MACHINE LEARNING ALGORITHMS FOR TELEGRAM SPAM FILTERING
Keywords:
Extreme Gradient Boosting, Light Gradient Boosting Machine, CatBoost, Support Vector Machine, K-Nearest NeighbourAbstract
With unprecedented usage of social media applications to interact in virtual communities, bad entities can now use these platforms to spread their malicious activities such as spam, hate speech, and even phishing to a very large population. Especially, Telegram is suitable for these kinds of activities because it is a new cloud-messenger that is highly popular among bloggers and media around the world, established by Pavel Durov in 2013. As a result, it is necessary for social media platforms to develop algorithms to filter these malicious contents. This paper employs Machine learning algorithms to filter spam messages in Telegram. Dataset obtained from Kaggle was used for the experiments in this paper. Five machine learning models were applied, namely, Extreme Gradient Boosting (XGB), Light Gradient Boosting Machine (LGBM), CatBoosting, Support Vector Machine (SVM) and K-Nearest Neighbours (KNN). Experimental results showed that SVM outperforms other machine learning models used for the study with a classification accuracy of 94%. This is an indication that SVM is a promising algorithm for Spam filtering in Telegram if adopted.
Published
How to Cite
Issue
Section
FUDMA Journal of Sciences
How to Cite
Most read articles by the same author(s)
- Abdullahi Bashar Abubakar, Danlami Gabi, Muhammad Garba, Nasiru Muhammad Dankolo, Abubakar Hassan, HYBRID PREDICTIVE MODEL FOR STUDENTS’ ACADEMIC PERFORMANCE BASED ON MACHINE LEARNING APPROACH , FUDMA JOURNAL OF SCIENCES: Vol. 9 No. 4 (2025): FUDMA Journal of Sciences - Vol. 9 No. 4