Arabic Text Categorization Based on Arabic Wikipedia

摘要

This article describes an algorithm for categorizing Arabic text, relying on highly categorized corpus-based datasets obtained from the Arabic Wikipedia by using manual and automated processes to build and customize categories. The categorization algorithm was built by adopting a simple categorization idea then moving forward to more complex ones. We applied tests and filtration criteria to reach the best and most efficient results that our algorithm can achieve. The categorization depends on the statistical relations between the input (test) text and the reference (training) data supported by well-defined Wikipedia-based categories. Our algorithm supports two levels for categorizing Arabic text; categories are grouped into a hierarchy of main categories and subcategories. This introduces a challenge due to the correlation between certain subcategories and overlap between main categories. We argue that our algorithm achieved good performance compared to other methods reported in the literature.