Inheritance networks are structures employed to represent common- sence knowledge. Previous work has developed framework for specifying declarative semantics of inheritance networks. This enables us to provide a model-theoretic specification of the meaning that is conceptually simple and computationally tractable. The current reasearch goal is to investigate implementation issues concerning our inheritance theories and their extensions. This work will include the design of efficient inheritance algorithms and analyses of their computational complexity. The issue of interfacing inheritance networks with an Expert System will also be considered. In particualr, this reserch will address the question of the kinds of queries that are posed to a network, and will try to develop special purpose algorithms for implementing them. Furthermore, there seem to be equally well-motivated, yet significantly different interpretations of certain network topologies. This is, in part, due to the fact that inheritance networks are not sufficiently expressive. We will study this aspect systematically, using the analysis to design a more expressive language that will make explicit the available choices. A good understanding inheritance reasoning will provide valuable insights to tackle challenging problems in the reals of nonmonotonic reasoning.
