Link Prediction in Online Social Networks Using Group Information

Abstract

Abstract. Users of online social networks voluntarily participate in different user groups or communities. Researches suggest the presence of strong local community structure in these social networks, i.e., users tend to meet other people via mutual friendship. Recently, different approaches have considered communities structure information for increasing the link prediction accuracy. Nevertheless, these approaches consider that users belong to just one community. In this paper, we propose three measures for the link prediction task which take into account all different communities that users belong to. We perform experiments for both unsupervised and supervised link prediction strategies. The evaluation method considers the links imbalance problem. Results show that our proposals outperform state-of-the-art unsupervised link prediction measures and help to improve the link prediction task approached as a supervised strategy.

A naïve Bayes model based on overlapping groups for link prediction in online social networks

“…Following the formalism showed in [12], for a network G, we denote by Lx,y and Lx,y the class variables of link existence and nonexistence, respectively, for a pair of nodes (x, y) ∈ V . The prior probabilities of Lx,y and Lx,y are calculated according to Eq.…”

“…Two standard evaluation measures are used to quantify the prediction accuracy [7,11,12]: AUC (area under the receiver operating characteristic curve) and precision. The AUC is interpreted as the probability that a randomly chosen and correctly predicted link has a higher score than other randomly chosen and wrongly predicted link.…”

“…When a node x belongs to a set of node groups Gα, this node is represented as x Gα . A node belongs to just one set of node groups [12].…”

“…The connection likelihood between a pair of nodes can be improved by identifying the different roles that their common neighbors play, for example, identifying their behaviors into the different groups that they belong to [10,12]. Hence, measures such as CN, AA and RA try to capture different roles from the set of all common neighbors.…”

“…However, in online social networks users usually belong to more than one community. Thus, in [12] are proposed three measures based on overlapping communities information: Common Neighbors Within and Outside of Common Groups (WOCG), Common Neighbors of Groups (CNG) and Common Neighbors with Total and Partial Overlapping of Groups (TPOG). The assumption of these measures is that common neighbors which belong to the same group of a pair of vertices are likely more influential in the existence of a future link between these vertices than common neighbors of different groups.…”