基于大数据的网络生成性资源优选策略研究

Driven by the purpose of boosting educational application of online generative resources and targeting at optimization of generative resource selection, this study designed “one focus” (selection technique), “two criteria” (resource quality and personalized features) and “three questions” (evaluation indicators of selection, feature vector correlation technique and feature information integration processing) to investigate the techniques and model of online resource selection, so as to meet the need of promoting generative resources in the network era. The study systemically examined the issue of optimized selection of online generative resources. First, we collected the achievements and cases of resource selection and formulated the criteria of generative resource selection from the perspectives of resource quality and personalized features through Delphi method and interpretative structural modeling, which resulted in an interpretative structural model of evaluation indicators of resource selection. Second, this study established a model of content knowledge library, label information library and user behavior ontology, which realized the multi-level correlation of resource feature vectors through content analysis, time series analysis and Hidden Markov Chain Model. Third, the study built a multi-dimension service framework of generative resource selection. The study adopted supervised learning and made aggregated analysis, dimension reduction analysis and cluster analysis through aggregation/cascading and principal component analysis, by which resource classification and personalized recommendation were achieved.

以推动网络环境下生成性资源教育应用为主旨，以生成性资源优选方法为研究对象，设计“一个中心（生成性资源优选技术）、两条标准（资源质量和个性化特征）、三个问题（资源优选评价指标、特征向量关联技术、特征信息融合处理）”，探究网络环境下资源优选的技术和模式，满足网络时代生成性资源的推广需求。系统研究网络生成性资源的优选问题：①梳理教育信息资源优选的成果和案例，从资源质量和个性化特征维度，利用德尔菲法和解释结构模型，制定生成性资源优选标准，形成资源优选的评价指标解释结构模型；②建立学科知识库、标签信息库和用户行为本体模型，利用内容分析、时序分析和隐马尔科夫链模型等方法，实现资源特征向量的多重关联；③建立多维度生成性资源优选服务框架，采用监督式学习，通过聚合或级联、主成分分析对资源特征向量进行聚合、降维和聚类分析，实现资源的分类和个性化推荐策略。

以推动网络环境下生成性资源教育应用为主旨，以生成性资源优选方法为研究对象，探究网络环境下资源优选的技术和模式，满足网络时代生成性资源的推广需求。系统研究网络生成性资源的优选问题：①梳理信息资源优选的成果和案例，从资源质量和个性化特征维度，将信息技术赋能教育划分为三个层次，从以教学媒体的形式赋能教育，通过增强教育中人的感知、记忆、思维等信息处理能力来赋能教育，再到重组课堂教学结构再造教学流程来赋能教育；从技术障碍角度分析信息化教学可能存在的问题和不足，探索影响在线学习效果的关键制约因素。②建立学科知识库、标签信息库和用户行为本体模型，建立计算机可以识别和处理的评价指标，并引入同伴评分，采用“1+X”的指标体系对资源质量进行评价，对生成性资源进行排序。③建立多维度生成性资源优选服务框架，从人口学特征、支持性特征、策略性特征和学习特征等建构学习者特征模型，从知识属性和社会属性角度建立了信息资源描述模型。