【摘要】 显性化测量与评估在线学习行为和情感状态是学习分析领域的研究热点。为克服单模态数据分析片面和多模态数据融合模糊等问题,全面感知和反馈在线学习过程状态,本研究构建了一种在线学习行为多模态数据融合模型。该模型利用自动化操作行为事件监听、表情识别、生理特征监测等原理,从行为、情绪和认知3个维度进行时序数据同步融合、分层递进诊断评估和统计聚类分析;考虑到对在线学习者具有较低的侵入性和干扰性,该模型在技术实现上采用分布式物联网技术和开放式Django Web服务器部署技术,形成学习过程状态数据的自动采集、分析、融合、评估和反馈等多层体系结构;应用该系统对MOOC环境下在线学习行为评测实验,讨论了本模型的准确性、易用性和有用性等问题。研究结果表明,本研究构建的模型能够为在线学习分析提供一种有效的技术解决方案,为同类研究可提供方法借鉴与技术实现参考。更多还原

【Abstract】 Explicit measurement and evaluation of online learning behaviors and affective status is a hot issue in learning analytics research. This study sets out to construct a model of integrating multimodal data from online learning behaviors with the aim of enabling a holistic perception of and feedback on online learning experience. Based on the principles of automatic action event listener, identification of emotions and monitoring of physiological characteristics, the proposed model implements synchronous integration of time series data, step-by-step progressive diagnostic assessment, and cluster analysis from the three dimensions of behavior, affect and cognition. To reduce intruding and interfering effects on online learners, the model adopts distributed Internet of Things and open Django Web server deployment technology, hence able to automatically collect, analyze, integrate and feedback on learning process data. The model is verified in the context of MOOCs and its accuracy, ease of use and usefulness are discussed.It is found to have the potential to be an effective solution to drawbacks inherent in traditional single-mode data analysis and multimodal data analysis.更多还原