Adapting hybrid machine translation techniques for cross-language text retrieval system

This research work aims in developing tamil to english cross - language text retrieval system using hybrid machine translation approach. the hybrid machine translation system is a combination of rule based and statistical based approaches. in an existing word by word translation system there are lot of issues and some of them are ambiguity,out-of-vocabulary words,word inflections,and improper sentence structure. to handle these issues,proposed architecture is designed in such a way that,it contains improved part-of-speech tagger,machine learning based morphological analyser,collocation based word sense disambiguation procedure,semantic dictionary,and tense markers with gerund ending rules,and two pass transliteration algorithm. from the experimental results it is clear that the proposed tamil query based translation system achieves significantly better translation quality over existing system,and reaches 95.88% of monolingual performance. © school of engineering,taylor’s university.