Large, complex system visualization using a semantic data model

The thesis to be defended in this dissertation is that a semantic data model can serve as the basis for the visualization of large, complex systems. The schema for the defense of this thesis is provided by the two major subdivisions of the dissertation.; In the first part of the dissertation a semantic data model is developed. A visualization architecture incorporating the semantic data model is also developed. The high-level structure of the architecture is given along with the necessary algorithms and data structures. Detailed examples complete the exposition of the ideas of this part of the dissertation.; In the second part of the dissertation a methodology for the integration of visualization into the software lifecycle is developed based on the results of the first part. The results obtained in this part are complementary to those of the first part in defending the thesis. In order for visualization to be realizable in large, complex systems a unifying methodology, based on automated tools where possible, must be available to the teams of programmers of the organizations who produce the systems.; The theoretical bases for automated tools supporting the methodology are given in three chapters on composition, decomposition, and analysis. Efficient algorithms for accomplishing these tasks are given. The closing chapter of this second part integrates the previous results and shows how they may be positioned in the software production lifecycle to introduce visualization in as painless a manner as possible.; The major result of this research is the introduction of a semantic data model for the visualization of large, complex systems. This semantic data model also connects semantic data model research to research on data flow diagrams (DFDs). This allows DFD-based design methods to be applied to semantic data models for the first time.; Another major result is the introduction of a polynomial complexity algorithm for the decomposition of DFDs. The previously published algorithm is exponential in complexity and therefore did not guarantee an optimal solution.; Secondary results include an algorithm for composition, introduced here as the inverse of decomposition, and the use of timed Petri-net methods for the analysis of DFDs and semantic data models.