A Comparative Analysis of Arabic Text Summarization Techniques: Evaluating Word Frequency, K-Means Clustering, and PageRank Algorithm

Abstract

This study explores methods to efficiently summarize extensive Arabic texts, addressing the growing need to condense large volumes of content across various fields. Three primary techniques are evaluated: Word Frequency Analysis, K-means Clustering based on Sentence Proximity, and the PageRank Algorithm. The research finds the PageRank Algorithm to be the most effective, delivering higher compression ratios while maintaining strong recall and precision metrics. In particular, the PageRank method achieved the highest compression ratio of 0.562 while maintaining a population standard deviation of 2.0, compared to other techniques. Evaluation metrics such as population standard deviation, F1 score, and compression ratio support these findings. The study also examines advanced approaches like fuzzy logic-based methods, transformers, and multi-document summarization, aiming to enhance Arabic text summarization.