Abstract Representation in Machine Translation

Abstract

This paper tries to introduce the notion of 'abstract representation', one of the key notions in Machine Translation (MT). This aim is going to be gained through giving some examples of abstract representation which have been found in different references. In the first systems of MT, experts in computer sciences and linguists tried to use 'abstract representation' to give a proper translation, but because of the complexity of this notion and the problems which early computers faced, some of them decided to forsake 'abstract representation' totally. They started using methods without abstract representation such as example-based and statistical approaches, but after a while, these methods, too, failed to solve the problem. Today again, the latest methods of machine translation are trying to apply 'abstract representation' besides new findings of other methods to solve the problems of machine translation.