Not Perfect But Good Enough: Experiences Using Machine Translation In A Multi-Lingual Institution
Machine translation is the process whereby a computer programme translates text from one natural language, the source language, into another, the target language. The number of paradigms for automatic translation has escalated during the past decade. Rule-based machine translation is characterised by linguistic rules used in translation, while statistic machine translation uses a highly developed theory of probabilities and supposition estimation. Corpus-based machine translation was an alternative way to address the knowledge acquisition problem of rule-based machine translation. Both memory-based and example-based machine translation systems use a database from which the translation system retrieves previously translated examples. This database can be compiled using different methods and various matches are retrieved during the translation process. Language management systems can be used to identify specialised terms for e.g. computers, law, the library and medicine. The EtsaTrans translation programme is a hybrid example-based machine translation system that uses a corpus and word list to translate a document from the source text into the target text. The EtsaTrans translation programme is being developed at the UFS. This paper discusses the experiences implementing machine translation as a means of making documents easily and speedily available in a mulit-language institution.
Keywords: Machine translation
Deputy Director, Library and Information Services, University of the Free State