She didn't discuss this at length, so the following is my understanding, all errors and bad ideas are mine :)
The process she described, as I understand it, starts with existing taxonomies, collecting them and then start to create entities or nodes for the ontology from them. She stressed that while taxonomies are often created by one person, ontologies need to be a group effort, so creating entities from taxonomies requires discussion and teamwork.
With a foundation of entites the next step is to create relations between the entities and/or properties for the entities, and then possible restrictions or constraints.
I rhink this is a nice, pragmatic starting point, and helps to get through the stage "what is this semantic anyway" and "why and what do we have to model".
A discussion later this evening focused me on the question of ontologies vs technology: many projects start with a focus on technology instead of actual meaning or semantics.
And that's what I think this process, starting from existing taxonomies or glossaries helps to overcome.
I often use examples from the book "Business Model Blueprints" to enphasize content over technology, or the example of contracts I stole from "Semantics in Business Systemss" by David McComb: most business people know that once you start do fight over the content of a contract, it's "all about the words", or more precie, all about the meaning of the words in your contract, or said another way: all about semantics.
Examples like this can help to explain and emphasize semantics to the business people, and a process like starting with existing material can help to show a pragmatic way to start a semantic project.
So I definitely will add this to my toolbox.