Paper: Example-Based Rescoring Of Statistical Machine Translation Output

ACL ID N04-4003
Title Example-Based Rescoring Of Statistical Machine Translation Output
Venue Human Language Technologies
Session Short Paper
Year 2004
Authors

Conventional statistical machine translation (SMT) approaches might not be able to find a good translation due to problems in its sta- tistical models (due to data sparseness dur- ing the estimation of the model parameters) as well as search errors during the decoding pro- cess. This paper1 presents an example-based rescoring method that validates SMT transla- tion candidates and judges whether the selected decoder output is good or not. Given such a validation filter, defective translations can be rejected. The experiments show a dras- tic improvement in the overall system perfor- mance compared to translation selection meth- ods based on statistical scores only.