| Autor: |
Jose, Teena, Babu, Suvanam Sasidhar |
| Zdroj: |
Journal of Ambient Intelligence and Humanized Computing; 20240101, Issue: Preprints p1-15, 15p |
| Abstrakt: |
The vast usage of social media makes it a familiar platform for malicious users referred as social spammers to overwhelm usual users with unwanted content. One efficient way for detection of social spammer is to construct a classifier based on social network and content information. However social spammers are adaptable and sophisticated to game the system with rapidly developing network and content patterns. The spammer detection is always a challenging issue on social network. The rigid anti spam norms have resulted in development of spammers. They look alike legal users who are difficult to recognize. In this study a novel spammer classification method based on LDA (Latent Dirichlet Allocation) a topic model is proposed. This method retrieves both the global and local data of topic distribution patterns which seizes the spamming essence. The method is tested on one dataset of benchmark and one self-gathered data set. This proposed approach outperforms other state of art approaches in terms of average FI-score. |
| Databáze: |
Supplemental Index |
| Externí odkaz: |
|
Full text from SpringerLink