Michael K. Tschudi

e-mail: tschudi@asu.edu
telephone: 602/829-9112


Current Research Interests

The Adood Ranch Project

Proposed Thesis: "Static Analysis of Active Database Rules: Evaluating Tools for Termination Analysis"

Active databases may contain rules triggered by events produced by other rules, perhaps leading to infinite rule triggering behavior. There are existing methods for static termination analysis, but all are too conservative in their methods, and thus may report non-existing rule- triggering cycles. With conservative methods, simply having an event of the same name produced by one rule and consumed by another rule becomes the basis for declaring a triggering relationship between the two rules.

Actual triggering is based on satisfiability conditions associated with rule parameters and other conditions. Research done at Arizona State University produced a set of advanced static analysis methods. Less conservative, these methods represent more accurately the actual state of a set of active rules. The research presented within this thesis implemented the advanced static analysis methods and extended them..

In particular, extensions were made to merge the advanced analysis methods with other methods of analysis, incorporate dynamically-changeable attributes into the static analysis, and demonstrate the feasibility of cycle-wide analysis in addition to pairwise analysis of rules in a triggering relationship..

Analyzing a sample database constructed for this thesis, this implementation discovered 744 potential cycles of rule triggering among ten active rules. Using several satisfiability analysis methods integrated into a single implementation, the software was able to determine that 740 of these cycles are not satisfiable, and thus will not actually cycle. This represents a tremendous savings in effort for an active rule designer..

Results indicate that the advanced static termination analysis methods and the extensions to blend them with other methods and with cycle-wide analysis were highly successful in discovering unsatisfiable cycles, whereas the extension to incorporate dynamic attributes was disappointing.



Publications


Resume

Employment Experience


Publications


Education


Other