English-amharic Document Translation Using Hybrid Approach

Machine Translation (MT) is a sub field of Natural Language Processing (NLP) thatrninvestigates the use of computer software to translate text or speech from one naturalrnlanguage to another. Recently, in Ethiopia English documents are translated to Amharicrnusing human translators, because of the absence of Automated Translation System. Duernto this, the process of document translation is so expensive, challenging, unsecured andrntime consuming. To solve those problems statistical based approaches were proposed torntranslate English to Amharic. However, the approaches have accuracy and understandabilityrnissues. In order to solve these problems we propose a Hybrid approach machinerntranslation (HMT) system that combines Statistical and Rule Based Machine Translationrnapproaches.rnIn this case using the hybrid approach achieved better accuracy and found to be advancedrnfor English - Amharic machine translation system over SMT approach. The proposedrnhybrid approach achieved 15% and 20% accuracy improvement for simple and complexrnsentences over statistical machine translation approach. This study also identified preprocessingrnthe inputs of SMT is more suitable to improve accuracy for complex sentencesrnwhile post-processing the outputs of SMT is more suitable for simple sentences.

Get Full Work

Report copyright infringement or plagiarism