A Methodology of Constructing Canonical Form Database Schemas in a Multiple Heterogenous Database Environment

Databases are usually developed independently by each group or organization to meet their own individual requirements. However, as interactions between groups and organizations become increasingly common, many applications end up requiring data not from one single database but from several related databases. Thus, users have to deal with databases whose designs have never been coordinated. Interoperability between these multiple heterogeneous databases has been a hot research topic. Previously in Lim et al. (1997), we proposed the Query Clearing House (QCH) model, which aims at assisting users in finding information in multiple heterogeneous databases. QCH plays the role of mediating between the end users' data needs and the local databases' data publication needs. The objective of this paper is to propose a uniform schema conversion method with which one can transform local relational database schemas into one cohesive format-i.e, into canonical form schema expressions. These canonical form expressions are then used as the basis for constructing two important resources for the QCH model: meta-data and mapping libraries. The meta-data includes not only the schematic information for the local databases, but also additional semantics that are not available in the database schemas themselves. The mapping library for each database has the mapping information between the general terms used in the meta-data and the specific terms used in the database schema so that a user's query can be systematically transformed into a database-specific query expression. We use L K ,which was initially presented in Shin (1991, 1994), as the representation formalism for expressing canoni-cal form schemas and the meta-data. L K is known for its flexible, descriptive power, which facilitates expressing general concepts at the desired level of granularity. The language is also known for its versatile association mechanism, which The Query Clearing House (QCH) model aims at achieving an ideal heterogeneous multiple database environment in which users can submit queries without concern for the location of the data or the specifics of the relevant database schemas. One important prerequisite for building such an environment is to organize local database schemas in a cohesive way so that a systematic method of determining data relevancy can be developed. In this paper, we propose a method for converting multiple local relational schemas into canonical form expressions. The resulting canonical form schema expressions include much more descriptive data semantics than what is offered by the original database schemas. These expressions also include information for mapping between …