Paper: Minimum Bayes Risk based Answer Re-ranking for Question Answering

ACL ID P13-2075
Title Minimum Bayes Risk based Answer Re-ranking for Question Answering
Venue Annual Meeting of the Association of Computational Linguistics
Session Short Paper
Year 2013
Authors

This paper presents two minimum Bayes risk (MBR) based Answer Re-ranking (MBRAR) approaches for the question answering (QA) task. The first approach re-ranks single QA system?s outputs by using a traditional MBR model, by mea- suring correlations between answer can- didates; while the second approach re- ranks the combined outputs of multiple QA systems with heterogenous answer ex- traction components by using a mixture model-based MBR model. Evaluation- s are performed on factoid questions se- lected from two different domains: Jeop- ardy! and Web, and significant improve- ments are achieved on all data sets.