User Attribute Inference in Directed Social Networks as a Service

Social networking has become a frequent activity with many users having accounts on multiple social networking sites. Profile attribute inference has gained popularity due to its usefulness in applications such as link prediction and recommender systems. This paper proposes to infer ego user attributes, by propagating the known attribute values of followers or followings within certain circles. With the ability to follow or be followed by any user, the possibility of many weak links being formed is high. We utilize tie-strength to address this and differentiate each users' influence in the ego user attribute inference. We show that our approach based on followers sub-network, following sub-network and all links sub-network yields better accuracy than friends sub-network, which is the converted undirected network obtained from a directed social network. We conduct extensive experiments and achieve better accuracy than the previous best method for attribute inference.