Enhancing accuracy of recommender systems through various approaches to local and global similarity measures