Paper: Inference of Phrase-Based Translation Models via Minimum Description Length

ACL ID E14-4018
Title Inference of Phrase-Based Translation Models via Minimum Description Length
Venue Annual Meeting of The European Chapter of The Association of Computational Linguistics
Session Main Conference
Year 2014
Authors

We present an unsupervised inference pro- cedure for phrase-based translation mod- els based on the minimum description length principle. In comparison to cur- rent inference techniques that rely on long pipelines of training heuristics, this procedure represents a theoretically well- founded approach to directly infer phrase lexicons. Empirical results show that the proposed inference procedure has the po- tential to overcome many of the prob- lems inherent to the current inference ap- proaches for phrase-based models.