A feature selection method for comparision of each concept in big data In this paper, we present a new feature selection method for comparison of each event in big data. The method selects appropriate features for comparing or evaluating each event when we prepare an event set. There are massive data on the Internet as big data. We have been able to retrieve appropriate datasets from them by using given keywords. However, it is still difficult to use these data to compare some concepts. For example, when you would like to compare the difference between your company and a rival company, it is difficult to compare them, because evaluation axes are needed to compare them. This method extracts evaluation axes as features in a data-driven manner. We can compare them by using the extracted axes as the features. This means that the system automatically selects the appropriate features which we should focus on when we compare them.