Font Size: a A A

Managing structural heterogeneity

Posted on:1995-07-11Degree:Ph.DType:Thesis
University:The University of Wisconsin - MadisonCandidate:Miller, Renee JeanFull Text:PDF
GTID:2479390014490959Subject:Computer Science
Abstract/Summary:
As information systems evolve, we are increasingly faced with the problem of finding flexible and painless ways of permitting new applications to use legacy data, that is data originally designed or structured without consideration for the applications or tasks we now have at hand. Such data may be stored in a data model that is foreign to new applications and their users. Even when the data model is unchanged, the choice of modeling constructs, the design methodology, and the way data is to be viewed may differ in the legacy system. Clearly, we cannot require that all new applications be constrained to use the modeling choices of the past. Neither is it feasible, in many cases, to migrate all data to a new system.;To address this problem, we need the ability to create a schema tailored to the needs of a specific application that can be used to access data stored under one or more legacy schemas. Furthermore, we need techniques that allow queries against the new schema to be automatically translated into queries against the legacy schema(s).;Two schemas that meet our requirements (in that one schema can be used as a view to access data stored under the other) will differ structurally in that they may use different data models, or different constructs of the same data model to represent related information. In this thesis, we develop techniques for managing such structural heterogeneity in schemas. We address the problem of testing if two arbitrary, structurally heterogeneous schemas can in fact represent the same data and if one can be used as a view onto data stored under the other. Additionally, we develop techniques that permit the creation, modification and maintenance of such schemas. We develop rigorous solutions that guarantee that any schema we create can indeed meet our requirements of allowing access to legacy data. Furthermore, we develop and implement practical solutions that apply to schemas that occur in practice, are efficient, and may easily be combined with nonrigorous solutions when such alternatives are required.
Keywords/Search Tags:Data, Schemas, New
Related items