Comment: peterme (Sep 25, 2003)
Where are data models?
Where are ontologies?
You've described ways to classify and apply attributes and metadata to content. You haven't discussed how to *relate* pieces of content though. I've been finding that classification (even faceted classification) is insufficient. Items are often best understood through relationships... This was very much the point that Andy Dillon made at the last IA Summit about having needing to get away from structural/spatial metaphors, and to come to grips with semantics.
Comment: victor (Sep 25, 2003)
Very handy Lou, thank you. It'll help explain a lot to the folks at work.
At one point you repeat "document auto-categorization"...intentional?
Regarding Peterme's comments, I'd say data models are a database administrator activity as they're more machine-oriented than user-oriented.
I think ontologies are just one form of content model, covered here. And relationships are achieved using metadata, also covered here. The diagram could go into more detail, but for this scope I think it's a fair amount of info without getting cluttered.
Comment: Lou (Sep 25, 2003)
Peter, content models are the closest thing to data models; very similar, but dealing with semi-structured text instead of data. That's where the *relating* you're referring to takes place: content models consist of content objects connected with contextual linking rules--relationships--generally powered by metadata.
Got that? (Recognize that the diagram is a summary of about three hours of content, so naturally it's not going to be completely obvious.)
Ontologies? Please define. ;-) I'm still awaiting the golden definition of ontology that actually makes sense.
Comment: victor (Sep 26, 2003)
Maybe it would be more clear to define ontologies in contrast to thesauri. Ontologies can have additional 'semantic' relationships, like Adam _buys_ toys, and Kaybee _sells_ toys. Rather than documenting these relationships separately from the concepts, an ontology integrates the two during information modeling prior to data modeling.
Ontologies also have other features similar to scope notes that are highly precise. This allows the ontology to be machine readable. So instead of having to update your info model AND data model every time a business changes, you update the ontology and feed it into the machine.
Longer explanation, in the context of CMS, here:
Is that helpful?
Comment: Lou (Sep 28, 2003)
Victor, you should receive a knighthood for bringing clarity to this thorny issue. To that end, what you and Peter call ontologies I call content models, which is in the diagram.
(BTW, to your earlier comment: "document auto-categorization" isn't repeated twice; one actually reads "document auto-classification".
Comment: victor (Oct 9, 2003)
Incidentally, I just read a favorable review of this in JASIST...
Metadata Fundamentals for All Librarians. Priscilla Caplan. Chicago, IL: American Library Association, 2003. 192 pp. $42.00. (ISBN: 0-8389-0847-0.)
Comment: Lou (Oct 9, 2003)
Victor, can you tell from the review whether the book gets into enterprise or at least distributed metadata?
Comment: John Porcaro (Oct 13, 2003)
Great to meet you at KM World today. This roadmap will be helpful to my team and me as we plow our way through the cornfields of IA at the Lazy M.
Comment: wang yuexiang (Dec 1, 2004)
recently i am preparing for my paper.
the roadmap really help me a lot. although,
some of them are suit for enterprise in my country,that is in china.i will be concerned on it later>
Comment spam has forced me to close comment functionality for older entries. However, if you have something vital to add concerning this entry (or its associated comments), please email your sage insights to me (lou [at] louisrosenfeld dot com). I'll make sure your comments are added to the conversation. Sorry for the inconvenience.