EVID - a Prolog Program for Interactive Defeasible Reasoning

Background Practical decisionmaking often depends on information which while not directly available to us, can be inferred from other information that is available. For instance, while driving, we may not directly experience the feel of a slick street, but can infer that the street is slick from the visible evidence that it has just begun to rain. This inference is based on the rule-of-thumb that, in typical situations, if rain has begun to fall on a previously dry street, then that street is now slick. There usually will be addi tional rules-of-thumb linking other kinds of evidence (e.g., an oil spill) to a given conclusion (e.g., a slick street). Such rules may have exceptions; for instance, a special kind of roadbuilding material may prevent slickness from occurring at the onset of rain. Thus, additional information that the street in question is made of this material, will defeat (block) the inference of slickness from the evidence that it has just begun to rain.

Invention Description EVID is a defeasible reasoning program written in the logic programming language Prolog. A defeasible rule is a special type of if-then sentence that permits one to infer a conclusion from a supporting evidential sentence, provided that this inference is not blocked (defeated) by other statements of defeating conditions.

Benefits

EVID is a tool which provides robust decisionmaking capabilities by allowing practical decisionmaking in situations where information may be lacking.

Features

In an application, EVID operates upon a knowledge base (KB) which consists largely, but not exclusively, of special defeasible rules, and rules about defeating conditions pertaining to some domain of practical interest. The combination EVID + KB is best employed as a decision support system: the user of EVID + KB can enter various factual data about some situation. EVID + KB defeasibly infers conclusions based on the factual data. These inferences are automatically updated in complex ways depending on changes in the factual information, which may either be evidential or defeating information. In this way, EVID + KB can perform as an advisor in decisionmaking about the domain of practical interest.

EVID?s main features for practical applications are described in Robert Causey's article, ?EVID: A system for interactive defeasible reasoning,? Decision Support Systems, Vol. 11, 103-131 (1994). The theoretical basis for EVID?s reasoning procedures is presented in ?The Epistemic Basis for Defeasible Reasoning,? Minds and Machines, Vol. 1, 437-458 (1991).

Market Potential/Applications EVID is intended to be used in decision support programs.

Development Stage Lab/bench prototype

UT Researcher Robert L. Causey, Ph.D., Philosophy, The University of Texas at Austin

Type of Offer: Licensing



« More Computer Science Patents

---