A rule based approach for implementation of bangla to english translation

Abstract

An initiative to model a Bangla to English (B2E) translation using Natural Language Processing (NLP) was proposed in our previous research. Here we implemented the model with a lot of modifications. A very successful translator is observed in Anubadok Online which is based on Penn Treebank annotation system and it can only translate English sentences to Bangla. Penn Tree Bank is the collection of English corpus, so Bangle linguistic processing is not observed there. Bangla is an Irregular Language. In our previous research, we proposed a case structure analysis for verb. There are a lot of influences of case in Bangla language. The relationship between verb and case elements is an important issue for Bangla language. But in our current implementation we used rule based approach. For Bangla-English translation first we performed morphological analysis for Bangla then we used rule based analysis where we considered a limited feature of case analysis. After that, using a dictionary we translate bangle words into English. To make an English sentence, we considered English SVO grammatical rules. Our current system is successfully implemented for the translation of Assertive-Affirmative, Negative and Interrogative sentences.