Characterization of Functional Relationships

Abstract

We have seen that QP theory in the form proposed by Forbus is too weak to answer many of the questions we would like to ask when reasoning about a physical system for control purposes. We have proposed that this weakness be addressed by extending QP quantity and relationship representations in linguistic form using soft quantitative information. The goal of this chapter is to develop an annotation technique that will permit us to resolve ambiguities in a constructed QP model of a situation. That is, we assume that standard QP models have been applied and ambiguity has been encountered during influence resolution. What we seek here is a mechanism for annotating the models using extratheoretic information capable of resolving the ambiguity. This chapter begins with an examination of the causes of the particular ambiguities uncovered in our example. We then present a method of resolving this class of problems based on linguistic characterizations of partial functional strength and fuzzy relational algorithms. Details of the algorithms involved are presented, and we conclude with examples showing how the undecidable questions raised earlier are resolved by these techniques.