Ontology
mapping for blogs
Man this is interesting. Is Dave Winer getting into the idea of
ontology merging? Read this
comment and judge for yourself:
?The hierarchy itself is separate ? you (could) publish an
OPML file of that hierarchy and put it in
a public place.
In answering the question, ?Why do it this way?? Dave gave an
interesting response ? the atomization of a blog into feeds would
allow users to merge the ?my world? of their blog with content from
the many ?their worlds? on the net. Such merged topical hierarchies
could then themselves be exported as an OPML file and a defacto statement of ?this is my
point of view, my information and other information from beyond my
domain that I think is important.?
There?s a number of folk hard at work on this problem: the
XFML folks, Paolo Valdemarin
& Matt Mower on their k-collector, and our very own
SWAD-E thesaurus project which discusses
mapping issues . Heck, we even discussed the issue in our
requirements specification. All at different levels of
sophistication and complexity. I suspect that something as simple
as blog categorization won't run into the
really hard thesaurus mapping problems, but that it won?t be
straightforward as Dave?s comments might lead one to think.
To expand on this theme, let?s think about how people might want
to share categories. Firstly, you might want to simply reuse
someone else?s categorization scheme. That?s fine as a bootstrap,
but what if you already have a scheme? What if the 2 schemes
overlap? What happens to your previously categorized blog entries?
You might, I suggest, want at least the ability to say ?these two
categories are the same?.
Then there?s the aggregation of categories. Without some sort of
mapping, two blogs using different categorization schemes are just
that ? 2 blogs.
A feasible approach is the decentralized ontology creation favoured
by Topic Exchange and
k-collector. Here, people suggest new topics/categories, and the
(ever growing) structure is shared among the community. A fine
idea, but one I fear is not scaleable beyond a very small
community.
Finally, there is the idea of ?semantic lenses? ? using 2 different
categorization schemes to view the same content.
Oh, one other thing. Reading on, I note ?Dave?s idea is that
supporting views other than reverse-chron gives new participants
entry points into the data?. I couldn?t agree more, this is what I
was trying to demonstrate with semantic view, navigation and query
on the semantic blogging prototype.