Combining Uncertain Estimates
This addresses the challenge of aggregating uncertain information in expert systems, but it appears incremental as it focuses on a specific method within an existing framework.
The paper tackles the problem of representing and combining unreliable and conflicting uncertain estimates in expert systems for decision-making, proposing a method that satisfies a set of specified properties.
In a real expert system, one may have unreliable, unconfident, conflicting estimates of the value for a particular parameter. It is important for decision making that the information present in this aggregate somehow find its way into use. We cast the problem of representing and combining uncertain estimates as selection of two kinds of functions, one to determine an estimate, the other its uncertainty. The paper includes a long list of properties that such functions should satisfy, and it presents one method that satisfies them.