A Hybrid Collaborative Filtering recommendation algorithm for Web-based Learning systems

The popularity of Web-based Learning (WBL) application has provided different learning resource for the online learning users in different sites. However, without proper guidance, it will be very difficult for the online users without enough background knowledge to choose satisfactory resources in their learning processes. Therefore, A novel Hybrid Collaborative Filtering (HCF) called Standard Deviation HCF (SD-HCF) for WBL prediction is proposed. We consider both user and resource similarity, and a weight is added to the similarity compassion to solve the overestimation of PCC similarity. Meanwhile, a standard deviation weight is integrated to prediction phase. The experimental results on a real-world WBL datasets, show that, SD-HCF can achieve high prediction accuracy as compared with other collaborative filter methods.