A Key Person Extraction Method for Chat Groups Based on Structure-Aware Graph Neural Network

With the rapid development of mobile Internet and communication software, chat groups are used more and more in daily life and work scenarios, and a large amount of chat data is generated, which has a great impact on many fields, such as network information dissemination, corporate marketing, etc., and the supervision of group information also poses certain challenges. At present, there are many studies on opinion leaders in social networks, but there are relatively few similar studies on chat groups. Therefore, it is of great value to analyze and mine chat groups and extract key figures. However, the message data in the chat group exists in the form of sequence, without the network structure of the social network, and the users in the chat group have less available information, so it is impossible to extract the key figures in the chat group by referring to the existing Research. Based on the above background and problems, this paper proposes a chat group key person extraction method based on the structure-aware graph neural network and improved PageRank algorithm, using the structure-aware graph neural network to learn the relationship between group users, construct the relationship graph, and then Use the improved PageRank algorithm based on the number of sent messages and the edge weight of the relationship graph to calculate user influence and extract key figures. The validity and rationality of this method are proved by experiments.