Knowledge Patterns

Abstract

At its heart, ontological engineering is a modeling endeavor. Creating a formal ontology requires not just creating a conceptual vocabulary but also specifying how terms in that vocabulary should be combined, and what conclusions should be drawn from a given set of assertions. As well as building specific ontologies, there is renewed interest in capturing useful modeling decisions at a more abstract level of reusable “patterns”, namely representations which capture recurring structure within and across ontologies.While a variety of different styles of patterns have been proposed, we have been working with a particular style which we here call “knowledge patterns”. Knowledge patterns are based on the observation that one often finds oneself repeating structurally similar patterns of axioms (e.g., about containers) when building a knowledge base. To make this shared structure explicit, and hence reusable, these patterns are explicitly represented as general theory schemata, and then used by transforming them (through symbol renaming) to create specific theories. In this way, the general theory can be reused in multiple ways, and the underlying modeling decisions about how the world seems to behave in accordance with those theories is made explicit.