How to Know the Best Machine Translation System in Advance before Translating a Sentence
Bibekananda Kundu, Sanjay Kumar Choudhury · International conference natural language processing · 2014
The aim of the paper is to identify a machine translation (MT) system from a set of multiple MT systems in advance, capable of producing most appropriate translation for a source sentence. The prediction is done based on the analysis of a source sentence before translating it using these MT systems. This selection procedure has been framed as a classification task. A machine learning based approach leveraging features extracting from analysis of a source sentence has been proposed here. The main contribution of the paper is selection of sourceside features. These features help machine learning approaches to discriminate MT systems according to their translation quality though these approaches have no idea about working principle of these MT systems. The proposed approach is language independent and has shown promising result when applied on English-Bangla MT task.