Multilingual social event detection method based on federal map neural network

Abstract

The invention relates to the technical field of social event detection, and discloses a multilingual social event detection method based on a federated graph neural network, comprising the following steps: S1, messages are extracted, messages in social information are extracted, and then messages related to the messages are extracted as nodes in a heterogeneous graph; S2, node edges are added, and edges between nodes are added according to the social information; s3, in a pre-training stage, a graph neural network is used for learning representation of a message, a message graph is initialized, and a model is initialized. Rich semantics and structure information in social information are fused together to obtain more knowledge, continuous social detection events can be coped with, the knowledge is expanded by using dynamic social communication, high-accuracy event detection in different language modal data environments can be realized, and the problem of minority-language event detection with few samples is effectively relieved.

Description

technical field [0001] The invention relates to the technical field of social event detection, in particular to a multilingual social event detection method based on federated graph neural network. Background technique [0002] Social events highlight major events in our daily life. These major events generally reflect social group behavior and widespread public concern. Social event detection is very important. It provides us with valuable opinions for timely responses to different events. Therefore, there are many applications in crisis management, product recommendation, decision-making and other fields. In recent years, social event detection has become a research hotspot in social media mining, and has attracted more and more attention and exploration from the industry. [0003] Since social events have attracted widespread attention, there have been many studies on social event detection. With the in-depth application of artificial intelligence in data mining, the dete...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the deficiencies of the prior art, the present invention provides a multilingual social event detection method based on a federated graph neural network, which solves the problem that the general detection method cannot effectively use the relationship between social data for social event detection, and cannot fully Mining the relationship between data cannot verify that federated transfer learning can effectively solve the problem of few samples in small languages

Images