Arabic text summarization using pagerank and word embedding algorithms

Arabic is one of the world's most frequently spoken languages, with over 200 million people using it as their first language, and it is the official language of 26 nations. Although Arabic text summarization (ArTS) has increased in popularity in recent years, the quality of current ATS systems need improvement. Graph-based techniques on Arabic natural language processing have clearly gained popularity in recent years. Because of their ability to arrange large and difficult structures into standard and formal ways, graphs may be used and developed in a helpful way to assist in conquering and minimizing Arabic language challenges. This study proposed a single-document Graph-based Extractive Arabic Text Summarization (GEATS). The PageRank method is used, along with word embedding. The similarity of any two sentences is calculated by ranking the sentences based on cosine similarity. The final score for each sentence is determined using PageRank scoring. Then, the summary includes the sentences with the highest ratings taking into account the compression ratio, which is 40% of the document's sentences. The EASC Corpus is used as a standard corpus to test the performance of this technique. ROUGE-1, ROUGE-2, and BLUE metrics are also employed in the evaluation process. The findings demonstrated that the proposed strategy outperforms state-of-the-art approaches.