SemanticWeb/Research

= Research on Semantic Web Stuff =

RDF Stuff

 * SuRF: Object-RDF-mapper
 * RDFlib: core python library

How do we integrate ontologies directly into domain model creation?
 * http://seth-scripting.sourceforge.net/l

Getting Started with RDF
virtualenv rdfenv pip -E rdfenv install -r pip-requirements.txt

pip-requirements.txt:

-e svn+http://rdflib.googlecode.com/svn/trunk@1744#egg=rdflib -e svn+http://surfrdf.googlecode.com/svn/trunk#egg=surf -e svn+http://surfrdf.googlecode.com/svn/trunk/plugins/surf.rdflib#egg=surf.rdflib
 * 1) usage: pip -E {pyenv} install -r pip-requirements.txt
 * 2) rdflib>=2.4,<2.4.99
 * 3) rdflib 2.5 before it got broken
 * 1) Surf
 * 1) surf
 * 2) surf.rdflib
 * 3) http://pypi.python.org/packages/2.5/s/surf.rdflib/surf.rdflib-1.0.0_r219-py2.5.egg
 * 4) rdfalchemy

Read Surf Docs: http://packages.python.org/SuRF/

To test try out: http://packages.python.org/SuRF/examples/rdflib.html

2010-02-03
http://etherpad.com/i4u2P6916w

Questions:

1. Reification in 4store -- in particular how do we add a uuid to every triple? - full reification doesn't seem too expensive, look at making a library for this using RDFLib - top priority here is probably making py4s rdflib compatible? - *** yes ***
 * This is important because for versioning and provenance (i.e. who did this), reification seems absolutely necessary

2. Wikis and rdfstores
 * What people really mean by a wiki is versioning and a decent UI
 * Ontowiki: http://ontowiki.net/Projects/OntoWiki <-- was looking at this today
 * there good people only downside is php + tie to virtuoso ...
 * Semantic mediawiki

3. How do you do domain modelling in RDF?
 * E.g. take that library app we started on ... (Work, Manifestation, User, ...)
 * rdfalchemy? or something better?
 * options seem to be rdfalchemy, surf or roll your own
 * also InfixOWL (only class/property defs) from FuXi and telescope (only query construction)
 * None of those seem nearly as mature as corresponding db orms but deifnitely a starting point

4. Full text search and triple stores ... full-text index though the query will run, results will not be as expected.
 * hard one. usually involves reading all of the triples. perhaps better to have a separate
 * It is possible to filter by regex but going to be costly
 * What is 4store support for count/group_by type stuff like?
 * count works, group does not. count also counts only all rows (not groups of rows) so even
 * mind that aggregates aren't even an official part of sparql spec yet
 * i know, only in 1.1!

5. semantic.ckan.net and ckan.net - easy enough, will do this tonight - CRA is still there now, need new store to put it in (and it needs some more work w.r.t.      uris for government departments, functions, etc) - suggest we throw up semantic.wdmmg.net (or semantic.wheredoesmymoneygo.org) using current code (which is semantic.ckan.net code) in wdmmg repo - suggest we start doing this on eu2.okfn.org (EC2 instance) -- this doesn't yet have anything production on it   - agreed
 * Regular syncing (how should we do this)
 * Would be nice to have ?format=rdf or similar as well as using accepts header
 * Should we remove CRA breakdown (or has this already been done) and if so should we change example sparql query?