SMWCon Fall 2011 impressions

(I had this post in draft for too long. Time to publish, as is)

The Semantic MediaWiki conference Fall 2011 in Berlin is over, so time to summarize some thoughts and impressions.

It was my first SMWCon at all. A bit late regarding that I did a Google Summer of Code project for SMW in 2010, but my finances were kind of inexistent then. Happy to get the chance now though.

Before turning to some of the individual talks, just a note of two general things I found interesting (I wish I would have time to review each of them, since there was so much interesting stuff...):

  1. There were a remarkable amount of talks on connecting SMW with the rest of the Semantic Web, through RDF, SPARQL etc. Cool, SMW is seemingly becoming a natural choice of platform for SemWeb publishing!
  2. The proportion of bio-people were also a bit remarkable. Apart from SNPedia founder Mike Cariaso, there were a whole bunch of others, including Salvadore from the GeneWiki project, Toni .... (and me) .... I guess it reflects how good SMW handles the need to give structure to very heterogenous datasets, so typical for the Life Sciences.

The talks

Note that you can now find slides and videos for most of the talks, online:

The conference started with one tutorial day, followed by the main days for talks. The tutorials turned out to be so interesting though, so that most people seemed to attend them as well.

Find below my very brief notes/impressions on some talks that I found specially interesting, for my use cases and interests:

Very nice UIs

Daniel Hansh from OntoPrise showed off their SMW+ Community Edition package, which includes SMW, Halo and other extensions. This is quite cool stuff, with really helpful and slick UIs, so let's hope it will remain open source! :) (Slides, Video)

Performance optimizations in SMW

Markus Krötzchs talked on "Saving C02: Top SMW Performance Issues and How to Address Them". The slides are cram full of link to more detailed in formation, so this I'll have to study in more detail. (Slides , Video)

SMW reworkings for better RDF support

As said, there were lots of RDF talks on the conf. One of which is some reworkings of the SMW internals to support the RDF and SPARQL models better. Markus Krötzch gave an overview of the role between SMW and RDF in his talk "Connecting SMW to RDF Databases: Why, What, and How?" (Slides)

Keeping track of changes, even at the fact level!

Jeroen De Dauw presented a new extension: "Semantic Watchlist", to replace a number of (in Jeroens opinion, somehow lacking) extensions. Looks very welldone! (Slides, Video)

Powerful transforming data from RDF on-demand

William Smith, Christian Becker and Andreas Schultz presented some very cool sutff: "Neurowiki: How we integrated large datasets into SMW with R2R and Silk / LDIF". The LDIF framework seems to do a lot of what RDFIO does, but in a much bigger and more capable framework. Really cool stuff! (Slides 1, Slides part 2, Video)

SMW classes as XML

Yaron Koren presented an idea to store "classes" in SMW as XML/JSON in one single location, rather than as now, in three different places (Category, Template and Form). (Slides, Video)

Towards a Semantic Wikipedia!

Denny Vrandečić and Daniel Kinzler presented the WikiData extension, as part of their work to make a "Semantic Wikipedia" a reality ... and based on another nice demo-project they did: Shortipedia. This is gonna be hot stuff! (Slides, Video)

Linked Data - increasingly important topic for SMW

Anja Jentzsch from the LODD presented Linked Data, and the best practices for how to publish RDF data, so that you really "get connected" to the evolving Semantic Web. An increasingly important topic, as shown by the increased interest in connecting SMW with the outside world. (Slides, Video)

RDFIO / Hooking up SMW with external tools via SPARQL

My talk ... (as blogged earler)

More RDF to Wiki Page title mapping strategies

Michael Erdmann also presented how they do Data integration with SMW+ and OntoBroker. They interestingly use a similar strategy as RDFIO in order to nicefy wiki page titles. Interesting! ... maybe there is a way to consolidate all these efforts in a reusable way? (Slides, Video)

More RDF to Wiki Page title mapping strategies

Jeff Pan talked on "Tractable Reasoning". Very interesting! They focus on "making reasoning reasonable", that is, computationally feasible ... and seem to have succeeded as well, their REL reasoner has shown to totally outperform reasoners such as Pellet for more or less any kind of ontology, cool! (I found out it's quite easy to beat pellet though, earlier, with SPARQL/ARC, and especially with PROLOG). (Video)

More RDF to Wiki Page title mapping strategies

Markus Krötzsch and Jeroen De Dauw talked about the next steps for Semantic MediaWiki. Many great things happening: Foundation started, ... (Slides, Video)

Excel-like statistics in SMW

Benedikt Kaempgen demonstrated some supercool stuff, something like a pivot browser for Semantic MediaWiki, in order to get more "Excel-like" statistics in SMW. They use the Spark extension by Jeroen, to query SPARQL endpoints and similar stuff, from javascript. Supercool! (Slides, Video)

SMW as a semantic browser

Benedikt also showed of their Semantic Web browser, that only requires "Equivalend URIs" to be defined for pages, and then let's you browse the data in the wiki in it's original format. Interesting since that matches perfectly with RDFIO, in that RDFIO complements this with also RDF export and querying in original format, using basically the same strategy! (Slides, Video)


Mike Cariaso talked about what's new with SNPedia ... lot's of cool stuff (apart from how cool SNPedia is just in itself!) (Video)

Lightning talks

Not to forget, there were also a whole bunch of very interesting lightning talks ... too many for me to have time to cover here now. One thing you really should not miss though, is the SPARK extension by Jeroen De Dauw, to query SPARQL endpoints via javascript, for visualizations. Wow! (Slides).

Also, for you Bio-people readint this, the SNPedia + GeneWiki mashup, you'll probably find interesting! (Video)

Well, you better watch them all anyway, they are only 5 minutes each, and there's just too much good stuff there, so I can't cover it all: