Expressive Extensions to Inheritance Networks

Even though much work has gone into explicating the semantics of inheritance networks, no consensus seems to have emerged among the researchers about their precise semantics. In fact, there are several different possible interpretations of the same network topology. So, from a practical knowledge representation standpoint, in the absence of any independently verifiable semantics, we wish to pursue the approach of enhancing the language of traditional inheritance networks to enable the user to choose among the various available options, to program in a more complete description of the input problem in the enriched language. This approach permits representation of certain networks that were not representable previously, and allows making subtle distinctions among networks that were not hitherto possible. In this paper we propose an annotated inheritance network language, develop its formal semantics by amalgamating harmoniously a family of related “local” inheritance theories, and discuss some implementation issues.