An Enhanced Text-Classification-Based Arabic Information Retrieval System

Abstract

This chapter presents enhanced, effective and simple approach to text classification. The approach uses an algorithm to automatically classifying documents. The main idea of the algorithm is to select feature words from each document; those words cover all the ideas in the document. The results of this algorithm are list of the main subjects founded in the document. Also, in this chapter the effects of the Arabic text classification on Information Retrieval have been investigated. The goal was to improve the convenience and effectiveness of information access. The system evaluation was conducted in two cases based on precision/recall criteria: evaluate the system without using Arabic text classification and evaluate the system with Arabic text classification. A chain of experiments were carried out to test the algorithm using 242 Arabic abstracts From the Saudi Arabian National Computer Conference. Additionally, automatic phrase indexing was implemented. Experiments revealed that the system with text classification gives better performance than the system without text classification.