Multiple userids identification with deep learning

Abstract

People are becoming increasingly active on social media. It is common knowledge that most users register more than one social media account. Some users use multiple accounts to publish fake reviews to disorder the regular social network. To identify multiple accounts of the same author, multiple userids identification problem is defined. Existing methods for multiple userids identification problem focus on human-designed shallow statistical features and do not fully utilize the deep semantic information in user-generated texts. This paper uses the deep learning model to extract deep semantic features at the document level and user level. Then the most similar user pairs can be selected by computing the similarity of their writing styles indicated by text features. The experimental results demonstrate that our deep learning method outperforms state-of-the-art methods.