Machine Learning for Imbalanced Datasets of Recognizing Inference in Text with Linguistic Phenomena

Recognizing inference in text (RITE) plays an important role in the answer validation modules for a Question Answering (QA) system. The problem of class imbalance has received increased attention in the machine learning community. In recent years, several attempts have been made on the linguistic phenomena analysis, however, little is known about the effects of imbalanced datasets with linguistic phenomenon in recognizing inference in text. The objective of this paper is to provide an empirical study on learning imbalanced datasets of recognizing inference in text with linguistic phenomena for a better understanding of the effects of imbalanced datasets with linguistic phenomenon in recognizing inference in text. In this paper, we proposed an analysis of imbalanced datasets of recognizing inference in text with linguistic phenomena using NTCIR 11 RITE-VAL gold standard dataset and development dataset. The experimental results suggest that the distribution of imbalanced datasets of recognizing inference in text with linguistic phenomenon could be dramatically varied on the performance of a machine learning classifier.