Catalogue of Artificial Intelligence Techniques
Aliases: Non-numeric Spreading Activation, Von Neumann Machine Connectionism
Keywords: local connectionism
Categories: Natural Language
Author(s): James A. Hendler
Based on early work by Quillian (1966), symbolic marker-passing is used an efficient search which finds paths through a Semantic Net. This search is usually characterised by being Breadth-first, massively parallelisable, and non-deductive. The canonical example of the need for marker-passing is found in the story `John wanted to commit suicide. He picked up a rope.' To find that the context of the story is `hang' requires the spread of information through the network, and the finding of the path: SUICIDE, KILL, HANG, NOOSE, ROPE. This formulation of marker-passing has been applied in natural language processing and planning research. Recent research has been examining integrating this technique with that of local connectionism. See also Marker-passing.
- Charniak, E., Passing markers: a theory of contextual influence in language comprehension
Cognitive Science 7 (1983), 171--190.