Arabic Abstractive Summarization Using the Multilingual T5 Model

Abstract

Natural language processing (NLP) is a branch of artificial intelligence (AI) that enables computers to comprehend, generate, and manipulate human language. Natural language processing can interrogate the data with natural language text or voice. Abstractive Text Summarization is based on the Natural Language Processing technique that tries to provide new and more concise textual summaries for huge texts. Artificial intelligence and deep learning techniques are used in abstractive summarization to examine the text's essential information and create a new summary that conveys the content more concisely and accurately. This type of summary differs from normal extractive summarizing in that it can generate new summaries beyond simply extracting key lines from the source text. This study presents an abstractive Arabic summarization based on the Multilingual T5 (MT5) model AASMT5 to address these concerns. This technique is based on deep neural networks and models like transformers, which have changed the ability to summarize text. This research has explored the various types of summarizations and highlighted the significance of two prominent techniques: abstractive and extractive summarization. This research describes a complete process for developing an Arabic abstractive summarization model using the MT5 architecture. Experiments on different datasets show that this model achieves state-of-the-art results across MT5.