Hybrid Movie Recommender System Based on Word Embeddings

A recommender system is an application intended to offer a user item that may be of interest to him according to his profile, the recommendations have been applied successfully in various fields. Recommended items include movies, books, travel and tourism services, friends, research articles, research queries, and much more. Hence the presence of recommender systems in many areas, in particular, movies recommendation. The problem of film recommendation has become more interesting because of the rich data and context available online, what advance quickly the research in this field. Therefore, it's time to overcome traditional recommendation methods (traditional collaborative filtering, traditional content-based filtering) wich suffer from many drawbacks like cold start problem and data sparsity. In this article we present a solution for these limitations, by proposing a hybrid recommendation framework to improve the quality of online films recommendations services, we used users ratings and movies features, in order to use two models into the framework based on word2vec and Knn algorithms respectively.