Advancement Through Research
Recommender systems need production recommendation methods with highest quality and for millions of users. If the new user of recommender system apply a kind of product, due to the lack of sufficient information about the interests of users in this system, recommending items for him being very difficult and has a high error rate. This problem also observes for new items and causes the users recommender system being dissatisfied. In order to solve the above problems, many of recommender systems use of clustering users methods and create the recommendations based on the opinions and views of the neighboring users. Following the we offer an approach in order to select the initial centers of clusters in the K-means clustering algorithm. Finally, the proposed approach is used for clustering the users of recommender systems and thereby using this method we can improve the quality of created recommendations. Experiments evaluating our approach are carried out on the real data set taken from movies recommendation system of MovieLens web site. Preliminary results suggest that our approach can improve prediction accuracy compared to existing approaches.
Clustering, K-means clustering algorithm, Recommender systems, similarity
