2018 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM) Social Network Mining for Recommendation of Friends Based on Music Interests Chenxi Fan, Huizi Hao, Carson K. Leung ( ) , Leslie Yu Sun, and Jennifer Tran Department of Computer Science, University of Manitoba Winnipeg, MB, Canada Email: kleung@cs.umanitoba.ca Abstract—With the rapid development of technology and software, social media have become a necessity in our daily lives as it is a way for people to keep in touch with friends and share about current events. Some of the most popular social media and social networking sites that people use include Facebook, Instagram, Snapchat, and Twitter. Finding compatible persons to be friends on social media can be a challenge as many of the people recommended to the user by social media are people who are already friends with them or have been followed. However, when users are looking for friends, the real concern is whether they have common interests or hobbies with each other and whether they often interact with one another. In this paper, we propose friend recommendation algorithms revolving around music interests and interactions in social media. Keywords—Big data analytics, implicit social graph, interactions, music interests, social network analysis, social network applica- tions, social network mining I. I NTRODUCTION With the rapid development of technology and software in the current era of big data [5], [10], [13], [16], [18], high volumes of a wide variety of valuable data which may be of different veracity (e.g., precise data, imprecise and uncertain data [3], [9], [17]) can be easily generated and collected at a high velocity. Embedded in these big data are useful information and knowledge. Rich sources of big data include social media and social networks [8], [20]. They have become a necessity in our daily lives as they are ways for people to keep in touch with friends and share about current events. Some of the popular social networking sites that people use include Facebook, Instagram, Snapchat, and Twitter [6], [7]. In a society, everyone has their own networks of friends re- gardless whether they met online or in person. These networks of friends mainly come from (i) people who they interact with (e.g., classmates, relatives, colleagues) or (ii) people who suggested by social network recommendation algorithms. For instance, the “People You May Know” feature in Facebook utilizes the implicit social graph, which captures interactions between users and their contacts [21]. In this paper, we focus on the number of mutual friends between users in a social network. Specifically, every user S in Facebook is considered a source node in the implicit social graph. His friend F is considered a node-to-source relationship, which is represented by an edge connecting the two nodes capturing interactions between the two corresponding users (S and F ) on social me- dia. In general, a user may periodically communicate with his friends by (i) commenting or liking their posts on Facebook, and/or (ii) providing contents in the form of links or pictures. Given a user’s social network with edge’s weight represents how close the node (friend F ) is related to its source (user S), the edge is bidirectional to represent the mutual friendship (cf. directional edges to represent ‘following’ relationships [14], [15]). The weight of the edge is computed based on the number of interactions between two users. Inspired by the interaction rank algorithm developed by Roth et al. [21] mainly for capturing the email interactions among users, we present a new interaction ranking algorithm for suggesting friends based on factors such as music interests by summing the users’ number of musicians liked, their current location, the number of times they attended a music event, and the number of music-related posts that has been commented or liked. Each of these factors is also associated with a corresponding weight capturing the importance of such factor in the generation of the preference matrix between users. Based on the weighted sum, we compute the similarity of the targets user tags and potential recommendee. We then rank and select the highest rated users as recommended friends. Key contributions of our paper are the three interaction rank algorithms that captures interactions among users and their potential friends. The algorithms recommend friends based on music interests. The reminder of this paper is organized as fol- lows. The next section provides background and related works. We then describe our friend recommendation in Section III. Evaluation is presented in Section IV. Finally, conclusions is drawn in Section V. II. BACKGROUND &RELATED WORKS A. Social Graph and Bipartite Graph There has been some previous work on social graph tech- nique. For instance, Roth et al. [21] presented a social graph with weighted edges, which express the relationship strength between a user and his implicit groups. The core criteria are to compute the edge weights with frequency, recency, and di- rection. In order to satisfy these criteria, an interaction ranking algorithm—which sums the number of friends’ comments and page likes between a user and another particular user that weighs each interaction (e.g., send or receive emails from groups of other users) between friends as a function of its recency—was developed. The algorithm was mainly developed IEEE/ACM ASONAM 2018, August 28–31, 2018, Barcelona, Spain 978-1-5386-6051-5/18/$31.00 © 2018 IEEE 833