I always thought rule-based was synonymous with logic-based AI. Logic has axioms and rules of inference, whereas rule-based ai has a knowledge base (essentially axioms) and if-then rules to create new knowledge (essentially inference rules).

But in their famous article "What is a Knowledge Representation?", Davis, Shrobe and Szolovits seem to imply that they are not:

Logic, rules, frames, etc., each embody a viewpoint on the kinds of things that are important in the world. Logic, for instance, involves a (fairly minimal) commitment to viewing the world in terms of individual entities and relations between them. Rule-based systems view the world in terms of attribute-object-value triples and the rules of plausible inference that connect them, while frames have us thinking in terms of prototypical objects.

Is this only saying that rule-based are propositional whereas logic-based is usually meant to mean predicate logic? Or is there more to it than this?

  • $\begingroup$ "rules of plausible inference" implies to me that these rules are probabilistic instead of logical. $\endgroup$ – BlindKungFuMaster Aug 9 '17 at 7:17

I want to preface this by saying that the distinction is not clear. Nevertheless, I'll tell you what I know about this, and I will attempt to make the further clarification:

The Structure of rule-based agents is: Take input from environment, pass through condition-based rules, and perform the action through actuators or anything which creates some action in the environment.


The structure of logical agents (based on internal state) is: Take the input from environment, from the internal knowledge checks of what the state is and how the model may evolve; next, infer what my action will do; finally, pass through condition-based rules and take action in the environment

SOURCE: Artificial Intelligence: A modern approach by Stuart Russell and Peter Norvig

Since this is my first answer here, if I have broken any rules, please let me know.

| improve this answer | |

Rule-based systems cover a wide range of systems. Some make use of boolean if/then/else rules, others may use weighting or even probabilistic inference. Some operate on frames, some on java objects, some on propositions that can be formulated in predicate logic. An example of a popular rule system is Drools.

Some rule systems can be expressed as a subset of predicate logic. For example, SWRL is a W3C standard rule language that extends OWL Description Logic (DL) with horn rules. Both OWL-DL and SWRL are expressible in first-order predicate logic. However, not all rule languages are directly expressible in this way, as rule languages encompass such a range of semantics. Even in cases like Prolog there are subtleties. Pure prolog is a subset of FOL, but actual existing Prolog implementations are not FOL subsets (e.g. order of precedence matters).

The W3C Rules Interchange Format (RIF) working group has done a lot of work attempting to unify these different perspectives. I'm not sure the best place to start to see a summary, but you can see some useful material on the group charter page.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.