Spam Detection in Arabic Tweets Using Artificial Intelligence Techniques

Abstract

this study explores the application of artificial intelligence in Natural Language Processing (NLP) to categorize tweets and identify spam. Traditional methods for spam detection face significant delays and inefficiencies, often-taking months to analyze large volumes of data, and typically rely on supervised learning, which demands verified datasets. Addressing the need for more effective spam detection, especially in Arabic dialects, we took it upon ourselves to study the range of differences between traditional machine learning algorithms (SVM, RF…) and deep learning ones (LSTM, BILSTM, ARABET). The end goal of this study is taking the accuracy a step higher than what we currently have, while also maintaining a great efficiency score of spam detection, potentially improving message filtering and information management in communication systems.