Authors : *1P. V. Nimbalkar, P. K. Butey
Page Nos : 967-977
Description :
Automatic methods of measuring MT performance give us the ability to evaluate various features and present objective results without worrying about agreement rates of humans[1]. The problem of human agreement has been shifted to the set of references that are developed for a set of documents and each test system can then be evaluated objectively since the set of references are common to all systems[2].
Keywords: Machine Translation, Natural language processing.