PROJECT TITLE :

Identifying User Relationship on WeChat Money-Gifting Network

ABSTRACT:

The identification or classification of real-life relationships between users has become very useful for many applications as a result of the proliferation of online social networks. Some examples of these applications include the detection of financial fraud. In the real world, it is common practice for individuals who do not share the same relationship to give one another gifts that have significant meanings on separate occasions. The red packet is a traditional form of monetary gifting across many cultures in Asia, but it is particularly prevalent in Chinese culture. People have gradually begun to give electronic red packets rather than paper ones as the means of money gifting on social network platforms as a result of the rapid development of the Internet. In light of the aforementioned motivation, the purpose of this paper is to advocate for a novel method that exploits the red packet interactions of users in order to identify users' relationships on WeChat, which is one of the most popular social platforms in China. Specifically, we perform an analysis of the WeChat red packets network and then mine the semantic information contained within the amount and sending time of each red packet to determine the types of real-life relationships that exist between WeChat users. On the one hand, we construct an Amount-Date Graph and apply the graph embedding method to learn embeddings of the amount and sending date of each red packet in order to better capture the red packet gifting behaviors that occur between users for the purpose of relationship identification. This allows us to more accurately capture the red packet gifting behaviors that occur between users. On the other hand, we propose a new sequential model called the Cross & Attention Sequence Model (CASM), which explicitly learns the interactions between the latent semantic information of each red packet's amount and sending date in the red packets sequence between two users. This model is called the Cross & Attention Sequence Model. In order to demonstrate the efficacy of our methodology, we run exhaustive tests on a real-world dataset called WeChat Users Red Packets. This dataset includes eight different types of actual relationships. The experiments demonstrate that the proposed method performs noticeably better than the baselines and achieves an accuracy of prediction of 81.70 percent.