I am building the events catalog site[1, 2, 3] based on Semantic Web technologies/approaches for my master's thesis (alpha version is avaialable at http://beaware.at). Looks like SemanticOverflow is the best place to get feedback/critics and eventually make the project better :)
I have posted a full list of unresolved problems and open questions at (2). But at this message I want to consult on ontology (here is it's actual version) related questions.
BeAware's ontology is built around events types hierarchy:
Here is a list of open questions concerning ontology structure that confuse me:
I would be very grateful for any feedback.
P. S. Any not related to ontology feedback is appreciated at BeAware googlegroup.
To give you a short and simple answer, forget about overly hierarchical structures, you have many ways to model this, and on the semantic web you do not need to bracket things off as having a single 'type', to illustrate and keep it simple, here's an example from a different domain.
The other way to look at this, is why make a distinction, something is either an event or it isn't, and it may have multiple topics:
What will eventually work I can't say, it's domain specific, however I can say that sticking with a rigid hierarchical structure and trying to shoehorn things in to a single type won't work, forget that and worry about describing things - you're working with graphs not trees :)
Best of luck.
answered 18 Jun '10, 01:00
There are no “right” and “wrong” taxonomies. There are only some that will work better for your use case, and some that will work not so well.
You should have rough targets for the shape of your taxonomy from the start—how many top-level concepts should it have approximately, how deep should it be, how many concepts in total? Should each concept contain roughly the same number of instances or does this not matter? Think about your use case to answer these questions.
Useful techniques for creating the taxonomy: Find a representative set of real instances, and choose top-level concepts so that roughly the same number end up under each concept (this will tell you wether “IT” should be a top-level concept or not). Use techniques from information architecture, such as card sorting, to figure out what goes under which concept.
answered 18 Jun '10, 07:55