Exploiting context-awareness and multi-criteria decision making to improve items recommendation using a tripartite graph-based model

Highlights：

• We propose a solution that exploits users’ contextual information and users’ feedbacks on items’ criteria to improve the prediction accuracy of recommender systems.

• We model the multi-dimensional available data in the form of a tripartite graph including three types of connected entities (users, contextual situations and criteria).

• We propose a context-aware multi-criteria recommendation approach that explores the idea of clustering contextually similar users evaluating items with respect to multiple criteria.

• We explore a novel way to predict cluster-based multi-criteria ratings for users involved in similar contextual situations by considering the dependence between contexts and also the correlation between criteria.

• We perform an intensive comparative evaluation with state-of-the-art baselines belonging to four categories of work: single rating based methods, context-aware based rating methods, multi-criteria rating based methods and context-aware multi-criteria rating based methods.