Skip to content

SMWCon in Amsterdam, the recap

The Fall 2010 Semantic MediaWiki Conference, or SMWCon, happened almost three weeks ago, but sadly I didn’t get to writing a recap until now. Although it’s turned out well, in that the videos from the conference were just uploaded yesterday, so now I can link directly to the talks, which is nice, since I don’t have to do as much describing.

First, my overall impressions: I’ve been to four Semantic MediaWiki-related gatherings now (Boston, Karlsruhe, Boston again, Amsterdam), and I can definitely say that each event has been more serious and more focused than the last one; which is due to the increasing maturity of the technology, and increasing awareness. This one was two full days of talks about serious corporate usages, and SMW-based software that’s working and in use, with very little time lost between talks. It was sort of a relentless barrage of information, which was nice: even the lunch breaks had semi-structured discussions. At the end, there was discussion about turning SMWCon into a three-day event, to better handle all the interest, and to possibly allow time for some development work.

For all the talks, it could be that the most impactful parts of the conference were the discussions that went on during the breaks, since there were a lot of relevant people and a lot to talk about. I know of a few different initiatives that may be happening as a result of talks that happened there. I’ll just mention one, which is the project to add better RDF/triplestore support to Semantic MediaWiki. The basic issue is this: SMW stores its data, and retrieves it, via the wiki’s standard relational database, which is usually a MySQL database. That works fine for most people, but since almost the beginning of the SMW project some people have been asking about being able to use what’s known as a triplestore instead: a database geared specifically for storing semantic triples of the kind that SMW deals in. Triplestores are superior to regular relational databases for handling triples because they allow for reasoning, inferencing and the standard academic stuff, something I’d known about for a while; but they also do faster querying, something I only found out recently.

For a while, there have been two solutions to at least let you export the data from Semantic MediaWiki into a triplestore, where it could then be queried by standard semantic-web tools: one was a built-in component of SMW, while the other was a (non-open-source) extension. But by some coincidence, in the four months before this last SMWCon, three more extensions, all open-source, were created to allow integration with a triplestore in one way or another, including one, LinkedWiki, that was released just two days before the conference. So there’s clearly been an increase in interest, and SMWCon was an ideal place to discuss next steps, especially since almost all of the developers of the different extensions were there (and the developers of one of the extensions, “SparqlExtension”, gave a talk at the conference about theirs).

The current plan, then, as I understand it, is to use the knowledge gained from the creation of these extensions, and probably some of the code as well, to add to SMW the option of directly using a triplestore to both store and query its own data; a solution that should appeal to the semantic-web geeks and the corporate bean-counters alike. That project now has its own page, “SPARQL and RDF stores for SMW”. I’m looking forward to seeing what comes out of it.

As for the talks – let me briefly list some of the talks I found noteworthy; though I encourage everyone who wasn’t there (and who’s interested in SMW, of course) to check out all the videos.

  • Markus Kr√∂tzsch gave an interesting keynote address on SMW: Past, Present and Future. Strangely, for all the discussions I’ve had with Markus and Denny (the main developers of SMW) I didn’t know the full story of SMW’s origins until this talk. It’s public knowledge that SMW was first proposed at the 2005 Wikimania conference in Frankfurt (i.e. the first Wikimania), but I didn’t know that the proposal, and the idea, came about only because Markus and Denny wanted to attend that Wikimania, which was close by, and wanted to present something interesting. I also didn’t know that the first sponsors of the development of the software were inspired to fund it by that very talk, or that Markus and Denny originally didn’t plan to do any of the coding. (Unfortunately for anyone looking to make a film version of the story, none of the people involved have since sued each other. :) ) It’s strange to think how everything could have turned out differently if it weren’t for that one conference. Actually, I’d say it’s an argument for holding Wikimania more often in the standard tech centers of North America and Europe rather than in more far-flung places; but that’s a subject for another day.
  • I gave a talk too, about random stuff relating to Semantic Forms, including my “Semantic Classes” proposal, which actually seemed to get renamed halfway through the talk to “Semantic Schemas”, after some audience feedback. I’m still very much set on the idea, though the name’s pretty much up in the air.
  • Joel Natividad, from TCG, gave a talk that was really more like a problem statement. (It was also probably also the best-put-together talk, featuring audio and video clips, etc.) The issue is workflow in SMW – being able to coordinate the actions of different components, to complete entire business processes. So, to take an example, a workflow could be: someone creates a document on a wiki, which then causes a manager to get notified by email; the manager then reviews the document, edits it and approves it; at which point some other application has a field modified in their database and does something accordingly. As Joel pointed out, it’s been an open question in the Semantic MediaWiki world for over two years now on how best to solve that problem; and now, his company has to implement a workflow solution for a system that has SMW as one of its components, so it’s no longer theoretical. The consensus of the talk seemed to be that the right answer is to use a 3rd-party workflow application, but which one to use, or how best to use it, remains an open question. I’ll be very curious to hear what solution they end up going with (maybe by the next SMWCon?), because that may well serve as a template for other projects.
  • Rudi van Bavel of KeyGene gave a nice talk on importing millions of rows of data into Semantic MediaWiki, and using the results. It was heartening for a lot of people in the audience to see, because he demonstrated that SMW can work well with that much data, and gave some tips on how to configure the database to lead to good results.

Note that I’m just mentioning the talks that especially struck me; there were other talks that were interesting, but of which I had already seen variations before, or knew the technology, so I’m not an ideal judge. On that note, let me plug the talk given by fellow WikiWorks member Jeroen De Dauw on the Maps and Semantic Maps extensions (he already uploaded it to YouTube). I’m very familiar with the technology, but if you’re not aware of it, it’s a real treat to see it in action.

Categories: Uncategorized.

Comment Feed

No Responses (yet)

Some HTML is OK

or, reply to this post via trackback.