A Novel Context-Aware Recommendation Approach Based on Tensor Decomposition

In the information age, the ability to analyze data has a fundamental role. In this field, recommender systems, that are able to provide suggests to users analyzing the information provided to system, play a central role. Moreover, the use of contextual information make recommender systems more reliable. This paper aims to describe a novel approach for context-aware recommender systems that exploits the tensor decomposition CANDECOMP properties in order to provide ratings forecasts. The proposed approach is tested on DePaulMovie dataset in order to evaluate its accuracy, and the numerical results are promising.