Iterative Use of Weighted Voronoi Diagrams to Improve Scalability in Recommender Systems

Abstract

Collaborative Filtering (CF) technique is used by most of the Recommender Systems (RS) for formulating suggestions of item relevant to users’ interest. It typically associates a user with a community of like minded users, and then recommend items to the user liked by others in the community. However, with the rapid growth of the Web in terms of users and items, majority of the RS using CF technique suffers from the scalability problem. In order to address this scalability issue, we propose a decomposition based Recommendation Algorithm using Multiplicatively Weighted Voronoi Diagrams. We divide the entire users’ space into smaller regions based on the location, and then apply the Recommendation Algorithm separately to these regions. This helps us to avoid computations over the entire data. We measure Spatial Autocorrelation indices in the regions or cells formed by the Voronoi decomposition. One of the main objectives of our work is to reduce the running time without compromising the recommendation quality much. This ensures scalability, allowing us to tackle bigger datasets using the same resources. We have tested our algorithms on the MovieLens and Book-Crossing datasets. Our proposed decomposition scheme is oblivious of the underlying recommendation algorithm.