Catching Dynamic Heterogeneous User Data for Identity Linkage Learning

Abstract

Benefitting from the development of social platforms, more and more users tend to register multiple accounts on different social networks. Linking user identities across multiple online social networks based on user behavior patterns is considerable for network supervision and information tracking. However, a user’s online behavior in a social network is dynamic. The user profile may be changed due to some specific reasons such as user migration or job changes. Thus, catching the dynamics of evolutionary user data and collecting the latest user features are important and challenging issues in the area of user identity linkage. Inspired by deep learning models such as word2vec and Deep Walk, this paper proposes an integrated framework to catch the dynamic user data by supplementing vacant features and updating outdated features in data sources. The framework firstly represents all textual and structural user data into Iow- dimensional latent spaces by utilizing word2vec and DeepWalk, then, integrates different user features and predicts vacant data fields based on late fusion approach and cosine similarity computation. We then explore and evaluate the application of our proposed method in a user identity mapping task. The results proved that our framework can successfully catch the dynamic user data and enhance the performance of identity linkage models by supplementing and updating data sources advance with the times.