Paper: Exploiting Social Q&A Collection in Answering Complex Questions

ACL ID W10-4117
Title Exploiting Social Q&A Collection in Answering Complex Questions
Venue Joint Conference on Chinese Language Processing
Session Main Conference
Year 2010
Authors

This paper investigates techniques to au- tomatically construct training data from social Q&A collections such as Yahoo! Answer to support a machine learning- based complex QA system1. We extract cue expressions for each type of question from collected training data and build question-type-specific classifiers to im- prove complex QA system. Experiments on 10 types of complex Chinese ques- tions verify that it is effective to mine knowledge from social Q&A collections for answering complex questions, for in- stance, the F3 improvement of our sys- tem over the baseline and translation- based model reaches 7.9% and 5.1%, re- spectively.