Machine and Deep Learning Algorithms for Twitter Spam Detection

Abstract

Twitter allows users to send short text-based messages with up to 280 characters which is called “tweets”. The reputation of Twitter attracts the spammers to spread malevolent programming through URLs attached in tweets. Twitter spam has become a critical problem. Spam refers to a variety of prohibited behaviours that violate the Twitter rules. In this paper, different machine and deep learning algorithms are used to detect if the tweet is spammer or not. The performance of six machine learning algorithms, namely Random Forest (RF), Naive Bayes (NB), Bayesian Network (BN), Support Vector Machine (SVM), K-Nearest Neighbour (KNN), and Multi-Layer Perceptron (MLP) and one deep learning algorithm which is Recurrent Neural Network (RNN) are evaluated. Different test options are used, namely cross validation and percentage split tests. Results show that RF predicts the best result with lowest error rate and highest classification accuracy rate with different test options comparing to all algorithms.