面向泛在学习智能推荐的多源数据融合用户偏好建模策略

Modeling Strategies for User Preferences from Multi-source Data Convergence in Ubiquitous Learning Personalized Recommendation

智能推荐是解决泛在学习资源过载、优化用户学习体验与提升服务个性化水平的有效途径。准确识别用户偏好是实现泛在学习智能推荐服务效果的关键,影响用户偏好模式形成的因素众多,基于多源偏好数据分析,分别从用户偏好建模的关键问题、多源偏好数据的抽取策略以及整合多源偏好数据的建模过程三个方面开展泛在学习智能推荐中的用户偏好建模策略研究。

Personalized recommendation is an effective way to tackle overloaded learning resources, optimize learning experiences and improve personalized services. The key to the effectiveness of the recommendation services of ubiquitous learning lies in the precise recognition of user preferences. There are many factors influencing user preferences forming. Based on multi-source preference data analysis, this paper explores user preferences modeling strategies in ubiquitous learning personalized recommendation in three aspects： key issues of user preferences modeling, strategies of preferences data sampling from multi-source data and modeling process of user preferences through integrating multi-source data.