Matrix Decomposition for Clustering and Collaborative Filtering

Since the consumers of today are flooded with choices for various products like movies on OTT platforms, onlineOnline music, and other onlineOnline shopping sites, so to increase user satisfaction and maintain loyalty, the retailers and content providers need to find ways to match users with their most preferred products of choice. So they use recommender systemsRecommender system which have been very successful in providing accurateAccurate suggestions of items to customers. The two main strategies used by recommender systemsRecommender system are content-based models and collaborative filteringCollaborative filtering. Matrix factorization is a collaborative filteringCollaborative filtering method to find the relationshipRelationships between items’ and users’ entities. Latent features, the association between users and movies matrices, are determined to find similarity and make a prediction based on both item and user entities. Matrix factorization is a way to generate latent features when multiplying two different kinds of entities. Since not every user gives ratings to all the items they use, there are many missing values in the matrix and it results in a sparse matrix. Hence, the null values not given by the users would be filled with 0 such that the filled values are provided for the multiplication. It has been observed that matrix factorization models are superior when compared to the nearest neighbor technique, for the generation of product recommendationsRecommendation. This is because it incorporates additional factors like implicit feedbackFeedback, temporal effects, and confidence levels into the recommendationRecommendation process. Therefore in this chapter, we see how the process of matrix decomposition works, in detail.