A formal systems approach to solver design-hill climbing method with push down stack

Abstract

This paper presents a formal approach to design of a solver of an intelligent management information system and its implementation. The approach implies set theoretic modeling based on the general systems concepts and implementation in the extProlog.

There are research efforts which attack (optimization) problems using the set theory and logics. Furthermore, they use logic programming languages for their implementation. Although their methods look quite similar to the approach of this paper, there are clear differences between them. This paper is interested in exploration of the solving system rather than algorithms.

The paper first presents a design and implementation procedure of a solver. Then, classification of problems is discussed. The least structured class of the classification is the target of this paper. A data mining system is an example of the class.

Formal theories are derived for the design procedure assuming the least structured case. A solving strategy, which is called a hill climbing method with a push down stack, is proposed on the theories.

A data mining system is used as an example to illustrate the results.

Finally, a full implementation in extProlog is presented for the data mining system.

Keywords

Solver / hill climbing / push down automaton / systems approach / set theory / Prolog

References

[1] Cantone, D., E. Onodeo, A. Policriti, Set Theory for Computing, Springer, 2001.

[2] Hooker, J. Logic-Based Method for Optimization, John-Wiley, 2002.

[3] Mesarovic M. D., Y. Takahara, Abstract Systems Theory, Springer, 1989.

[4] Takahara, Y., H. Kubota, “A Framework of Conceptual Design of Data Processing System”, Office Automation, Vol. 10, No. 1, 1988 (in Japanese).

[5] Takahara, Y., N. Shiba, Y. Liu, “General system theoretic approach to data mining”, International Journal of General Systems, 2002.

[6] Takahara, Y., Y. Liu, J. Hu, H. Shimazu, M. Okada, “Intelligent data mining system”, Proceedings of International Conference on E-Business, Beijing, 2002.

[7] Takahara Y., Shimizu A.. A foundation of problem solving EUD-conceptual framework. J. of JASMIN, 1996, 5(3): 3

[8] Takahara, Y., Y. Liu, “An extended prolog for end user development”, Proceedings of World Multiconference on Systems, Cybernetics and Informatics, Orlando, 1999.

[9] Witten, I., E. Frank, Data Mining, Morgan Kaufmann, 1999.