习评价、资源适配、教学媒体、教学策略、教学交互、扩展资源、学习体验、学习成效、资源更新和教师特质；在各类数字教育资源中，音乐艺术类资源质量评价最好，计算机类和经济管理类资源质量评价整体较低。本研究为数字教育资源质量评价提供有效方法。
【关键词】数字教育；LSTM模型；BERTopic模型；意见挖掘；质量评价
Research on Quality Evaluation of Digital Educationat Resources Based on Online Review Opinion Mining: Taking the National Higher Education Smart Education Platform as an Example
ZHANG Lin1, JIANG Qiang1, ZHAO Wei1 and ZHAO Yan2
(1.School of Information Science and Technology, Northeast Normal University, Changchun 130117, China; 2.School of Education, Changchun Normal University, Changchun 130032, China)
Abstract: How to scientifically and effectively evaluate the quality of digital educational resources to promote their optimized construction and effective supply has become a critical issue urgently needing resolution in the current educational field. This article collected learner online review data in six categories: Music& Art, Humanities & Social Sciences (History, Philosophy, and Law), Education & Teaching, Medical & Health Sciences, Computer Science, and Economics & Management from the National Higher Education Smart Education Platform. A topic clustering method was adopted to identify evaluation dimension for digital educational resources quality, followed by an opinion mining model integrating topic feature vectors to calculate quality scores across these dimensions. Results show that the most critical dimensions influencing resource quality are content organization and linguistic expression, followed sequentially by knowledge explanation, instructional materials, learning assessment, resource adaptability, teaching media, instructional strategies, teaching interaction, supplementary resources, learning experience, learning outcomes, resource updating, and instructor characteristics. Among all the categories, Music& Arts resources received the highest quality evaluations, while Computer Science and Economics & Management resources were rated comparatively lower. By employing opinion mining to analyze textual reviews of National Higher Education Smart Education Platform resources, this article provides an effective methodology for evaluating digital educational resource quality.
Keywords: digital education; LSTM model; BERTopic model; opinion mining; quality evaluation
下载： 
基于在线评论意见挖掘的数字教育资源质量评价研究.pdf
