Extending the instance-based data model : semantics, performance and security considerations

• Main Author: Su, Jianmin, 1963-
• Other Authors: Memorial University of Newfoundland. Dept. of Computer Science
• Format: Thesis
• Language: English
• Published: 2010
• Subjects: Database design; Database security; Data mining; Data structures (Computer science); Database management; Newfoundland studies; University of Newfoundland
• Online Access: http://collections.mun.ca/cdm/ref/collection/theses5/id/12788

Thesis (Ph.D.)--Memorial University of Newfoundland, 2011. Computer Science Bibliography: leaves 157-163. Current databases are typically designed for particular predetermined purposes. However, users may need to use the same dataset for multiple and changing purposes, some of which may not be known when the database is designed. To handle multiple purposes in traditional data models, it is often necessary to construct multiple databases or views. When new information needs arise, additional databases or views may need to be constructed. -- The instance-based data model (IBDM) supports instances independent of any classes to which those instances might be assigned. The model adopts a two-layer approach to data organization (instance layer and class layer), so that an instance may belong to more than one class or, alternatively, none of classes defined in a database schema. The model makes it possible to construct multiple and flexible schemas for a dataset to support multiple and changing needs of users. However, previous research on the instance-based model does not address a number of issues related to the strengths of separating instance and class layers in the IBDM in fulfilling the needs of particular applications, including supporting database administration issues such as providing more flexible security policies. -- In this thesis, we propose theoretical and practical enhancements to the instance-based model. First, we extend the semantics and implementation methods of data expressed in the instance-based model. The semantic extension of components of the instance-based model clarifies the definition of the model and the implementation of the components simplifies applications to real database systems. Second, we provide a theoretical comparison and an empirical simulation to show that the instance-based model is more efficient than the relational model on some typical queries. Third, we propose a security model to address security issues in multilevel security applications using the instance-based approach. To ensure the model's security, we also provide operating methods and rules for the proposed model. Finally, we evaluate the proposed model and prove that the model is secure. By applying the instance-based model to the multilevel security area, the research forms the foundation for using the instance-based model to construct multiple schemas and to support multiple applications.