Paper: Confidence-based Active Learning Methods for Machine Translation

ACL ID W14-0312
Title Confidence-based Active Learning Methods for Machine Translation
Venue EACL Workshop on Humans and Computer-assisted Translation
Session
Year 2014
Authors

The paper presents experiments with ac- tive learning methods for the acquisition of training data in the context of machine translation. We propose a confidence- based method which is superior to the state-of-the-art method both in terms of quality and complexity. Additionally, we discovered that oracle selection tech- niques that use real quality scores lead to poor results, making the effectiveness of confidence-driven methods of active learn- ing for machine translation questionable.