Paper: A Principled Approach to Context-Aware Machine Translation

ACL ID W14-1013
Title A Principled Approach to Context-Aware Machine Translation
Venue Workshop on Hybrid Approaches to Translation
Session
Year 2014
Authors

This paper presents a new principled approach to context-aware machine translation. The proposed approach reformulates the posterior probability of a translation hypothesis given the source input by incorporating the source-context information as an additional conditioning variable. As a result, a new model component, which is referred to as the context-awareness model, is added into the original noisy channel framework. A specific computation- al implementation for the new model component is also described along with its main properties and limitations.