A Cantonese Rumor Detection Method Based on Deep Semantic Awareness Graph Convolutional Networks

Abstract

The invention relates to the technical field of rumor detection, and specifically discloses a Cantonese rumor detection method based on a deep semantic perception graph convolution network. First, a plurality of groups of healthy Cantonese rumor keywords are constructed, and a Web crawler is constructed to detect relevant tweets, users, Forwarding and comment information are obtained, and after completing the data annotation, a dataset Net‑CR‑Dataset is constructed; secondly, a deep semantic perception graph convolutional neural network model SA‑GCN is designed; according to the unique language features of Cantonese, BERT Chinese is pre-trained The model is optimized, and the BERT pre-training model is further pre-trained and fine-tuned by using a large number of collected Cantonese corpus, so as to extract the semantic feature vector of the tweet; and the improved GCN network is used to extract the structural features of the tweet. Structural feature vector; finally, the SA‑GCN model fuses the structural feature vector and the semantic feature vector to obtain the final classification result. The invention is superior to other commonly used detection methods in terms of detection effect and early detection capability.

Description

technical field [0001] The invention relates to the technical field of rumor detection, in particular to a Cantonese rumor detection method based on a deep semantic perception graph convolution network. Background technique [0002] Social media provides a platform for people to follow hot events, express opinions, and make friends, and plays an indispensable role in people's lives. According to the "Digital 2021" report, as of January 2021, global social media active users have reached 4.2 billion, accounting for about 53.6% of the world's total population. Thanks to the huge influence of social networks on public opinion, rumors emerge endlessly in social networks, which will not only disrupt the network order and cause social panic, but also bring economic losses in the real world and endanger national security. In addition to the common English and Chinese rumors, Cantonese rumors are also a big stubborn disease in the current social network. As a branch of Chinese, Ca...

AI Technical Summary

Problems solved by technology

Most of these detection methods based on graph structure do not consider the fusion of multiple features, which will lead to the loss of some important information, such as text content, user information, etc.

Meanwhile, commonly used shallow GCN networks cannot capture the features of distant neighbors

Moreover, the current series of research lacks the exploration of the field of Cantonese rumors, and Cantonese, as a major branch of Chinese, has a wide distribution of users, and Cantonese rumors in social networks also emerge in endlessly, so the present invention is based on graph structure and features Fusion method to detect Cantonese rumors in Twitter

[0016] However, most of the existing detection methods based on graph structure ignore the fusion of multiple features, resulting in the loss of some important information (such as text content)

Meanwhile, commonly used shallow GCNs may fail to capture the features of distant neighbors

Images