ドメインオントロジーを用いたデータマイニング向け層別化の一手法

抄録

Many kinds of technique for data mining have been suggested, and are actually utilized on various domains. However, most of the techniques support data mining process only from the side of mathematical task knowledge, including Statistics, Machine-Learning, and so on. To evaluate and interpret the results appropriately in the context of the backgrounds of the dataset, it is necessary to design and control the data mining based on domain knowledge. In this paper, we discuss stratification, which often depends on arbitrary decisions, and introduce our framework for the stratification by using a domain ontology.