A Framework for Arabic Tweets Multi-label Classification Using Word Embedding and Neural Networks Algorithms

Abstract

The need for classifying tweets is essential for many people like tourists, tourism companies and governments. In this paper, we propose a framework for Arabic Tweets multi-label classification using word embedding technique and deep leering algorithms. We built our dataset using 160k Arabic tweets gathered from Twitter. We compared two deep learning methods, Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN). Our results show that it is possible to classify tweets using our methodology without any significant difference in results of accuracy scores and hamming loss for both types of networks. The accuracy scores and hamming loss were nearly 90% and 0.02, respectively.