Finding Compatible People on Social Networking Sites, a Semantic Technology Approach

Abstract

the significant feature of a social networking website is the primary reason they are made for: connecting people and friends via internet. “Friend recommender systems” are wisely designed for finding people, most of whom tend to be with the same interests and backgrounds. These systems use a set of predefined items from which users specify their preferences simply by selecting from a fixed list. As a result, they can’t put it in their own words. Moreover, these systems only consider the “exact similarity matching” among the users’ interests to find and recommend new friends. The main focus of this paper is to introduce a new approach for matching more compatible friends on social networking websites. Contrary to existing approaches, our system let users specify their interests in their own words. Thus, users do not need to select their preferences from a predefined list. In addition, we define “compatibility” by introducing two new relations between users’ interests: “semantic” and “complementary” relations for the purpose of matching compatible users. We chose 50 members from Live Journal social network as our experimental case in this study and calculated compatibility degrees between each pair of them. The results show that the average error of this system is 0.2 which is acceptable in comparison with the similarity matching friend recommendation systems in which the average rate of error is 0.6.