Gaming fluency: evaluating the bounds and expectations of segment-based translation memory

Authors:
John Henderson;William Morgan
Affiliations:
The MITRE Corporation, Bedford, MA;The MITRE Corporation, Bedford, MA
Venue:
ParaText '05 Proceedings of the ACL Workshop on Building and Using Parallel Texts
Year:
2005

Citing 4
Cited 0

Comparing representations in Chinese information retrieval

Proceedings of the 20th annual international ACM SIGIR conference on Research and development in information retrieval
Towards a unified approach to memory- and statistical-based machine translation

ACL '01 Proceedings of the 39th Annual Meeting on Association for Computational Linguistics
The Alignment Template Approach to Statistical Machine Translation

Computational Linguistics
ORANGE: a method for evaluating automatic evaluation metrics for machine translation

COLING '04 Proceedings of the 20th international conference on Computational Linguistics

Quantified Score

Hi-index	0.00

Visualization

Abstract

Translation memories provide assistance to human translators in production settings, and are sometimes used as first-pass machine translation in assimilation settings because they produce highly fluent output very rapidly. In this paper, we describe and evaluate a simple whole-segment translation memory, placing it as a new lower bound in the well-populated space of machine translation systems. The result is a new way to gauge how far machine translation has progressed compared to an easily understood baseline system. The evaluation also sheds light on the evaluation metric and gives evidence showing that gaming translation with perfect fluency does not fool bleu the way it fools people.