An Evaluation methodology for English to Sinhala machine translation

Abstract

This paper presents evaluation methodology for English to Sinhala machine tran slation system. The English to Sinhala machine translation system has been developed by using Multi Agent Approach and powered through the concept of “Varanegeema”. Translation system works through the communication among nine agents namely English Morphological Analyzer Agent, English Parser Agent, English to Sinhala Base Word Translator Agent, Sinhala Morphological Generator Agent, Sinhala Parser agent, Transliteration agent, Intermediate Editor agent, Message Space Agent and Request agent. The evaluation was conducted through three steps. As the first step, evaluation was conducted through the white box testing approach and tested each module in the machine translation system through the developed testing tools. Then, evaluated the system performance and calculated the error rate through the result of the evaluation test bed. Finally, Intelligibility and the Accuracy test will be conducted through the human support. The experimental result shows 89% accuracy of the overall system and 7.2% word error rate and the 5.4% sentence error rate. Details of the evaluation and results are given in the paper