Query Expansion For Afaan Oromo-English Cross Lingual Information Retrieval System

Abstract

Cross lingual information retrieval is very important to retrieve free text document which is different from query language. In traditional information retrieval system it is only possible to access the system?�?s document by one language. Today especially English language is the dominant technology language i.e. the documents are stored by English language which can be assessable only by English query. So this is the problem for the people like Afaan Oromo native people who want to assess the data but not have the ability to create query using English language. To solve the problem query translation for Afaan Oromo English cross lingual information retrieval system study have been conducted. So, in Afaan Oromo English cross lingual information retrieval system English document have been assessable by Afaan Oromo queries by the help of dictionary based approach. In this research work we used dictionary based approach to develop the prototype of Afaan Oromo English cross lingual information retrieval system. The reason we used this approach is it is easy to construct dictionary to translate Afaan Oromo words to English words. In addition to dictionary based approach we applied query expansion on this system. In this study we used query expansion in the procedure of reformulating user?�?s query to expand by using the method of thesaurus which contains the translation of Afaan Oromo words to English words. The method we used to construct thesauri is by using synonym words which are similar with each other. So to increase the performance of the prototype developed the synonym of Afaan Oromo words have been translated to single English word or single Afaan Oromo word have been translated to synonym English words or synonym Afaan Oromo words have been translated to synonym English words. This all used as a query expansion for the system during query time. To evaluate this system we used precision and recall evaluation method and we get good values 75% and 73% respectively. However, the result of the system is courageous it can be increased more than this if the problem of phrasal translation, words out of dictionary and inflection of words are solved.