A hybrid recommendation technique optimized by dimension reduction
Dong-ru Ruan, Tian-hong Meng, Kai Gao · 2016
Providing high quality recommendations is significant for e-commerce systems to assist users in making effective selection decisions from a plethora of choices. Collaborative filtering (CF) is one of the most well-known and successful techniques to generate recommendations. However, CF suffers from an inherent issue that does not think over the auxiliary information such as item content information. This paper proposes an approach to combine the similarity of auxiliary information with the similarity of items based on user-item ratings. Specifically, due to the high dimensions and linear correlation of the auxiliary information, this paper uses Principal Component Analysis (PCA) to reduce the dimension to improve the predictive accuracy. In addition, Pearson correlation is better than the Euclidean distance as the similarity measurement to predict accuracy of preference value through the analysis of the different experimental results. Experimental results on real-world data sets demonstrate that the effectiveness of our approach.