The term Semantic Distance refers to the technique of relating a concept to meaning in reference ontologies. Existing methods have something as true or not: right on the nose or far away. With our approach, we have statements with many simultaneous meanings, some close and some a bit “less close.”
We’ve identified four key problem areas addressing the need for this human-like capability: a suitable logical framework; a palette of supportable engineering principles for implementations; an effective visual vocabulary for user interfaces to convey the new concepts we encounter; and new ontological principles. The Semantic Distance component of the project addresses the last of these; the ontological problems are three.
1) Current approaches to ontology are focused on standardization and “semantic web”- capable implementations; this results in a tight linkage to first order logic and entity-centric abstractions. Behavior or resulting state are second class, associated with these entities. Useful metaphors for this old approach is that they are noun-oriented, Newtonian, Cartesian. This is handy for the document-centric web which is being extended to message oriented infrastructure. But it poorly suits nature and human-like reasoning, often based on the concept of change — the verbs — rather than discrete states of entities. We need ontologies that are as balanced as the way we think, using both nouns and verbs.
2) Ontologies assume absolute (non-intuitionistic) truth, which you can color with probabilities if you wish. Analogical reasoning — which can almost be defined as real reasoning — is impossible in current systems. We need a way to fold analogies and ambiguities. We need ontological support for non-monotonic reasoning. We need a way to interpret in terms of narrative structures.
3) Facts in current approaches are considered true when stripped of their context — an absolute measure. This is not the way the world works. Ontologies should support the notion of interpreted semantics, interpretation being done in the context of applicable situations and attitudes. This will enable notions of introspection and reflexivity; without that, our systems can't really help us with reasoning problems.
The term “Semantic Distance” was devised for a workshop held At the US National Institute for Standards and Technology and cosponsored by us in the Fall of 2003. Since then, topoiesis [More] has made the need more acute. The plan is to incorporate current ontological work where possible The Semantic Distance notion is proposed as a discussion group within the ontolog forum [More ].