Paper: Model Combination for Machine Translation

ACL ID N10-1141
Title Model Combination for Machine Translation
Venue Human Language Technologies
Session Main Conference
Year 2010

Machine translation benefits from two types of decoding techniques: consensus decoding overmultiplehypothesesunderasinglemodel andsystemcombinationoverhypothesesfrom different models. We present model combina- tion, a method that integrates consensus de- coding and system combination into a uni- fied, forest-based technique. Our approach makes few assumptions about the underly- ing component models, enabling us to com- binesystemswithheterogenousstructure. Un- like most system combination techniques, we reuse the search space of component models, which entirely avoids the need to align trans- lation hypotheses. Despite its relative sim- plicity, model combination improves trans- lation quality over a pipelined approach of first applying consensus decoding to individ- ual systems, and then ...