Combination of Rough Set Theory and Maximum Entropy Model for Conjunctive Structure Detection in QA System
Shixi Fan, Xuan Wang, Xiaolong Wang · 2007
We introduce a combination model as conjunctive structures detection for question pre-processing in Q&A system. Conjunctive structures detection can be treated as a pattern recognition problem. The rough set theory is used for selecting effective features and the ME (maximum entropy) model is used for building a pattern classifier to get high accuracy. The training and testing data are collected from some discussion groups in the internet. A simple ME model is used for baseline system. The best Precision is 0.932 with 0.042 higher than the baseline system.