Conclusion
- I've presented a parser implemented entirely in simulated
neurons.
- The fLIF neural model is a reasonable approximation to biological
neurons and the topology is not entirel unrealistic.
- Words, rules and instances are all implemented as Cell
Assemblies.
- The parser is stackless, making use of simulated memory decay.
- It follows Jackendoff's tripartite theory and is consist
with Lewis' parsing model.
- It gets correct semantic results, has realistic timing,
resolves PP attachment ambiguities using semantics,
and is used in a functioning agent.
- This work has a long way to go, but this is a sound
Pyscholinguistic Model of Natural Language Parsing Implemented in
Simulated Neurons!
- Thanks to the CABot team: Roman Belavkin, Peter Passmore, Emma Byrne,
Fawad Jamshed and Kailash Nadh.
