ABSTRACT

In the various methods of knowledge representation, discussed in the last few chapters, it has been presumed that the database consists of precise data elements and the knowledge base contains no uncertainty among its constituents. Such an assumption restricts the scope of application of the proposed techniques. In fact, for many real world problems, imprecision of data and uncertainty of knowledge are, by nature, part of the problem itself and continuing reasoning in their presence without proper modeling tools may lead to generating inaccurate inferences. This chapter discusses the tools and techniques required for handling the different forms of inexactness of data and knowledge. We here cover the stochastic techniques including Pearl’s evidential reasoning and the Dempster-Shafer theory, the certainty factor based schemes and the fuzzy relational algebra for modeling imprecision and uncertainty of data and knowledge respectively. Examples have been given to illustrate the principles of reasoning by these techniques.