The OWL format seems the most common to express ontologies. 
Is it possible to import new concepts in ChatScript from a specialized OWL ontology?  How?
Can one replace the WordNet subset in ChatScript with a custom ontology?

I have been interested in the integration of OWL (Semantic Web Technologies) and chatbot technologies for a number of years. I have attempted to get research funding for a serious attempt at this goal, but with no success. My past emphasis has been on integration with AIML. Dr. Wallace did create an AIML 2.0 Triple Store as a first step in the integration process. But, lack of funding prevented any serious continuation of the effort, which would be considerable. Possibly there is an integration path with ChatScript (with which I’m not familiar). I would be interested in working with any effort that might ensue.

I don’t know what the actual format of your data is.  XML is a tedious format to read. CS can represent triples and it can represent any JSON entity (including arrays which OWL doesnt do). And JsonReadCSV can process a tab delimited column file, so many things are possible.

The great promise of using OWL ontologies is that they are a formal machine-readable descriptive logic representation. That means formal reasoning can be done across the ontology to potentially “discover” new facts that aren’t explicitly expressed in the ontology. This is a very powerful capability that offers the potential for chatbots to make a major leap forward in their capabilities to understand and respond to human interactions.

We’ve got Chatscript hooked up to the Jena/Fuseki triplesbase back end to give ourselves a more powerful semantic triples environment and also to be better able to dynamically create and store triples during conversations and retain them over system reboots and user restarts. Fuseki imports triples defined in Turtle quite nicely and through Curl we can have access to the full SPARQL query language, although we’ve also replicated the ^query commands so they work with the external triplesbase. Haven’t looked yet at doing OWL imports (although on my list) but may be a way to go.