Abstract: Recommender systems support users in the overwhelming task of examining through large quantities of data in order to select appropriate information or items. Unfortunately such systems may be matter to attack by spam users who want to operate the systems recommendations to outfit their needs: to encourage their own items/services or to originate trouble in the recommender system. Attacks can cause the recommender system to become untrustworthy and unreliable, resulting in user dissatisfaction. Traditional recommender systems rely on like-minded neighbors irrespective of their preferences/tastes when computing predictions and assume users are independent and identically distributed and completely ignore the social activities between users which are not reliable. In reality people heavily rely on their friends recommdations since, social networks demonstrate a strong community effect. Furthermore, people in cluster/group tend to trust each other and share common preferences with each other more than those in outside the groups. Based on this intuition in this framework, architecture of trusted-community recommender system is proposed. Users preferences expressed by incorporating trusted neighbors within community of the target user are merged in order to find the similar preferences. In addition, the worth of merged ratings is measured by the confidence considering the number of ratings inside the community and the percentage of clashes between negative and positive views. Further, the rating confidence is incorporated into the computation of user similarity. The prediction for an unrated item is computed by aggregating the ratings of similar users within community. Experimental results on real-world data set validate that our method overtakes other complements in terms of accuracy.
Satya Keerthi Gorripati and Valli Kumari Vatsavayi, 2017. A Trusted-Community Based Framework for Collaborative Filtering Recommender Systems. Journal of Engineering and Applied Sciences, 12: 6095-6103.