A Comparison Between Ontology-Based and Translation-Based Semantic Search Engines for Arabic Blogs

Abstract

The information-retrieval community has focused on the retrieval of documents based on their Latin content, and have emphasised keyword-based approaches. These approaches are prone to low retrieval precision for Arabic content. The goal of our research is to develop a domain-dependent semantic-based search engine for Arabic blogs named Mudawanati, a process devised to increase the precision of results by combining natural language processing with ontologies. We have focused on the two main approaches adopted in the development of Arabic search engines: translation-based and Arabic ontology-based, and we investigated how they affect the accuracy of results returned. Prototypes are developed for the two approaches and experimental results are presented to reflect the performance and precision rates of both approaches. Precision is a measure of the importance of a semantic web document.