Paper: Controlled Ascent: Imbuing Statistical MT with Linguistic Knowledge

ACL ID W13-2809
Title Controlled Ascent: Imbuing Statistical MT with Linguistic Knowledge
Venue Workshop on Hybrid Approaches to Translation
Session
Year 2013
Authors

We explore the intersection of rule-based and sta- tistical approaches in machine translation, with a particular focus on past and current work here at Microsoft Research. Until about ten years ago, the only machine translation systems worth using were rule-based and linguistically-informed. Along came statistical approaches, which use large cor- pora to directly guide translations toward expres- sions people would actually say. Rather than mak- ing local decisions when writing and conditioning rules, goodness of translation was modeled numer- ically and free parameters were selected to opti- mize that goodness. This led to huge improvements in translation quality as more and more data was consumed. By necessity, the pendulum is swing- ing towards the inclusion of linguistic features in MT...