In this investigation, we discuss how to make conceptual formation from databases as background knowledge. More specifically given a set E of objects e_1, ..., e_n, we generate conceptual description to capture the meaning that E has. Database schemata contain common knowledge that are carried by all the instances. However, they had been designed much before various instances are generated, then the schemata sometimes couldn't capture the exact meaning of the instances. To adjust such conceptual differences, we utilize relevant information(attributes, attribute values)to make suitable conceptual description by means of database schemata and their expressions. In this work, we introduce Conditional Sum of Product(CSOP) expressions on object types and discuss how to generate conceptual description based on extended decision tree technique. Also, we show some experimental results.

本文データは学協会の許諾に基づきCiNiiから複製したものである

301 学術雑誌論文 ＞ 自然科学