Integrating machine translation with institutional repositories: a case study of university of Nairobi

Abstract

The general aim of this research is to study the viability and challenges that would face the use of machine translation for digital collections in the University of Nairobi Institutional Repository. The premise of the study is that as universities are churning out their output through IRs, these should be available in the language of the user. Since university funding is getting tighter, open source software should be considered for some tasks such as translation. The specific objectives of the study include: finding out the suitability and convenience of using machine translation in translating IR resources in the UoN repository; evaluating the most suitable open source machine translation system for translating IR resources from English to Kiswahili; identifying the benefits and/or constraints encountered when doing so, and; generally identifying the future role of machine translation in digital information dissemination. The exploratory nature of this study makes it be conducted using a qualitative approach. Purposive sampling is used to choose the texts for study. To examine the perception of quality of machine translation, two open source MT software are used and their quality of translation compared against each other by highlighting their errors. The data emanating from the study is analyzed to identify the translation strategies used by the MT software. It is hoped that the results of the research will galvanize efforts towards offering translation assistance to make IR resources more meaningful to all researchers. Perhaps by having a way through which a user can switch between different language versions one could be able to translate the resources into the language most desirable and therefore eliminate the processes involved in copying and pasting into a translation app or even trying to decipher the meaning.