Accession Number : ADA183064

Title :   Keyword Cluster Algorithm for Expert System Rule Bases.

Descriptive Note : Technical rept.,

Corporate Author : AEROSPACE CORP EL SEGUNDO CA COMPUTER SCIENCE LAB

Personal Author(s) : Lindell,Suzanne

PDF Url : ADA183064

Report Date : 22 Jun 1987

Pagination or Media Count : 23

Abstract : An algorithm is described for automatically organizing a fairly unstructured expert system rule base in order to facilitate updating and debugging by programmers. The algorithm operates on a structure called an Assertion Cluster Graph (ACG) which consists of nodes for every assertation in the rule base and of arcs connecting the assertations that are dependent on each other for their values. The algorithm reduces the complexity of ACG by replacing related groups of assertions in the graph by a single summary node. The assertions are clustered into groups according to Keywords contained in their English Text. The algorithm is used to create an interactive program which displays the summarized version of the ACG and can expand the clusters on command. It is anticipated that this expert system display tool will not only be helpful to programmers, but will also enable users to better understand how the system works. Keywords: Assertion Cluster graph; Keyword cluster algorithm; Expert systems; Display tool; Rule based systems; Knowledge based systems; Knowledge engineering.

Descriptors :   *DEBUGGING(COMPUTERS), *ARTIFICIAL INTELLIGENCE, *COMPUTER PROGRAMMING, ALGORITHMS, PROGRAMMERS, TOOLS, INTERACTIONS, CLUSTERING, GRAPHS, INDEX TERMS

Subject Categories : Cybernetics
      Computer Programming and Software

Distribution Statement : APPROVED FOR PUBLIC RELEASE