A study on handling sparseness in collaborative filtering

With the advent of the Internet, the number of choices that are opened to us online is constantly increasing. Movies, books, recipes, world news..., as many sets where we need to select without the possibility of considering all the necessary information. So how to choose? As we are not only faced with the same choice, if anyone has similar tastes to ours and he liked such a recent film, the chances that we also liked the film seem bigger. It is therefore possible to take advantage of available information on choice of other agents to induce preferences over our own choices. Now with the availability of Internet and major databases on user preferences, it becomes possible extending to large-scale, the concept of word of mouth. The formalization and operation of this intuition are the subject of collaborative filtering. Collaborative Filtering (CF) has become one of the most used filtering technique used to cope with the" information overload" problem. However, CF suffers from important bottlenecks: privacy, cold-start, sparsity... Many researchers have proposed methods for handling latter problem but it remains a great and important research area. Keywords: Collaborative Filtering, Cold Start, Sparsity Problem. A STUDY ON HANDLING SPARSENESS IN COLLABORATIVE FILTERING Yegwende Vincent TIEMTORE Department of Computer Engineering Anadolu University, Graduate School of Sciences, May, 2017 Supervisor: Prof. Dr. Yaşar HOŞCAN