Lilia has put up an excellent list of points to consider when building an ontology. If you’re unaware, an ontology is a representation of knowledge (learn more).
I’ve been dealing with some of these issues over the past week and so Lilia (and others), I hope this helps. My comments are in italics against the base of Lilia’s list.
- one expert vs. group involved – If you have an expert then start with them, otherwise a group is going to be necessary. However, the expert may be in the form of an existing ontology. Struggling for categories/topics for this site I’m considering using what km.brint.com has for alignment with others. No doubt groups will take longer but they may include knowledge the expert misses. Also, an expert in a field may not be an expert in classifying what they know.
- expressiveness (detailed and complex ontology) vs. maintainability (simple ontology) – Recently Jim McGee posted about maps and that may be useful here. If you keep in mind the ontology as a map you will find it easier to keep unwanted complexity out. It’s when you start bringing the knowledge itself into the ontology that you get a rapid growth in complexity.
- automatic vs. manual – Can you consider automatic generation in kind with the use of an expert. That is, as a starting point. The thought of automatic ontology generation scares me but I can’t pinpoint why.
- group vs. organisational ontology – I’dd add personal ontology here as well. I’m trying to map what I know about knoweldge management and learning. It’s difficult here by myself but still just as valuable.
In PersonalBrain I have no ontology save that of the TypeOf pattern I’ve applied. But I can still find things quickly. And that’s the rub. Nobody else can find them as quickly because what little ontology there is – is mine.
