Interoperable Query Processing among Heterogeneous Databases

The proliferation of database systems based on diierent data models and query languages has created the need for techniques that support interoperability so that information may be shared among the database systems. This dissertation proposal describes a technique to support interoperable query processing when multiple heterogeneous databases are accessed. We will focus on the problem of supporting query transformation transparently, so a user can pose queries locally, without any need of global knowledge about diierent data models and schema. To support interoperable query transformation, we need to resolve the connicts (i.e., het-erogeneities) between diierent databases. The connicts exist because each database has been designed and populated independently of one another, so semantically equivalent concepts may have been deened in many diierent ways in each model/schema. Furthermore, we also need to consider the diierent query language utilized by each database to provide interoperability. We propose two kinds of parameterized canonical representations as a means of classifying and resolving heterogeneities. 1 The rst canonical form resolves heterogeneity based on the query language. The second canonical form resolves representation heterogeneity based on using two diierent schema. We initially focus on object and relational schema. The query transformation mechanism can operate on these canonical forms to produce the proper query for a heterogeneous remote database. We will describe an architecture for supporting interoperability and certain functional modules. The rst module described is an extractor module (EM) which parses a source query and extracts semantics, corresponding to some primitive expressions in the query which are represented using a canonical form. The second is a heterogeneous mapping module (HTM) which maps among entities in diierent schema; this module uses mapping rules which reeect global knowledge of the models, schema, and query languages. We use F-logic, a high-level logic representation , to represent the global dictionary knowledge in a canonical form. We will provide some examples of transforming queries from a relational schema to an equivalent object schema. This proposal outlines several topics that need to be studied in order to solve the problems described above. The rst area of investigation is the extension of the current canonical representation for source and target (transformed) queries to cover more constructs in the SQL and XSQL query languages. A second area is the extension of the canonical representation for mapping information to cover further solutions for schema connict, in particular, connicts among several object schemas. We describe our current …