Arabic interactive cross-language information retrieval via natural language processing

The aim of the research is to test the possibilities of improving the quality of retrieval results in two different environments: (a) in monolingual information retrieval (IR) by applying morphological analysis module for Arabic, and (b) in testing dictionary-based methods in cross-language information retrieval (CLIR) with Arabic as a source and target language besides English. The Arabic language has received little attention in IR research.; In this dissertation we describe an English-Arabic/Arabic-English complete interactive cross-language and monolingual retrieval system that takes in Arabic queries as well as English queries and retrieves relevant documents in Arabic or English from the corresponding document collection. Then, we present the complete interactive system, through which a user is able to select a source and target languages, enter the appropriate query, review the retrieval results of the query translation, and conduct retrieval using the revised queries based on relevant documents. The system uses a combination of automatic and user-assisted methods to build and improve cross-language queries. The system has been successfully tested on Qatari Al-Raya and United Arab Emirates Al-Watan news articles and a set of 20 queries. The test for Arabic used an Arabic document database and a morphological analyzer for Arabic. Then, were determined effective ways to maximize a user's performance interactively. The query translation component of our system was evaluated with a set of queries using full words (retrieval without morphology) and a set of queries using stems (retrieval with morphology). The set of queries used in the evaluation have been all manually translated into Arabic and the translations produced by the system are good translations in 80% of the cases. Next, we briefly present some results on evaluations of the system. And finally, we describe the design and implementation of a Multilingual Dictionary Browser that has been tested on Arabic, English, and Spanish.