Software tools for Uncertain Reasoning: An Introduction
The three papers in this section address a common thread: the development of operable software systems which do reasoning under uncertainty. It is clear that such systems will necessarily be embedded within a larger mosaic of system design issues than are typically addressed in research forums on uncertainty in artificial intelligence. Each paper in this section provides its own perspective on issues of development, delivery, and integration. The paper by Bonissone discusses the Dynamic Classification Problem and in rapid fire addresses a host of issues associated with developing and deploying an uncertain reasoning system. The paper by Booker et. al. also examines development of a classification problem solver (BaRT) from a Bayesian perspective. Like Bonissone, the BaRT team recognizes the need for integrating multiple representations in the reasoner. The final paper in this section by Bradshaw et. a/. addresses integration of a knowledge acquisition tool into a decision analysis environment. This paper also begins to address the issue of model-building in uncertain reasoning system based on the preferences, information, and alternatives of a specific user and problem instance. This section highlights system design issues which encompass an uncertainty management method. Topics in knowledge acquisition, representation, integration, compilation, presentation, and user interface have at least as much effect on the overall usefulness of a reasoning system as the formal characteristics of the underlying uncertainty mechanism. As research in uncertainty in artificial intelligence achieves some clarity on the characteristics of available alternative uncertainty formalisms, it is hoped that these papers will initiate a trend in research on implementation and delivery of uncertain reasoners in complex environments.
PDF Link: /papers/89/p235-breese.pdf
AUTHOR = "John Breese
TITLE = "Software tools for Uncertain Reasoning: An Introduction",
BOOKTITLE = "Uncertainty in Artificial Intelligence 5 Annual Conference on Uncertainty in Artificial Intelligence (UAI-89)",
PUBLISHER = "Elsevier Science",
ADDRESS = "Amsterdam, NL",
YEAR = "1989",
PAGES = "235--235"