TOPIC ANALYSIS OF SCHOLAR PAPER USING EARTH OBSERVATION VOCABULARY

摘要

This paper presents a pilot analysis for Asian Conference on Remote Sensing (ACRS) proceedings by using Earth Observation Vocabulary (EOV) which consists of 146 observation items and its definitions. Although the papers in ACRS proceedings are classified into official session topics, the topics change every year and also have wide varieties including academic theme or project-based categories. So only with the official classification, it is difficult to analyze over years. Therefore, in this paper we attempt to characterize the ACRS proceedings from the viewpoint of observation items in EOV. For the analysis, we use Naieve Bayes which is a popular text classification model. In this paper, firstly we evaluate the accuracy of the Naieve Bayes in order to find the suitable conditions. Subsequently, we apply the Naieve Bayes model for proceedings and calculate the similarity to EOV. The result shows the priority of observation items and the trend in ACRS.