Bangla to English Machine Translation using Fuzzy Logic Md. Musfique Anwar Computer Science and Engineering Department, Jahangirnagar University, Bangladesh Email: manwar@juniv.edu Abstract- Transfer in machine translation (MT) plays an important role for producing correct output. This paper presents a technique to address about structural and lexical mappings from different types sentences of Bangla language for machine translation. Machine translation requires analysis, transfer and generation steps to produce target language output from a source language input. This paper deals with the syntactic transfer and generation for Bangla simple, complex and compound sentences into English. Structural representation of Bangla sentences encodes the information of Bangla sentences and a transfer module has been designed that can generate the English sentences from a corpus based automatic Bangla machine translator using Fuzzy logic. The effectiveness of this method has been justified over the demonstration of different Bangla sentences and the success rates in all cases are over 90%. Keywords- Machine Translation, Structural representation, Fuzzy Logic, Corpus. 1. INT RODUCT ION Machine translation (MT) refers the translation from one natural (source) language to another (target language). It is an important area of Natural Language Processing (NLP). MT is a challenging job due to building up a successful translator for producing exact target language output from a source language. At a minimum, transfer systems require monolingual modules to analyze and generate sentences, and transfer modules to relate equivalent translation representations of those sentences. To interpret language we need to determine a sentence structure. To do this we must know the rules of how language is organized and have an algorithm to analyze language given on those rules. Parsing serves in language to combine the meanings of words and phrases. A grammar captures the legal structure in a language and thus allows a sentence to be analyzed. Parsing a sentence then involves finding a possible legal structure for sentence. The result is usually a tree (referred to as parse tree) or structural representation (SR) [1]. Analysis and generation are two major phases of machine translation. There are two main techniques concerned in analysis phase. These are: Morphological Analysis Morphological analysis is the determination of the grammatical categories (noun, verb, adjective, adverb, etc) of the words of sentences. That means, it incorporates the rules by which the words are analyzed. To give an English example, the words analyzes, analyzed and analyzing might all be recognized as having the same stem analyze and the common endings –s, -ed, -ing. The result of morphological analysis then is a representation that consists of both the information provided by the dictionary and the information contributed by the affixes. Morphological information of words are stored together with syntactic and semantic information of the words. Syntactic Analysis Syntactic Analysis involves the inclusion of a few rearrangement rules in the basic word by word approach such as the inversion of ‘noun -adjective’ to ‘adjective-noun’. Rearrangement rules may take into account fairly long sequences of grammatical categories, but they do not imply any analysis of syntactic structure like the identification of a noun phrase. Complete syntactic analysis involves the identification of relationships among phrases and clauses within sentences. Syntactic analysis aims to identify three basic types of information about sentence structure: 1) The sequence of grammatical elements, e.g. sequences of word classes: article + verb + preposition ………, or of functional elements: subject + predicate. These are linear (or precedence) relations. International Journal of Computer Science and Information Security (IJCSIS), Vol. 16, No. 11, November 2018 156 https://sites.google.com/site/ijcsis/ ISSN 1947-5500