Procedural renunciation and the semi-automatic trap

Abstract

This paper addresses two contemporary issues which could threaten the usefulness of conceptual graphs and their widespread acceptance as a knowledge representation. The first concerns the recent debate over the place of actors in the formalism. After briefly summarising arguments on both sides, I take the position that actors should be retained, and marshal four supporting arguments. An example shows that (slightly enhanced) actor nodes can greatly simplify the delivery of external control signals, without excessively complicating the denotation of the graphs they contain. The second issue concerns an epistemological problem which I have called the semi-automatic trap. This is our tendency to continue constructing systems of logic that depend on human involvement beyond necessity to the point at which such involvement is impractical, unscaleable and theoretically problematic. Two important escape routes from the semi-automatic trap are pointed out, involving more emphasis on automatic graph construction from primitive data, and empnasis on automatic interpretation of conceptual graphs. Practical methods for both are suggested as ways forward for the community.