摘要
面对面协作知识建构的会话分析面临分析视角扩展、分析场景迁移以及分析方法改进等现实诉求。研究将协作知识建构会话分析与自然语言处理方法相结合,探索建立会话内容智能挖掘框架、流程与方法,并对真实课堂环境中的协作知识建构会话进行分析与挖掘。研究结论认为:知识建构会话内容挖掘重点应聚焦于知识建构中的概念分布与关联、观点分类与演变以及话题演化三个层次;基本分析阶段分为层级确立、要素解构、数据映射、模型构建、可视表征五个阶段;会话内容挖掘结果可从纵向时序与横向组间维度呈现词汇、观点与话题的内容结构与时序变化。未来研究可以关注两个问题:如何通过优化观点编码框架和训练数据集来提升观点分类的精度;如何将会话内容挖掘方法植入互动分析系统中,从而实现对知识建构互动会话内容的伴随式分析与反馈。
At present, the discourse analysis of face-to-face collaborative knowledge building confronts the realistic demand of analysis perspective expansion, analysis context migration and analysis method improvement. This study harmonizes the analysis of collaborative knowledge building conversation and natural language processing methods to explore the framework, process and methodology of conversation content mining, and then applies this approach for collaborative knowledge building conversation mining and analysis in real classroom environments. The results conclude that: knowledge building conversation content mining focuses on three levels of concept distribution and connections, idea category and evolution, and topic evolution;the basic analysis stage includes hierarchy establishment, element deconstruction, data mapping, model construction, and visual representation;conversation content mining presents content structure and sequential variation of vocabulary, ideas and topics from vertical temporal and horizontal intergroup dimensions. In the future research, two issues might be addressed: how to improve the accuracy of idea classification by optimizing the idea coding framework and training data sets, and how to incorporate the conversation content mining method into the interactive analysis system to achieve the simultaneous mining and feedback of knowledge building interactive conversation content.
作者
马志强
郭文欣
王萌
MA Zhiqiang;GUO Wenxin;WANG Meng(Jiangnan University,Wuxi Jiangsu 214122;East China Normal University,Shanghai 200062)
出处
《现代远距离教育》
CSSCI
2022年第1期3-13,共11页
Modern Distance Education
基金
教育部人文社会科学研究项目“面向心理健康的文本情绪识别研究”(编号:21YJA740037)。
关键词
知识建构
会话分析
观点挖掘
人工智能教育
学习分析
Knowledge Building
Discourse Analysis
Opinion Mining
Artificial Intelligence Education
Learning Analysis
