Most Influential Contextual-Features [MICF] based model for Context-Aware Recommender System

Recommender system is an information filtering system that finds its applications in various e-commerce related fields. It recommends a list of items to an end-user from a potentially overwhelming collection of choices. Since the preferences of a user is different from the likings of other users, traditional recommender systems that recommend toprated entities to all the users, may not suffice in anticipating the needs of a user. Therefore, contextualization of recommender system is required to act more efficiently and in a user-specific manner. In an effort to deliver personalized recommendations shaped by user's contextual information, we have devised a novel methodology to incorporate contextual information into the recommender system. The proposed algorithm presents a framework for identifying the relevant contextual-variables and generating the cluster of contextual-features that depict similar rating-pattern for each class of entities. Thereafter, determining the set of Most Influential Contextual-Features that exhibit same rating-pattern as the end-user across all classes and predict the rating an end-user will give to an item, he has not rated before. Our algorithm not only renders intelligent and personalized recommendations but also alleviates cold-start, sparsity and newitem problem of traditional recommender system.