A Deep Learning Model for an Arabic Text Summarization

Abstract

Recently, with the availability of a vast amount of information and texts on the Internet, the comprehensive reading of long texts has become difficult and complicated. Based on this, summaries have become crucial for extracting the essence of textual documents. This study proposes a model for the automatic generation of summaries for Arabic texts based on the mBART deep learning model. Our methodology aims to produce summaries that synthesize the main ideas of the texts while maintaining coherence with the original text. The XLSum dataset for Arabic was used to train and evaluate our model, and its performance was also measured using ROUGE metrics, which assess the quality of the summaries by comparing the similarity between the generated summaries and human references. The results highlight the promising performance of our model, paving the way for future advancements in the field of automatic summarization of Arabic texts.