A Proposed Clustering Technique for Arabic Text Summarization

Since the internet revolution, the content of information is growing quickly in this digital world. Therefore, the need for techniques to benefit from that amount of information becomes very required. Automatic Text Summarization (ATS) is one of these techniques which is derived from the Text Mining area. It aims to shorten the size of the original text as much as possible. Consequently, we propose a Clustering-Based Technique for Arabic Text Summarization. The framework for this technique combines the cosine measure and phrase to cluster phrases using the k-means algorithm. Then, to rank these phrases, the framework uses a Modified Page Rank algorithm (MPR) with the value of the position sentences. Finally, it generates a summary that includes the most important sentences. This Technique of Arabic Text Summarization could produce a complete summary of Arabic documents, including most of the ideas of the original text. The evaluation process is done using a dataset derived from different domains, like education, sport, religion, music, and environment. The experiment results prove that the proposed approach gives better performance than previous works in the same domain.