Methods for boosting recommender systems

Online shopping has grown rapidly over the past few years. Besides the convenience of shopping directly from ones home, an important advantage of e-commerce is the great variety of items that online stores offer. However, with such a large number of items, it becomes harder for vendors to determine which items are more relevant for a given user. Recommender Systems are programs that attempt to assist in such scenarios by presenting the user a small subset of items which she is likely to find interesting. We consider in this work a popular class of such systems that are based on Collaborative Filtering (CF for short). CF is the process of predicting user ratings to items based on previous ratings of (similar) users to (similar) items. The objective of this research is to develop new algorithms and methods for boosting CF based Recommender Systems. Specifically, we focus on the following four challenges: (1) improving the quality of the predictions that such systems provide; (2) devising new methods for choosing the recommended items to be presented to the users; (3) improving the efficiency of CF algorithms and related data structures; (4) incorporating recommendation algorithms in different application domains.