Social Link Prediction

Abstract

This chapter contributes toward introducing some learning automata-based algorithms for link prediction in social networks. Since one of the common link prediction methods for predicting hidden links use a deterministic and static graph where a snapshot of the network is analyzed to find hidden or future links, we study link prediction in social network which their structures are dynamic, online, and non-deterministic and introduce learning automata models as a powerful tools for such issues. The first learning automata approach for link prediction, which introduced in this chapter, is designed for stochastic social networks in which edge weights of graph are modeled as random variables. Another LA-based approach for link prediction, considered a weighted graph representation instead of a binary graph representation and the generalization of this approach is applied for fuzzy social networks. The link prediction in time series social networks is also introduced as well.