See also: IRC log
<Ashok> scribenick: chimezie
<Ashok> RESOLVED: Accept last week's minutes -- add Ahmed to the Roll
-> http://lists.w3.org/Archives/Member/member-xg-rdb2rdf/2008Apr/att-0003/04-rdb2rdf-minutes.html Last Weeks Minutes
Wolfgang: Michael proposed to have f2f with ISWC in October
speaker: the date might be too late
Ashok: willing to host before or after the conference?
wolfgang: could be arranged
Ashok: I might have difficulty before but should be okay doing it afterwards
Ashok: could you get back that to us on this?
Ahmed: how about STC '07 in May in San Jose
Ashok: we could possibly do both
Ahmed: dates are May 18 - 22nd
DavidOrchard: will not be available, but that shouldn't hold up that as an option
Ahmed: will investigate if we can host and confirm the dates
Ashok: we now have a wiki
<Ashok> Wiki: http://esw.w3.org/topic/Rdb2RdfXG
<satya > URL for the presentation: http://esw.w3.org/topic/Rdb2RdfXG?action=AttachFile&do=view&target=RDB2RDF-Knoesis.pps
Ashok: there was a request to do a state of the art or literature survey
scribe:This is what we are supposed to be doing
[ agreement that we should do this ]
Ashok: we need volunteers
Satya: i'll volunteer
Wolfgang: i'll volunteer as well
DavidOrchard:Will this go on the wiki?
Ashok: yes
<satya> URL for the presentation: http://esw.w3.org/topic/Rdb2RdfXG?action=AttachFile&do=view&target=RDB2RDF-Knoesis.pps
satya: in biomedical domain. large dataset (entrez genes)
scribe: two approaches (one without an ontology and another without)
Ashok: was OWL used as well?
satya: in the first case, OWL was not used
scribe: on Entrez gene screens, implicit relationships are shown
Ashok: where did you specify what you were looking for
satya: each gene corresponds to a record in the DB
satya: we are trying to model the relationships explicitely in RDF
scribe: we started with the XML
representation of the source data
... a very large XML file
... starting from the XML representation made the mapping
effort much easier than via SQL calls against the original
database
satya: the data was from 2 database and we were able to use named entities to make the links
DavidOrchard: were these new relationship or previously known?
satya: in method 2, the instances extracted conform to a created OWL-DL model for the data source
scribe: it spanned information model and domain concepts
Ashok: how were the connections made (via what analysis)?
satya: the diagram demonstrates the result of a query (SPARQL)
scribe: then some post-processing was done to derive the diagram
satya: this was an application-driven approach
Ashok: you were able to get pretty good performance from the queries dispatched on the dataset?
Satya: yes
Satya: in the first method, inference was used
scribe: in the second case inferencing was not used
Jenny: what is the next step?
Satya: we have started work on integrating uniprot data
scribe: requires extending the ECOM schema
Jenny: what is the current usage pattern
Satya: we gave the result back to the biological researchers
scribe: after confirmation we will make this available for their queries, however they are not familiar with SPARQL
Orri Erling: did the dataset change much and was subsequent work needed to change the XSLT?
Satya: creating the XSLT was a recursive process (10 days initially)
Satya: not much work was needed beyond the original formulation of the XSLT
Ashok: we need a volunteer to present next week
Orri: I would like to reserve a slot to speak 2 weeks hence
<whalb> My email is: wolfgang.halb@joanneum.at
rssagent, create the minutes