Year of Graduation
Similarity Fusion in Recommender Systems
Applied Mathematics and Information Science
In this paper proposed method based on the aggregation of most popular metrics for building recommendations using support matrix. Considered and compared item-based and user-based recommender systems based on collaborative filtering. Implemented 8 most popular similarity measures. By experimentation on real data from movielens.org showed that algorithm which use similarity fusion performs better than each metric. Moreover calculated performance of all implemented algorithms.