Prime rule-based methodologies give inadequate control

ABSTRACT

The use of rule-based methodologies in the development of Expert Systems is widespread. In order to provide good explanations in these systems it is desirable that the rules be prime. The difficulty of expressing control in such rules, and thus arriving at a desirable sequencing of events, has led to pragmatic additions to the basic methodology. Recent developments in the theory of decision processes have provided new insight into the form of a desirable sequencing. Prime rules, even when augmented by sophisticated control strategies, cannot generate from backward chaining all these desirable sequencings. Furthermore, if one of these desirable sequencings happens to be generated from prime rules it may be by luck rather than design.

REFERENCES

• 1: Chen, K., and Ras, Z., (1983) "Homogeneous Information Trees." University of North Carolina at Charlotte, Math. Dept., Technical Report.
• 2: Chen, K., and Ras, Z., (1982) "DDH - Approach to Information Systems." Proc. of CIS8 in Princeton N.J.
• 3: Chen, K., and Ras, Z., (1983) "Dynamic Hierarchical Data Bases." Proc. of ICAA in Taipei, Taiwan.
• 4: James R Cockett, File Handling for Detail, Extent and Sub-Tasks in the Implementation of Decision Processes, University of Tennessee, Knoxville, TN, 1983
• 5: J. R Cockett, The Theory of Discrete Decision Processes, University of Tennessee, Knoxville, TN, 1984
• 6: J. R Cockett, Decision Expression Optimization, University of Tennessee, Knoxville, TN, 1985
• 7: Heard, S. R., (1985) "DECIDE: A Decision Expression Intepreter.' University of Tennessee Knoxville, Dept. of Computer Science, Ms. Thesis.
• 8: Bernard M. E. Moret, Decision Trees and Diagrams, ACM Computing Surveys (CSUR), v.14 n.4, p.593-623, Dec. 1982 [doi>10.1145/356893.356898]
• 9: B. E. Moret , M. Thomason and R. C. Gonzalez, The Activity of a Variable and Its Relation to Decision Trees, ACM Transactions on Programming Languages and Systems (TOPLAS), v.2 n.4, p.580-595, Oct. 1980 [doi>10.1145/357114.357120]
• 10: Quin}an, J. R., (1984) "Inductive Inference as a Tool for the Construction of High Performance Programs." In Michalski, R. S., Mitchell, T. M., and Carbonell, J., (Eds.), Machine Learning. Palo Alto, Calif.: Tioga.

INDEX TERMS

Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
      I.2.4 Knowledge Representation Formalisms and Methods
          Subjects: Representations (procedural and rule-based)
      I.2.8 Problem Solving, Control Methods, and Search
          Subjects: Graph and tree search strategies

General Terms:
Design, Theory, Verification