Ensemble Machine Learning Algorithm for Telegram Spam Detection

Abubakar Hassan1*, Muhammad Abatcha1 and Emmanuel Gbenga Dada2

1Department of Computer Engineering, University of Maiduguri, Maiduguri, Borno State.
2Department of Computer Science, University of Maiduguri, Maiduguri, Borno State

*Corresponding author’s Email: abubakarhassan@unimaid.edu.ng, doi.org/10.55639/607.06050402


ABSTRACT

Telegram is one of the popular Instant Messaging (IM) platforms widely used today in the world. It is largely used due to its advantages of high speed, strong security and good privacy, key features for both public and private messaging. But telegram spam messages have become a significant concern for users which not only inundates user’s storage space but also exposes them to security threats, including malicious content and phishing attacks. This leads to the need to develop more effective spam detectors for the modern digital technology platforms. Machine learning algorithms have proved to be a robust approach for solving the problem of spam messages causing concerns to internet users. This paper proposed an ensemble machine learning algorithm for Telegram Spam Detection using Random Forests and Logistic Regression as base learners. Experiments were conducted in jupyter environment (Python 3) using dataset obtained from Kaggle. The models were evaluated using the metrics: accuracy, precision, recall, and the F1 measure, which offer a holistic appraisal of the model’s efficacy. Experimental results demonstrated that the proposed ensemble model and the Random Forests algorithm achieved 94% accuracy compared to the Logistic Regression model (93%) on the benchmark dataset.

Keywords:

Ensemble Learning,
Machine Learning,
Random Forests,
Logistic
Regression,
Telegram