基于多维关联分析的教育精准服务模式研究

The crux of of realizing precise educational services is the representation of complex relationship among various data and the comprehensive analysis of the relationship between learners and resources. This project focuses on the overall data and proposes a context-aware personalized recommendation algorithm to provide suitable educational resources, and explores a multi-clustering algorithm in high-dimensional space to construct adaptive collaborative learning communities, thus the precise educational services could be provided for learners. Firstly, the learner model, which includes the learner's personality, behavior, and context, will be constructed based on the Interpretive Structural Model (ISM). Meanwhile, the educational resource model, which integrates metadata, social tagging, and social network analysis, will be constructed by combining expert-based metadata and user-based collaborative tagging. According to the learner model and resource model, some tensor-based representation models for learners and resources shall be presented, and further be integrated with the context tensor to a higher dimensional fusion tensor model. Therefore, for the fusion tensor model, the personalized recommendation and multiple clustering algorithms by exploiting the multi-dimensional correlation analysis method based on higher order singular value decomposition could be implementd. Finally, the precise educational service model embedded ubiquitous learning process will be proposed to push the on-demand recommended content to learners by selecting suitable service approaches. By enjoying the precise educational services, everyone can obtain personalized educational resources and adaptive development, further achieve the educational process fairness.

实现教育精准服务的关键在于全面考虑数据之间的复杂关系，对学习者和资源进行多维关联分析。项目从整体数据出发研究情境感知的个性化推荐算法以提供适合的教育资源，探索高维空间下多视图聚类方法以构建自适应的协作学习社群，为学习者提供适时适地和应时应景的精准教育服务。首先基于解释结构模型方法，构建涵盖学习者个性、行为和情境特征的学习者模型；结合专家元数据与用户协同标注方法，构建融合“元数据-社会标注-社会网络分析”的教育资源模型；据此利用张量理论对学习者、资源进行高维表征并与情境进行关联融合，通过基于高阶奇异值分解的多维关联分析方法，研究教育资源个性化推荐和学习者多视图聚类；最后探索嵌入泛在学习过程的个性化教育服务模式，将所推荐的服务内容以恰当的服务方法按需推送给学习者。通过教育精准服务促使人人都能获得个性化教育资源和自适应发展，达到教育过程公平。

实现教育精准服务的关键在于全面考虑数据之间的复杂关系，对学习者和学习资源进行多维关联分析。项目从整体数据出发研究情境感知的个性化推荐算法以期为学习者提供适合的教育资源和精准的教育服务。首先基于解释结构模型方法，构建涵盖学习者偏好特征、行为绩效、学习情境和学习动力系统的学习者特征模型；以专家分类元数据为框架，结合用户情境语义协同标注的学习资源表征方式，引入社会网络分析方法梳理、重构学习资源特征的语义关系和层次结构，形成自上而下的专家分类与自底向上的情境语义协同标注的学习资源表征方式，达到用户协同和情境融合；依据多维关联分析原理，利用张量理论对学习者、资源进行高维表征并与情境进行关联融合，通过基于高阶奇异值分解的多维关联分析方法，实现学习者和资源之间的精准匹配，达到教育资源个性化推荐；最后探索面向泛在学习过程的个性化教育服务模式，将所推荐的服务内容以恰当的服务方法按需推送给学习者。通过教育精准服务促使人人都能获得个性化教育资源和自适应发展，达到教育过程公平。