Name: Aram Ibrahim Al-Anazi
Detection of Adult Content in Arabic Tweets Using Machine Learning Models
Abstract:
Deep learning (DL) and machine learning (ML) signaled a turning point in content moderation. This work addresses particular linguistic and cultural issues by evaluating the performance of several machine learning and deep learning models in spotting adult content in Arabic tweets. We implemented and compared Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory networks (LSTM), and AraBERT using a 33,691 Arabic tweet dataset. Data was extensively preprocessed—cleansing, tokenizing, segmenting into training, validation, and test sets among other things. Model efficacy was evaluated using performance criteria including accuracy, F1 score, and confusion matrices. Arabert proved best in capturing spatial patterns for content classification and attained the highest accuracy—100%). With accuracies of 94.27% and 94.22%, respectively, CNN and RNN also performed well; LSTM obtained an accuracy of 88.37%. These results highlight AraBERT’s effectiveness in content moderation within Arabic digital platforms, so promoting safer online environments.
Keywords:
Arabic Tweets; AraBERT; Convolutional Neural Networks (CNN); Long Short-Term Memory (LSTM); Natural Language Processing (NLP); Recurrent Neural Networks (RNN); Text Classification
Download file pdf:
