Contextual Sub-network Extraction in Contextual Social Networks

Abstract

Predicting the trust between a source participant and a target participant in a social network is important in many applications, e.g., assessing the recommendation from a target participant from the perspective of a source participant. In general, social networks contain participants, the links and trust relations between them and the contextual information for their interactions. All such information has important influence on trust prediction. However, predicting the trust between two participants based on the whole network is ineffective and inefficient. Thus, prior to trust prediction, it is necessary to extract a small-scale contextual network that contains most of the important participants as well as trust and contextual information. However, extracting such a sub-network has been proved to be an NP-Complete problem. To solve this challenging problem, we propose a social context-aware trust sub-network extraction model to search near-optimal solutions effectively and efficiently. In our proposed model, we first present the important factors that affect the trust between participants in OSNs. Then, we define a utility function to measure the trust factors of each node in a social network. At last, we design an ant colony algorithm with a newly designed mutation process for sub-network extraction. The experiments, conducted on two popular datasets of Epinions and Slashdot, demonstrate that our approach can extract those sub-networks covering important participants and contextual information while keeping a high density rate. Our approach is superior to the state-of-the-art approaches in terms of the quality of extracted sub-networks within the same execution time.