Cross-Network User Identification Using Global Seed and Optimal Local Extension

Abstract

Cross-network user identification aims to identify the accounts owned by the same user across multiple networks, which is significant in friend recommendation, network security and link prediction. Existing methods mainly make full use of a small set of seed users and iteratively identify the other users. However, limited by the scale of seed users, these methods can??t reach a satisfactory accuracy with low time complexity. A method of cross-network user identification using global seed and optimal local extension (GLE) is proposed. Firstly, in order to effectively solve the cold start problem, this paper proposes a global seed expansion method (GSE) to expand the seed set. Secondly, to ensure higher accuracy at a lower time cost, this paper proposes a local search range expansion method for candidate searching. Finally, experiments demonstrate that this method can significantly improve the recall and precision of user identification at a lower time cost, and effectively solves the identification problem when the scale of seed users is insufficient.

Keywords

• user identification
• social network
• global seed expansion model
• optimal local expansion model