华东师范大学学报(教育科学版) ›› 2023, Vol. 41 ›› Issue (8): 79-89.doi: 10.16382/j.cnki.1000-5560.2023.08.008

• 学习科学 • 上一篇    下一篇

面向思维培养:基于精准标注技术的智能化课堂教学分析及应用

宋宇1, 许昌良2, 朱佳3, 柴少明4   

  1. 1. 华南师范大学人工智能与课堂教学交叉研究中心,广州 510631
    2. 广州市华侨外国语学校,广州 510095
    3. 浙江师范大学浙江省智能教育技术与应用重点实验室,浙江金华 321004
    4. 华南师范大学国际商学院,广州 510631
  • 出版日期:2023-08-01 发布日期:2023-07-25
  • 基金资助:
    国家自然科学基金项目“学习分析视角下面向高阶思维发展的课堂互动分析与评测”(No.61907017)

Thinking Training Oriented: Analysis and Application of Intelligent Classroom Teaching Based on Accurate Labeling Technology

Yu Song1, Changliang Xu2, Jia Zhu3, Shaoming Chai4   

  1. 1. South China Normal University, Research Center of Artificial Intelligence and Classroom Teaching, Guangzhou 510631, China
    2. Guangzhou Overseas Chinese Foreign Language School, Guangzhou 510095, China
    3. Zhejiang Normal University, Key Laboratory of Intelligent Education Technology and Application of Zhejiang Province, Jinhua 321004, China
    4. South China Normal University, International Business College, 510631, China
  • Online:2023-08-01 Published:2023-07-25

摘要:

课堂是人才培养的主阵地,提升课堂教学质量是促进基础教育高质量发展、培养创新型人才的重要抓手。新型课堂教学以课堂对话为载体,以思维培养为主要目标,智能技术为课堂教学分析提供了有效的工具手段。为了发挥智能技术的作用、赋能课堂教学分析和高阶思维发展,本研究创设了基于音视频转录文本的课堂教学自动标注方法,以面向思维培养的课堂教学评价体系为依据,借助卷积神经网络与双向长短期记忆网络相结合的混合神经网络技术实现了大规模课堂数据的快速精准标注,能够有效提炼课堂教学中的思维特征。在此基础上,发展了适应课堂教学场域的序列模式挖掘技术,揭示了课堂教学的过程性发展模式和思维进阶规律。为了验证以上智能技术在课堂教学分析和学生思维培养中的作用,本文以广东省A学校为例进行了为期一年的试验并进行了全过程监测,通过对比第一次和最后一次监测结果发现,在智能技术的加持下,涉及高阶思维的课堂对话比例得到显著提升,思维链条更长且能体现由低阶思维朝向高阶思维进阶的规律,其中较为显著的长链条对话呈现出知识习得→观点表达→分析阐释→总结归纳→迁移创新的进阶模式。未来智能化课堂教学分析可以从以下三方面着眼:发展以课堂对话为主、多模态数据融合的协同标注与分析技术;以发展评测体系为基础,研究覆盖多样态课堂的教学模式;智能技术的选择和运用要更加精准科学地服务于思维发展、认知能力提升等教育教学目标,从而有效推动课堂教学转型,创建优质高效的课堂。

关键词: 课堂教学, 高阶思维, 智能技术, 自动标注, 序列模式挖掘

Abstract:

Classroom teaching is the main position of talent training. Improving the quality of classroom teaching is important to promote the high-quality development of basic education and cultivate innovative talents. This study systematically puts forward a classroom teaching evaluation system for thinking training. Based on this, an automatic labeling method of classroom teaching based on audio and video transcripts has been created. With the help of the hybrid neural network model combining CNN and BiLSTM model, the rapid and accurate annotation of large-scale classroom data is realized, which can effectively refine the thinking characteristics in classroom teaching. At the same time, sequential pattern mining technique suitable for evaluation of classroom teaching has been developed, which can reveal the sequential pattern of high-quality classroom teaching and the advanced law of implicit thinking. This paper takes a school in Guangdong Province as an example to conduct a one-year experiment. By comparing the first and last monitoring results, it is found that with the blessing of intelligent analysis technology, the proportion of classroom dialogue involving high-level thinking has been significantly increased, the thinking chain is longer, and it can reflect the law of advancing from low-level thinking to high-level thinking, Among them, the more significant long chain dialogue is the advanced mode of basic knowledge acquisition→ personal opinion expression→analysis and interpretation→summary and induction → migration and innovation. The future intelligent classroom teaching analysis should focus on the following three aspects: developing technology based on classroom dialogue analysis and multi-modal data cooperation; the selection and application of intelligent technology should serve the education and teaching objectives, so as to effectively create a high-quality and efficient class.

Key words: classroom teaching, high level thinking, intelligent technology, automatic labeling, sequential pattern mining.