ConceptualGraph

This pattern extends the WholePartPattern to address meanings (referents) of natural language clauses - the wholes and parts of spatial or temporal situations (states, events, and processes). This is obviously a huge domain space, but it can be formally partitioned by using two relatively tiny sets of primitive concepts drawn from linguistics:


 * 1) All the situations you actually need within your domain, named by nominalized verbs. Typically, this is a few to a few dozen at most. This design pattern begins when you list their names in English only, then turn that listing into formal association types.
 * 2) The pattern's surprise is the small, fixed set of role types needed for them.  These are NOT classes, but types of Object Property that denote complements for each related verb sense. A couple of dozen can handle every verb case in ALL domains within English, and psycholinguists confirm that this same fixed set works universally over ALL natural languages.

In BioPAX domains, the core equivalents to (1) would be the major types of interactions (chemical or otherwise) and processes (pathways, steps, other changes). The main equivalents for (2) would often be "Participants".

Currently, BioPax discussions focus too much on what these ARE (as classes or class constraints). A better plan under CG patterns would focus on the case roles they fill in the context of a change or state model, and the facts or functionality implied for them by playing that role within the scope of that one association type.

This would facilitate interoperation with other ontologies and indirectly contribute implications phrased in their terms. It would also enable formal models of speech acts (with authors, truth values, etc.) for related association instances.

John Sowa gets credit for popularizing the CG modeling patterns, which arose decades earlier from Charles Fillmore. They have recently blossomed into an ISO recommendation known as Common Logic, in which the graphs comprise one of its legal (and semantically equivalent) dialects.