Abstract: In the interoperation processing concerning multiple heterogeneous data sources, schema merging spanning them is one effective aspect in providing both the data querying/manipulation and the application program interfaces in a consistent, unified style. Until now there has been no perfect directive theory relevant yet. In this paper, on the foundation of object data model, the authors tackle the problem by setting forth from the features of attribute, the minimal unit in schema construction. The different attribute/type equivalencies are defined first in a bottom up way, and then the basic schema merging methodologies based on them are presented in a top down way, which explores a new approach to creating a consistent object view schema over multiple data sources.