User Identity Linkage Across Social Networks via Community Preserving Network Embedding

Abstract

User Identity Linkage (UIL) across social networks refers to the recognition of the accounts belonging to the same individual among multiple social network platforms. Most existing network structure-based methods focus on extracting local structural proximity from the local context of nodes, but the inherent community structure of the social network is largely ignored. In this paper, with an awareness of labeled anchor nodes as supervised information, we propose a novel community structure-based algorithm for UIL, called CUIL. Firstly, inspired by the network embedding, CUIL considers both proximity structure and community structure of the social network simultaneously to capture the structural information conveyed by the original network as much as possible when learning the feature vectors of nodes in social networks. Given a set of labeled anchor nodes, CUIL then applies the back-propagation neural network to learn a stable cross-network mapping function for identities linkage. Experiments conducted on the real-world dataset show that CUIL outperforms the state-of-the-art network structure-based methods in terms of linking precision even with only a few labeled anchor nodes. CUIL is also shown to be efficient with low vector dimensionality and a small number of training iterations.