Performance of the Google Desktop, Arabic Google Desktop and Peer to Peer Application in Arabic Language

TL;DR

This paper evaluates and improves the performance of Google Desktop and peer-to-peer applications for Arabic language document retrieval, addressing morphological complexities by considering root words in search and indexing.

Contribution

It proposes an update to Google Desktop to account for Arabic roots and compares peer-to-peer indexing methods with and without root consideration.

Findings

Root-based indexing improves retrieval accuracy.

Google Desktop's performance is enhanced with root-aware search.

Peer-to-peer application performance varies with indexing method.

Abstract

The Arabic language is a complex language; it is different from Western languages especially at the morphological and spelling variations. Indeed, the performance of information retrieval systems in the Arabic language is still a problem. For this reason, we are interested in studying the performance of the most famous search engine, which is a Google Desktop, while searching in Arabic language documents. Then, we propose an update to the Google Desktop to take into consideration in search the Arabic words that have the same root. After that, we evaluate the performance of the Google Desktop in this context. Also, we are interested in evaluation the performance of peer-to-peer application in two ways. The first one uses a simple indexation that indexes Arabic documents without taking in consideration the root of words. The second way takes in consideration the roots in the indexation of Arabic documents. This evaluation is done by using a corpus of ten thousand documents and one hundred different queries.