Title :
Nonmonotonic inheritance systems
Author :
Yager, Ronald R.
Author_Institution :
Mach. Intelligence Inst., Iona Coll., New Rochelle, NY, USA
Abstract :
The use of frames and other hierarchical representational schemes has become fairly pervasive in intelligent systems. Central to these representations is the inheritance of properties from one object to another by way of is-a-kind and is-an-element-of links. Complexities and difficulties arise when the values inherited are typical (defeasible) and are transmitted through multiple links; in such cases, many of the issues inherent in nonmonotonic logics arise. Some procedures for making inferences in inheritance network that have both default (defeasible) and absolute knowledge (nondefeasible) are suggested. The basis for these suggestions is the use of possibility qualification for representing typical or default values
Keywords :
artificial intelligence; knowledge representation; artificial intelligence; default values; frames; inferences; inheritance network; knowledge representation; nonmonotonic inheritance systems; nonmonotonic logics; Fuzzy reasoning; Fuzzy sets; Intelligent systems; Knowledge representation; Logic; Qualifications;
Journal_Title :
Systems, Man and Cybernetics, IEEE Transactions on