FEMTI Taxonomy for Evaluating Machine Translation Models

The aim of our experiment is to present interrelationship between some of the characteristics of MT (Machine Translation) systems and the quality of their output. The emphasis is placed on the machine translation models that determine output quality of these systems. In order to achieve our goal, a couple of MT systems have been tested on different text types prepared in two languages: English and French. The procedure of our evaluation is in strict conformity with a Framework for the Evaluation of Machine Translation in ISLE (International Standards for Language Engineering)[7]. In our study, we also consider user population as a factor that indicates different levels of the adequacy-oriented output to the users' needs. In conclusion, the backbone is made of the further study on incorporation of the database types and both the target as well as the source-language formalism which should be critical for designing universally usable MT system.