Invited Paper: A Logician Looks at Expert Systems: Areas for Mathematical Research (at LICS 1986)

Authors: Anil Nerode

Abstract

The study of expert systems and knowledge based systems has naturally divided itself into three areas:

• knowledge acquisition (including updating),
• knowledge representation, and
• automated inference (including measures of belief).

These problems have been dealt with by a variety of logical models for knowledge representation, procedures for logical inference, and measures of belief. Many expert system writers feel that the only sticky points are with knowledge acquisition. This is probably due to the logical simplicity of a lot of current applications.
Are there common mathematical themes worth pursuing, and is it likely that mathematical and logical investigations can distill out coherent theories in these areas? We believe so. We discuss in each area the available (and also the missing) mathematical, logical, statistical, and algorithmic tools needed to rationalize the subject. We emphasize tree search and pattern matching as the touchstones for constructing a common theory.
We conclude with some mention of our current work. This comprises some statistical logic corresponding to the probability logics of Keisler-Hoover, and also design of extensions of predicate logic theorem provers designed for incorporating efficiently a broader range of information than at present (such as images and music). All involve doing the elementary model theory, completeness proofs, and automated proof procedures for previously undeveloped applied first order logics that can be associated with theses subjects. The hope is to implement PROLOG's efficient for such problems.

BibTeX

  @InProceedings{Nerode-ALogicianLooksatExp,
    author = 	 {Anil Nerode},
    title = 	 {A Logician Looks at Expert Systems: Areas for Mathematical Research},
    booktitle =  {Proceedings of the First Annual IEEE Symposium on Logic in Computer Science (LICS 1986)},
    year =	 {1986},
    month =	 {June}, 
    pages =      {120},
    location =   {Cambridge, MA, USA}, 
    note =       {Invited Talk},
    publisher =	 {IEEE Computer Society Press}
  }