Applying question classification to Yahoo! Answers

• Published in: 2008 First International Conference on the Applications of Digital Information and Web Technologies (ICADIWT) pp. 229 - 234
• Main Authors: Blooma, M.J., Goh, D.H.-L., Chua, A., Zhiquan Ling
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.08.2008
• ISBN: 1424426235; 9781424426232
• DOI: 10.1109/ICADIWT.2008.4664350

Question classification is an important part in modern Question Answering systems. Most approaches to question classification are based on handcrafted rules. Recent studies classify simple questions using machine learning techniques and recommends SVM as on of the best performing classifiers. This study applies a hierarchical classifier based on the SVM machine learning algorithm on questions posed by users, drawn from Yahoo! Answers. The significance of this study is that we attempted to directly classify complex questions with multiple sentence questions posed by real users. We report the accuracy achieved using both a coarse-grained classifier and fine-grained classifier to illustrate the effectiveness of our approach on complex questions. We also present a confusion matrix to analyze the results made by our classifier.