Fusing Partially Inconsistent Expert and Learnt Knowledge in Uncertain Hierarchies
This paper presents an approach to reasoning with learnt and expert information where inconsistencies are present. Information is represented as an uncertain taxonomical hierarchy where each class is a concept specification either defined by an expert or learnt from data. We present this as a good framework within which to perform information fusion. We show through a simple example how learnt information and uncertain expert knowledge can be represented and how conclusions can be reasoned from the fused hierarchy. This reasoning mechanism relies on a default assumption to rank conclusions based on the position of contributing information in the class hierarchy.
Unable to display preview. Download preview PDF.
- 1.Cao, T.H., Rossiter, J.M., Martin, T.P., Baldwin. J.F.: Inheritance and Recognition in Uncertain and Fuzzy Object-Oriented Models. Proc. of the 1st Int. Joint Conf. of the Int. Fuzzy Systems Ass. and the North American Fuzzy Information Processing Soc., IFSA/NAFIPS, (2001)Google Scholar
- 3.Rossiter, J.M., Cao, T.H., Martin, T.P., Baldwin. J.F.: Object-oriented modelling with words. Proc. of the Tenth IEEE Int. Conf. on Fuzzy System, FUZZ-IEEE 2001, (2001)Google Scholar
- 4.Cao, T.H., Baldwin. J.F., Martin, T.P., Rossiter, J.M.: Towards Soft Computing Object-Oriented Logic Programming. Proc. of the Ninth IEEE Int. Conf. on Fuzzy System, FUZZ-IEEE 2000, (2000), 768–773Google Scholar
- 5.R.C. Jeffrey “The Logic of Decision” McGraw-Hill, 1965Google Scholar