Thaddeus Julius Kowalski

Old Friends Revisited

The programs in this chapter are old friends to programmers who may well have written them in each language that they learned. The rule-based programming paradigm is different from most algorithmic languages in at least two respects: 1. The next step in the calculation is determined by pattern matching the data against the set of rules. Because any rule may fire, the idea of sequentially following the program text (i.e., tracing) is not an effective way of predicting or analyzing program behavior. 2. The programs are (quasi) non-deterministic. At any time several instantiations may contend for the next action. (Strictly speaking the program is deterministic because, when presented with the same data, the same sequence of actions occurs.)

Rule-based programming

Programs. Tables. Figures. Preface. Introduction. Old Friends Revisited. Search and Backtracking. AI Paradigms. Language Processing. Parts of Prolog and LISP. Machine Learning. References. Appendix: Quick Reference on C5-R 2.0. Index.

Search and Backtracking

Search is one of the most general problem solving techniques. The essence of search is ‘seek and ye shall find.’ That is, if a solution to a problem exists then one is sure to find it if one examines all possibilities. A good way of representing and visualizing search problems is as a graph where some vertices in the graph represent solutions, other vertices are partial solution states, and a path in the graph represents a sequence of actions moving through partial solutions. If the search is successful, then the path will lead to a solution vertex.

Parts of Prolog and Lisp

Each language and metaphor has strengths and weaknesses that are unique to it. This is as true of programming languages and systems as it is for so-called natural languages. As the previous chapters have shown the strengths of the rule-based paradigm as embodied in OPS5 help provide good solutions to a range of problems.