Design and implementation of summary data management system on temporal databases

Abstract

There are many applications of temporal databases, such as CAD/CAM, Office Automation, and Software Engineering, that require not only current data but also past data. For such kind of applications, a database management system (DBMS) must manage both current data and the evolution history of the data. Databases managed by such DBMS are called temporal databases. In order to support temporal databases, we must solve many problems: formal semantics of time suitable temporal data model (including operations on the model), temporal query language, user friendly interface, new storage structures and access methods for satisfying huge storage requirement, query processing and optimization, new mechanisms for concurrency control and recovery, and so on. This thesis proposes a data model for temporal databases, describes the formal semantics, and also defines operations on the model. To guarantee the independence between conceptual structure and internal structure, the model is built by using model theoretic approach. The model can be viewed as a cubic structure, which will be one of the inherent characteristics of temporal databases. Temporal databases will frequently used to analyze past and current data and to forecast the future. Thus temporal databases are good sources of data for statistical analysis procedures. In this thesis, mechanisms to support statistical applications is studied. Aggregate functions are extended to calculate summary statistics of temporal databases. Summary tables are suggested for storing and providing the summary statistics efficiently. This thesis also implements a graphic interface displaying the data of summary statistics obtained from summary tables (or aggregate functions) in graphical forms such as histograms, graphs, and diagrams. These forms are more suitable for statistical analysis thus can help users to make a decision.