A Hybrid Approach of Recommendation via Extended Matrix Based on Collaborative Filtering with Demographics Information

In view of the growth in the use of methods based on matrix factorization, this research proposes an hybrid approach of recommendation based on collaborative filtering techniques, which exploits demographic information of the user and item within the factorization process, considering an extended rating matrix in order to generate more accurate prediction. In this paper we present an approach of collaborative filtering that is at least as accurate as the biased matrix factorization models or better than them in terms of precision and recall metrics. Several experiments involving different settings of the proposed approach show predictions of improved quality when extended matrix is used. The model is evaluated on three open datasets that contain demographic information and apply metrics to measure the performance of the proposed approach. Additionally, the results are compared with the traditional bias-based factorization model. The results showed a more expressive precision and recall than the model without demographic data.