Research on Definition Extraction Based on Over-Sampling Using Distance Distribution Information of Instances
Xu Pan, Hongbin Gu, Zhiqing Zhao · 2010
For the purpose of extracting definitions of all terms from aviation professional corpus, we introduce a classification method. This method includes a novel approach to over-sampling minority instance using distance distribution information, building balanced training set using random under-sampling majority instance, constructing aggregating classifier with C4.5 decision tree. This method achieves the best score with 65% in F1-measure and 78% in F2-measure. In the end, we analyse the influence of feature selection method on classification results.