Showing papers on "Abductive reasoning published in 1985"

Answer Justification in Diagnostic Expert Systems - Part I: Abductive Inference and Its Justification

Abstract: Answer justification refers to the ability of an expert system to explain how or why it arrived at certain conclusions (such as a patient's differential diagnosis or treatment recommendations). In this paper, we describe an "abductive" inference method suitable for use in medical expert systems. We then demonstrate how this method can support a clinically plausible form of answer justification in functioning expert systems. A companion paper (Part II) provides the technical details of how the answer justification method described in this paper is implemented, and compares it to previous answer justification methods developed during the last several years.

Syllogistic Reasoning in Fuzzy Logic and its Applications to Reasoning with Dispositions

Abstract: : A fuzzy syllogism in fuzzy logic is defined in this paper to be an inference schema in which the major premise, the minor premise and the conclusion are propositions containing fuzzy quantifiers. This paper may be viewed as an initiation of a study of syllogistic reasoning in the context of fuzzy logic. Such reasoning has a direct bearing on the rules of combination of evidence in expert systems and, in addition, provides a basis for inference from commonsense knowledge by viewing such knowledge as a collection of dispositions. The results presented in this paper are preliminary in nature. The issue of syllogistic reasoning in fuzzy logic has many ramifications which remain to be explored.

Methods of Automated Reasoning.

Abstract: This chapter introduces into various aspects and methods of the formalization and automation of processes involved in performing inferences. It views automated inferencing as a machine-oriented simulation of human reasoning. In this sense classical deductive methods for first-order logic like resolution and the connection method are introduced as a derived form of natural deduction. The wide range of phenomena known as non-monotonic reasoning is represented by a spectrum of technical approaches ranging from the closed-world assumption for data bases to the various forms of circumscription. Meta-reasoning is treated as a particularly important technique for modeling many significant features of reasoning including self-reference. Various techniques of reasoning about uncertainty are presented that have become particularly important in knowledge-based systems applications. Many other methods and techniques (like reasoning with time involved) could only briefly — if at all — be mentioned.