Hierarchical Classification of Heterogeneous Data

T. Furukawa and M. Kuzunishi (Japan)


Classification Hierarchies, Data Modeling, Data Semantics, Information Retrieval


Hierarchical classification based on semantics is efficient approach to organize data, which usually assume that the semantics levels of data are the same. This paper discusses the schema of classification hierarchies for heterogeneous data, objects at various semantics levels. Since each object of a class is usually expected to belong to one child class (sufficiency) and not to multiple child classes (exclusivity), objects appear only in terminal classes by sufficiency and there is no duplicate object by exclusivity. If objects are heterogeneous, however, there may be objects which can not be classified into lower classes. There can be two in terpretations for "the objects of a class," the objects clas sified into the class and the objects whose semantics is the same as the semantics of the class. Classification hierar chies proposed in this paper cope with these problems by allowing non-terminal classes to keep objects and dividing the objects in a class according to their semantics levels. By the proposed classification hierarchies, users' requirements concerned with semantics levels are also satisfied.

Important Links:

Go Back