Implementation of Socially-Driven Learning-Based Prefetching in Mobile Online Social Networks

Abstract

In this paper, we discover the problem of auxiliary efficient access to social media contents on social network sitesfor mobile devices deprived of requiring mobile users to be onlineall the time.In order to offer the qualityof experience provision for mobile OSN services, in this paper,we suggest a socially-driven learning-based framework, namelySpice, for the media content prefetching to decrease the accessdelay and enhance mobile user’s satisfaction. Over and done with a largescale data-driven analysis over real-life mobile Twitter tracesfrom over 17 000 users during a period of five months, we disclose that the social friendship has a great impact on user’s mediacontent click behavior. To capture this effect, we conduct thesocial friendship clustering over the set of user’s friends, andthen develop a cluster-based Latent Bias Model for sociallydriven learning-based prefetching prediction.