Huge number of adaptive web applications have evolved in the recent years. These web applications attempt to understand the behavior of active user and then provide suggestions based on opinions of friends, relatives and peers etc to facilitate their decisions. Such applications are known as personalized web applications or recommender systems [21]. In recent times, massive number of recommender systems have been developed in both the academia and the industry. The estimated figures of varied recommender systems developed in different application domains [20], are approximately 21 for recommending the Web, 13 for movies, 11 for news, 10 for Document and Information, 6 for music, 5 for information filtering and sharing etc. The most prevalent techniques used to develop the recommender systems include Content based filtering, Collaborative filtering and hybrid techniques. Other methods include Demographic filtering, Utility based recommender systems and Knowledge based recommender systems.
Content-based filtering techniques analyze the past preferences of the active user to generate future recommendations. Nevertheless Collaborative filtering techniques assist users in effectively identifying the content of their interest by utilizing the opinions of the community of users [2]. Such systems do not rely on the contents or the features of the products. Thus they have been widely used for recommending items where item profiles are either missing or they are too precise to be useful. Numerous methods have been proposed in the literature for developing collaborative recommender systems like the most original correlation based methods [7], latent semantic indexing (LSI) [9], and Bayesian learning [12] etc.