Linking User Identities Across Social Networks via Frequency Domain Analysis

Abstract

User identity linkage refers to linking different social accounts belonging to the same natural person. Now user identity linkage across social networks based on spatiotemporal data has attracted more and more attention. However, the existing methods have some problems, such as trajectory processing is not suitable for sparse data, and grid processing leads to information loss and abnormality. Because of the above problems, we propose an accurate and efficient method of user identity linkage via wavelet transform, WTLink, which expresses the user identity in the form of several critical points obtained through a novel wavelet transform application mode. Then the user identities are linked by calculating the similarity between their representations with a proposed metric. We compare this method with several existing user identity linkage methods based on spatiotemporal data on real datasets. The results show that this method exceeds the baseline methods in terms of effectiveness and efficiency.