REMOVE: REcommendation Model based on sOcio-enVironmental contExt

Abstract

Recommender Systems (RS) suffer from the typical new user and data sparsity problems. In order to reduce these issues, a RecommEndation Model based on sOcio-enVironmental contExt called REMOVE is proposed in this paper. By elaborating the state-of-the-art recommendation algorithms, we conclude that the merge of both social and environmental information should be taken into consideration in a recommendation model. Hence, in our model, we argue that modifying the vector of characteristics of items as well that of the users can significantly improve recommendation quality. To the best of our knowledge, there has been no prior work that investigated matrix factorization by modelling item vector characteristic and considering social and environmental information. For the experimentation, we compared our approach to those that integrate contextual information. This evaluation provided encouraging results in terms of Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE), and demonstrates that’s REMOVE can better handle the concerned issues.