So, time to get something installed on my Ubuntu system, based on Samuel description here:

1. sudo aptitude install mediawiki-extensions php-apc imagemagick texlive-latex-base gs-gpl cjk-latex
2. configure MediaWiki
   1. zcat /usr/share/doc/mediawiki/README.Debian.gz | more
   2. follow instructions on http://www.mediawiki.org/wiki/Manual:Installing_Mediawiki_on_Ubuntu_10.04
   3. create a MySQL account as described here
   4. do config stuff at http://localhost/mediawiki/config/index.php
3. set up a save place for local checkouts for MediaWiki extensions
   1. sudo mkdir -p /usr/local/lib/mediawiki/extensions
   2. sudo chown egonw:egonw -R /usr/local/lib/mediawiki
4. configure Semantic MediaWiki
   1. cd /usr/local/lib/mediawiki/extensions
   2. svn checkout http://svn.wikimedia.org/svnroot/mediawiki/trunk/extensions/SemanticMediaWiki/
   3. cd /var/lib/mediawiki/extensions
   4. sudo ln -s /usr/local/lib/mediawiki/extensions/SemanticMediaWiki
   5. sudo nano /var/lib/mediawiki/LocalSettings.php
      1. add: include_once("/var/lib/mediawiki/extensions/SemanticMediaWiki/SemanticMediaWiki.php");
      2. add: enableSemantics('localhost');
      3. add: $smwgShowFactbox= SMW_FACTBOX_NONEMPTY;
   6. visit http://localhost/mediawiki/indSpecial:SMWAdminSpecial:SMWAdminex.php/Special:SMWAdmin

Next step is to install Samuel’s extension in its requirements:

1. set up the required extensions SMWWriter and POM:
   1. cd /usr/local/lib/mediawiki/extensions
   2. svn checkout http://svn.wikimedia.org/svnroot/mediawiki/trunk/extensions/SMWWriter/
   3. svn checkout http://svn.wikimedia.org/svnroot/mediawiki/trunk/extensions/PageObjectModel/
   4. cd /var/lib/mediawiki/extensions
   5. sudo ln -s /usr/local/lib/mediawiki/extensions/SMWWriter
   6. sudo ln -s /usr/local/lib/mediawiki/extensions/PageObjectModel/
2. apply the patch (where is it filed upstream?)
   1. cd SMWWriter
   2. wget http://saml.rilspace.com/smwwriter-fixesfor-rdfio-20100716-r2.patch
   3. patch -p0 < smwwriter-fixesfor-rdfio-20100716-r2.patch
3. set up the RDFIO work from Samuel
   1. cd /usr/local/lib/mediawiki/extensions
   2. mkdir RDFIO
   3. cd RDFIO
   4. svn checkout http://smwrdfio.googlecode.com/svn/trunk/ .
   5. sudo nano /var/lib/mediawiki/LocalSettings.php
      1. and add the RDFIO blob given on Samuel’s Install page
4. install ARC2
   1. cd /usr/local/lib/mediawiki/extensions/SemanticMediaWiki/libs
   2. wget http://code.semsol.org/source/arc.tar.gz
   3. tar zxvf arc.tar.gz
5. install the RDFIO pages onto the main page for easy access, add:
   1. [[Special:ARC2Admin|ARC2Admin]]
   2. [[Special:RDFImport|RDFImport]]
   3. [[Special:SPARQLEndpoint|SPARQLEndpoint]]

That’s it. Not quite the 5 minutes that Samuel promised me, but I’m happy to have this available for my conference tour next month! I installed the default RDF, and got the nice default wiki page to which I can now start adding manual annotation:

If you are wondering about the use case, this RDF import is ideal for building up knowledge bases, as detailed in my Critical mass for Open Notebook Science wikis by prepopulation with RDF data post last month. Just aggregate the info on the web you can find (yes, that’s another story), put it in your wiki and complement it with your local knowledge, import into Bioclipse, and run your analyses to verify your hypotheses!

Thanx to Samuel for this really great work!

This also applies to wikis. WikiPedia clearly has enough critical mass, but is struggling with scaling, which is another problem you can encounter. Building a dedicated, domain specific wiki allows you to do things the way you like. For example, the can be used to build up the knowledge base around a specific problem. For example, HIV drugs (see HIVdb as an example knowledge base). The wiki could contain information you need for your data analysis. The developers team will likely be small, but the data will be highly curated. You are basically both the developer and the user community. Hopefully, in good Open Notebook Science manner, things will grow, and others will join in on building the knowledge base.

Semantic MediaWiki bridges the human readable wiki content, with machine readable RDF content. This wiki does require some additional annotation, but at the same time makes important facts machine readable. The Bioclipse Wiki is using it, though we are not yet taking advantage of the RDF exposure.

Now, returning to the critical mass aspect we started with. Samuel, who did a project with me on reasoning with Prolog in Bioclipse, is now doing a Google Summer of Code project (is there any significance in the fact that three Bioclipse developers are student or have been mentor in the GSoC? :) within the Semantic MediaWiki project. In particular, one of the things he was just working on, is the import of RDF into a wiki.

Oh goodies! That is pretty cool, don’t you think? Just pull together RDF from your field of interest, drop it on the wiki, and you have your initial wiki the experts can continue to curate, add facts, etc. He already told me about his plans to add a SPARQL end point for the Semantic MediaWiki software.