Find Your Online Social Friends from Mobile Internet Traffic

• Published in: 2018 International Conference on Network Infrastructure and Digital Content (IC-NIDC) pp. 457 - 461
• Main Authors: Zhang, Yi, Qiao, Yuanyuan, Zhang, Yanting, Shu, Nanfei, Song, Yizhe, Yang, Jie
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.08.2018
• Subjects: Base stations; Feature extraction; Friend Recommendation; Global Positioning System; Human mobility; Internet; Mobile Internet; Predictive models; Social network; Social network services; Uniform resource locators
• ISBN: 9781538660669; 1538660660
• ISSN: 2575-4955
• DOI: 10.1109/ICNIDC.2018.8525736

Increasingly more mobile Internet traffic is produced which contains ample personal information related to user mobility and website browsing behavior. Prior research has attempted to recommend friends based on Global Position System (GPS) in location based social networks (LBSN). However, the study of friend recommendation in general social network according to position from the base station is relatively understudied. This paper introduces a novel feature set extracted from mobile Internet traffic according to base station location and Uniform Resource Locator (URL). We train classification models using these features to predict friendship between pairs of Weibo users. Results show that both base station location and URL when acted alone can already effectively reflect friendships even in general social network. We further show that by fusing the two features together, the model obtains even better performance. Finally, we demonstrate that the location and URL features can improve prediction performance than only using the common friends.