DC08

Was at the Dublin Core conference in Berlin. Well, I didn't get to see much of the conference as such, or of Dublin Core, I was there for the workshop and panel on user created metadata, which Wikimedia Germany organized. I gave a short overview of metadata support in Wikipedia and MediaWiki, or rather, the lack thereof. Other people on the panel were

• Jochen Topf of OpenStreeMap -- this is for maps what wikipedia for encyclopedias. But they don't simply collect maps, they collect map data, and generates mapf from that - or one huge world map. And they don't only have streets and cycle way, they also have buildings, post boxes... you can add whatever you like.
• Andreas Hotho of Bibsonomy -- bibsonomy is a free(er), academic alternative to delicious and citeulike. I have been using it fro a while now, and must say that the user interface could really use some polishing. But the work they are doing in analyzing and semantifying tags is quite interesting. And that's what Andreas told us about.
• Georgi Kobilarov of DBpedia -- this project analyzes infoboxes on Wikipedia, extracts the parameters and converts the resulting data to RDF. This way, they collected about 100 million statements about people, places, tv-shows, etc.
• Karen Coyle of OpenLibrary -- the open library is cataloging the books of the world. They are importing library and publisher catalogs, and also allow people to add and modify records freely, in true wiki spirit. I really want to integrate this wiki Wikipedia, decent support for bibliographies, book summaries, etc would make Wikipedia an excellent tool for scientific research. The also have a scanservice for books that have entered the public domain. nice.

These are a bunch of very interesting projects and very nice people - with them, as well as Jakob Voß and Patrick Danowski, who organized the event, we had some inspiring discussions before and after the session. One very interesting aspect that kept coming up was: how to connect and integrate projects more closely? To be more specific:

• What data can be shared? Who has what, who wants what? What's useful? I particularly want bibliographic records, and bibliographies, in Wikipedia.
• What exchange formats and interfaces can and should be used? The number of integration modules grows exponentially with the number of parties involved.
• How to integrate seamlessly, on the UI level? Using external widgets, like google offers for maps, calendars, etc? What identity could be used across projects? I think Wikimedia should really start to support OpenID.
• Push or pull? Automatic or semi-automatic? That is, would it be OK, if, say, people could edit Wikipedia pages right from inside OpenStreeMap? Or OpenLibrary records from inside Wikipedia? Going to another site to do the editing is bad, but having a local community for managing the editing is also valuable...

My own presentation didn't go too well -- I had way too many slides, and they were aimed at a semantic web audience, while most of the people attending were actually librariens. But it did spark some intersting discussions, and I hope people got some pointers what to look at, if they are interested. I covered traditional document metadata, structural data (links) and structured records (infoboxes) and described various ways to access them: starting with Special:Export and api.php, the RDF extension (why don't we have that active on Wikipedia?), and went on to describe DBpedia, Semantic MediaWiki, WikiWord, and WikiData. I guess it was a bit of a wild ride, especially for a not-so-technical audience. Hopefully some of the answers I gave during the panel helped to clear things up. There were some interesting discussions anyway, but somehow I got the impression it would have been more interesting to most of the audience if I had talked just about WikiWord. Maybe next time.

Free Content