An effective link prediction method in multiplex social networks using local random walk towards dependable pathways

Link Prediction Problem (LPP) is defined as the likelihood of a future connection between two nodes on a network that does not currently have a link between them. Usually, the description of LPP in online social networks is done in a single-layer, while this can lead to incorrect modeling of real-world details. Hence, LPP is described as multi-layer networks that involve performing predictions on a target layer taking into account the information of other layers. Multiplex networks are a common type of multi-layer network with the same nodes and different links in each layer. Because the topological structure is correlated between different layers in MSN, the LPP solution can be improved by combining information from different layers. This paper introduces a method for LPP on Multiplex Social Networks (MSN) based on discovering dependable pathways. Dependable pathways are defined by considering the importance of each link in the communication paths between nodes, where we present the importance of links by network mapping as a weighted network. The weight of the links is calculated based on the topological structures between the different layers in MSN by formulating the interlayer and intralayer links. Finally, we propose a new similarity measure with the development of the Local Random Walk measure based on dependable pathways and weighted network. This measure records network structure using pure random walk to calculate similarities and discovery of unknown links. Experimental studies have been performed on seven MSN real-world datasets. The results show the superiority of the proposed method compared to some state-of-the-art methods such as MLRW and LPIS.