Results of SMWCon and possible future of Semantic MediaWiki

SMWCon Fall logo. SMW logo blended with Berlin TV Tower

SMWCon conference have just ended and I’ve successfully recovered after it. It was the biggest conference ever: we’ve got almost 90 participants. It was also a very interesting one: we’ve got business talks and scientific talks, talks about open governement and talks about enterprise wikis and a lot more.

Semantic MediaWiki remains to be one of the rare projects that related to Semantic Web and at the same time doesn’t rely solely upon grants to be alive. In fact, most of the core developers of the platform have nothing to do with linked data and semantic web and I think that’s good.

I was a Program Chair, in other words I was responcible for a content of the conference. Unfortunately I couldn’t reach anyone from DataRangers team: these dudes would be most welcomed guests in the conference because they really can talk how to turn semantic wiki into a focused business solution. Still, Alexander from Gesinn.it was amazing too: you could never guess that his semantic::apps solution is SMW inside.

I think that for Semantic MediaWiki future several things are quite important and I’m going to contentrate on them:

  • proper positioning. What SMW is good for? It’s not clear from the website.
  • outreach. MORE PEOPLE. We need the community as big as Joomla/Drupal/Wordpress
  • funding of the development. It’s an old notice that if you put some money into some project, it may become more successfull

About the positioning I’d say that we have to distance from Semantic Web. It was good to have big grants, back then in 2006, but now the dissapointment in Semantic Web grows – despite the fact that many SW-technologies have become part of everyday Web. In my opinion we have to focus the following audience:

  • open government data people. There are some good use cases of that, nost notably NYCPedia
  • enterprise wikis. Most of them are behind firewalls and it’s sometimes not easy to contact the people from there. But they really have proper requirements and funding too.
  • Documentation projects. Webplatform is the one, recently Parson Communications started to use SMW for technical writing.
  • not Semantic Web research projects from bioinformatics, neuroscience, engineering. For example I’ve just stumbled upon the Texas Instruments wiki. BlueBrain project also wants to use us.
  • consultancies that use SMW for supporting some particular methodology. Examples are WikiVote with standards and roadmaps, Gesinn.it with Business Process Modeling

The funding part is especially interesting. So far I see the following ways to bring money and/or workforce into Semantic MediaWiki:

  • If you’re a professor you can have your bachelors to make projects that affect Semantic MediaWiki. Ideal project here will be to measure the efficiency, latency, speed for different storages.
  • If you’re using SMW for your research and have a grant you can use the grant money to sponsor features you need.
  • If you’re developer you can try applying for Wikimedia Individual Engagement Grant or be a mentor in Google Summer of Code. This is how I and Stephan pushed Semantic Glossary a little bit forward.
  • If you can’t spend your money but have qualifield people that ready to make improvements in SMW, please commit them back to core! The whole Linux Kernel works this way: there is a lot of people from Intel, NVidea and IBM people that write open source code in their working hours.
  • If you’re person or a company and you need small feature (e.g. new result format), hire one of the developers: he will write it for you and make it open source thus providing support
  • If you’re a company and you need BIG feature (e.g. speeding up, support for new store, new parameters for #ask, etc.) you can try to ask how many more companies need that. If it’s a common demand we can create a fund where every company put some money.

During the conference I’ve proposed this last model of groupfunding and I’m very eager to try it in action. I haven’t heard of this model in any of the projects but I think that it can just work in our case. Here is the poster I’ve presented at the conference about that:

smw-poster-3-5.1_crowdfund

What I’m thinking about now is the pilot project that will be supported with this groupfunding. It should be something medium-sized and long awaited, something that will interest many companies. Some candidates for the pilot feature I’ve come up so far

  • measuring performance. That is load testing. Many parties want to know how the amount of properties and subobjects affects latency/responce time. How does it work for RDF store? Is it quicker? How much quicker? How about the memory consumption? With and without cache? 
  • Developer documentation. This is tricky, every time sombody ask Jeroen about the proper way to do anything he answers that currently the code is a big mess and SMW will have new cool developer API soon. But anyway have description of something would already be good.
  • fully-fledged SPARQL support. That is, support make inline SPARQL quesries to work with all kinds of result formats
  • Stuff from our questionnaries, for example:
    • support displaying linked properties like “?a.b”,
    • greater support for ORs and ANDs,
    • more up-to-date display of data,
    • free-text search features in queries
    • brackets/braces for complex queries
    • things for forms and page schemas, for example visual editing of forms, WYSIWYG support in textareas, automatic escaping for form field content
    • Access fucking control. I know, it depends on many other factors. 🙁
  • Support for a new storage that will boost the performance. Maybe MongoDB?
  • Semantification
  • Custom datatypes

Any other ideas?