Re: [BioRDF] Meeting Notes Feb 27, 2006

Some short remarks:

* It might be useful to find some 'best practices' for the conversion from non-RDF data to RDF data. I have found the following to be useful: 1) If the data does not exist in XML, convert to XML first. Excel sheets can be directly exported to XML format (you have to write a XML schema first and map it to the rows and columns), tabular data can be wrapped with tags by a simple script (Perl, PHP etc.). Having all of the data available as XML before doing the conversion to RDF gives a better overview and eases integration between data sources tremeduously. The XML files and XML schemas might have other uses, too. 2) If there is some kind of manual mapping to be done between databases, these mappings should also be written in XML. 3) The conversion from XML to RDF/OWL is done using XQuery that outputs RDF/XML. XQuery is much better suited for this task than XSLT (complex queries, multiple datasources, conditionals etc.). Don't waste your time with XSLT. 

* Regarding use cases: I think various forms of describing and inferring 'connectivity' might be good use cases for RDF and OWL if you want something that transcends descriptions on basic molecular levels. See

http://www.lucs.lu.se/Ikaros/Subprojects/BrainDatabase/index.html

for a pilot project. The CoCoMac datasets might be used to provide experimental data:

http://www.cocomac.org/

The topic of connectivity is very in tune with the datamodel of RDF and the Semantic Web. However, this is still far away from the 'bedside'.

kind regards,
Matthias Samwald



http://neuroscientific.net

Received on Tuesday, 28 February 2006 22:09:47 UTC