Intelligence Information Retrieval System Modeling for Afaan Oromo

Abstract

Due to today’s information overload, finding and retrieving desired data or information has become difficult. Due to this new phenomenon, users will have a harder time locating and retrieving relevant information. It is now the norm rather than the exception to have data and information in multiple languages. It is critical in Ethiopia, where huge amounts of Afaan Oromo data are produced daily. With massive document collections come archival and search issues, putting these data and information characteristics to the test. Afaan Oromo users could easily search for and retrieve data that was relevant to their needs and interests using a hybrid information retrieval system developed for Afaan Oromo. Performance of information retrieval systems (IRS) has been under-researched in the Afaan Oromo linguistic domain. The efficiency of these systems has been found to be significantly lacking. For these issues, the AOIR system’s performance was improved by integrating various IR system approaches. The prototype includes the current indexing and searching subsystems. On-line news articles (Oromia Broadcasting Network, VOA Afaan Oromo), websites, books, and the Afaan Oromo Bible were used to gather 1000 Afaan Oromo text documents for the experiment. To identify terms and vocabulary that contained content, the text was subjected to text operations such as tokenization, normalization, stop-word removal and stemming and calculated using tf-idf term weighting scheme. After the experimental analysis in Python 3.7, precision, recall, and F-measure were all greater than 96.6 percent. The polysemy issue continued to affect the system’s overall performance. The paper also advocates for more work on system performance optimization to advance the AOIR.