Advances on Uncertain Reasoning in Intelligent Systems

T IS widely accepted that a core competence of an intelligent system situated in an application environment is the ability to represent and update uncertain knowledge on the environment, to reason about the state of the environment, and to take actions that affect the environment in directions that best reflect the interest of the user. The research on uncertain knowledge representation and inference formalisms, on methods for acquisition and learning of uncertain knowledge from human experts as well data, on reasoning algorithms, and on deliberation and planning has been a strong driving force in the field of artificial intelligence. The Uncertain Reasoning (UR) Track at The International Florida Artificial Intelligence Research Society Conference (FLAIRS) was started in 1996 and has been run successfully for six consecutive years as a high quality international research forum. The papers collected in this special issue are selected from the UR Track at FLAIRS’2001. They reflect a cross-section of current research in the uncertain reasoning area. Each paper was subjected to two rounds of reviews, with two to three reviewers for each round. Papers are revised after each round of reviews in response to critical comments provided by reviewers. Two papers in the special issue deal with fundamental issues in uncertain knowledge representation. R. T. Cox’s justification of subjective probability has long been cited as the motivation to use probabilistic models for uncertain belief. However, in recent years, the soundness of Cox’s justification has been questioned. In his earlier work, Paul Snow defended the soundness of Cox’s proof. In this special issue, Paul Snow answers new objections against the reasonableness of Cox’s result. The second paper, by Ronald Yager, discusses the use of a class of nonadditive measures of uncertainty, called fuzzy measures, as a unifying frame