Spammer Detection Based on a Heterogeneous Multiple-mini-graph Neural Network

Nowadays spamming has become a worldwide issue on the internet and is used for fraudulent purposes. The main purpose of this chapter is to distinguish spammers fromnormal users by way of their posts, comments, or behavior pattern on social media. To solve the issue a deep learning-based graph convolutional network approach that uses a heterogeneous convolutional network with vanilla embedding to extract the features from the graph nodes is proposed. This handles both the homogeneity and heterogeneity of networks simultaneously. After performing extensive experiments on the Cora dataset (which is the textual contents taken from different social media platforms), it is found that the efficiency of the model is very good when compared with the existing methodologies in the field of Spammer Detection using deep learning models.