Paper: Towards Heterogeneous Automatic MT Error Analysis

ACL ID L08-1145
Title Towards Heterogeneous Automatic MT Error Analysis
Venue International Conference on Language Resources and Evaluation
Session Main Conference
Year 2008

We describe a set of experiments to explore statistical techniques for ranking and selecting the best translations in a graph of translation hypotheses. In a previous paper (Carl, 2007) we have described how the hypotheses graph is generated through shallow mapping and permutation rules . We have given examples of its nodes consisting of vectors representing morpho-syntactic proper sentence, including, among other things, n-gram language models. In contrast to most purely statistical MT systems1 , we use rule-based methods to generate an AND/OR graph of translation hypotheses (Carl 2007) and a beam search algorithm to traverse the graph and feature functions to rank the paths, as e.g. in the MISTRAL system (Patry et 1A number of recent SMT architectures are described ht...