Welcome to the inaugural issue of the International Journal of Approximate Reasoning (IJAR)! I want to take a moment in this space to describe the journal, its scope, and its goals. A good way to start might be to discuss the title of this journal. According to Webster's, reasoning is "the drawing of inferences or conclusions from known or assumed facts." The term approximate indicates the presence of uncertainty. Uncertainties may arise in various ways--as measurement errors, as "noisy" and/or incomplete data, because a process is random, when the semantics of a natural language used to describe and define a process are ambiguous, and so on. Thus, approximate reasoning might be literally taken as the drawing of inferences or conclusions from known and unknown, or at least partially known, facts. I JAR will contain articles about this process.
How do we accomplish or facilitate approximate reasoning? And more importantly for readers of and contributors to I JAR, how can we attempt to have computer systems imitate this faculty for assimilation, integration, analysis, and exploitation of apparently incomplete and inconsistent information about processes under investigation? Initially, we propose systems of logic as formal models of reasoning that attempt to mimic the thinking process humans seem to use in arriving at solutions to (i.e., conclusions about) problems that are at once enormously complex and ill defined. Then we represent these models in computers and devise algorithms that manipulate the available information much as humans might. Humans are able to draw inferences and make decisions about processes in the absence of complete, consistent, and sufficient data and knowledge by using approximate reasoning; yet this reasoning is not well specified, formalized, or even identifiable by most of us.
Articles of interest to I JAR will include studies and formalisms that may eventually enable computers to assimilate, integrate, analyze, and exploit what appear to be the available relevant facts about a process, incomplete and uncertain as they may be, and subsequently to draw conclusions, or, perhaps less ambitiously, to supply, via a person-machine interface, estimates for the most likely conclusions. From these remarks one might infer that I JAR will provide a forum for researchers interested in modeling the human decision process on computers. It is currently fashionable to call such devices "intelligent systems." However, I would argue that we are a long way from building what most of us might be willing to call a truly intelligent machine; thus, I JAR will be content to discuss pieces of systems that mimic some facet of intelligent behavior. This latter aim evokes a somewhat broader interpretation of approximate reasoning than the definition advanced earlier. I JAR will certainly contain discussions of