See also: IRC log
<kei> agenda HCLS KB update [Matthias, Adrian]
<kei> scribenick Rob
Rob is on the phone as well
I may be ??P15
<kei> http://esw.w3.org/topic/HCLSIG_BioRDF_Subgroup/HCLS_KB_Thoughts
kei: URL contains set of
potential set of hcls kb thoughts to discuss, distilled from
email thread over past week
... Agenda:first a short intro, then hcls kb review, then query
federation discussion, then summary of last ontolog semantic
media wiki telcon, finally Q&A/discussion
mattias: looking into optimizing
performance of triple store; having trouble with certain
queries
... problems with denial of service attacks
<mscottm> ..especially queries that don't include the named graph in the query
kei: would like URL of kb
site
... not exactly clear on kb content - would like to see list of
datasets that have been loaded
... would be nice to see list in order to write SPARQL
queries
matthais: has neurocommons content, no additional contents
<matthiassamwald> DERI HCLS KB SPARQL endpoint: http://hcls.deri.org:8080/sparql
<matthiassamwald> test
<AdrianP> http://neuroweb.med.yale.edu/senselab/
<matthiassamwald> DERI HCLS KB web page: http://hcls.deri.org (down momentarily)
adrian: AllegroGraph installed, has loaded some data, need to open a port so that it is accessible via SPARQL Protocol externally, not yet accessible
<AdrianP> http://agraph.franz.com/allegrograph/
<matthiassamwald> Description of the datasets in the Neurocommons KB (which is at the moment all we have on the DERI endpoint): http://neurocommons.org/page/Bundles
kei: interesting to compare the
different triple store technologies as part of HCLS KB
effort
... this comparison will be helpful to the community
... what are the unique features of each? (Virtuoso &
AllegroGraph)
... support for graph operations - e.g. finding protein in a
graph with high connectivity
<AdrianP> AllegroGraph supports Named Graphs for Weights
eric: Oracle may help for that
<AdrianP> and social network analysis, e.g. http://agraph.franz.com/support/documentation/current/reference-guide.html#header3-68
<ericP> Oracle's engine was designed initially for graphs (e.g. maps queries). likely the best topology analysis suppport
scott: federation may be an issue when you add multiple types of triple stores
DARQ or SimWEB
<AdrianP> AllegroGraph has a Sesame SPARQL interface
support federation over any set of SPARQL accessible triple stores
eric: virtuoso already has
federation to some degree - expensive, makes local caches
... two problems: user directs federations (indicates location
of data, not as difficult), user does not direct federation
(does not provide location of graphs, more difficult)
<mscottm> 1+
rob: DARQ and SimWEB are in the later category (support arbitrary queries and automatically decompose)
<AdrianP> federation support by AllegroGraph
+1 on supporting user directed federation to start
<AdrianP> http://agraph.franz.com/support/documentation/current/federation-tutorial.html
<ericP> an experiment in making the user separate the federated sources
adrian/kei: will use a rule layer on top of the triple stores to handle the federation of the queries
the rule layer is executed within a middleware framework
kei: will any triple store technologies allow people to publish RDF data as linked data?
matthais: Univ. of Berlin has a service that will publish RDF as linked data
<matthiassamwald> Pubby: http://www4.wiwiss.fu-berlin.de/pubby/ (exposes resources in a SPARQL endpoint as "linked data")
scott: try federated query using just Virtuoso before executing over multiple distributed SPARQL endpoints
eric: worked with Virtuoso for Banff demo on queries over SQL repos
<ericP> yeah, one of the sources was something that i'd stuck in a nearby MySQL db
kei: how soon will the AllegroGraph repo be externally accessible?
adrian: they need to open externally accessible port; will be delayed by holidays - so, likely beginning of Jan
kei: need to identify which data
sets we want to store in which repos and how the federated
queries will be executed over those data sets
... start with the Banff demo data sets?
eric: start with relational data and expose as RDF
jun: what is the goal of the HCLS KB? Public service?
kei: good questions; what can be
done depends on interest and resources of group; demo is the
short term goal
... can look to incorporate datasets from drug group
... providing a stable public service is different level of
effort; can consider that after we get past the demo stage;
will depend on level of resources
... don't know that we have enough resources to do both
<matthiassamwald> PDSP Ki
scott: issue when creating
queries on PDSP: trying to find substances with high
affinities; encountered naming differences on key property in
the query
... PDSP was one of the databases that Matthais integrated last
year; this issue was never fully resolved
... would be a good idea to go back and make certain that this
naming issue is understood/resolve and that a similar issues
are not encountered integrating other dbs
kei: believes the issues has been
resolved
... does not know the technical details but they have done work
to query across multiple datasets including PDSP
... how much do we want to try and standardize URIs/naming
across datasets when we look at the query federation issue
scott: sounds like good news
kei: going to summarize the ontolog telcon on the SMW that was held last Thurs
http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2008_12_11
<ericP> [slide 2]
kei: first talk was on SMW by
lead engineer
... was a high-level overview of SMW
... second talk covered a variety of SMW extensions developed
by Yaron Koren: Semantic Forms, Semantic Drilldown, Semantic
Results Forms, Semantic Google Maps, compound queries/data
transfer
<ericP> my impression is that most folks who use SMW use Halo
kei: third talk was on the Halo
extension (funded by Vulcan, implemented by Ontoprise)
... Ontoprise delivers the Halo extension as SMW+ which
includes the Halo extensions and some others (e.g. semantic
forms)
... forth talk was on a rules prototype in SMW
... fifth was on a SMW extension to support purple
numbers
... last main talk was on concept modeling in SMW via
templates
... one lightning talk on "Ask the Wiki"
There will be another talk on 1/22/2009: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2009_01_22
<ericP> note, as far as i know, there are no stand-alone paths for ont development in SMW; all require stepping through protogé for the actual modeling
Semantic Wiki Applications & Use Cases (1): vertical applications
<ericP> the wiki is mostly used for posting change requests
<matthiassamwald> (I cannot participate via audio anymore, noise office environment...)
<Jun> (sorry, have to go)
kei: how might semantic wikis intersect with hcls kb scenario?
rob: could have a use case where a semantic wiki is use to extend/customize the content of the hcls kb
kei: can explore that scenario
more offline
... interesting to look at using wikis for query
construction/execution
<AdrianP> bye
This is scribe.perl Revision: 1.133 of Date: 2008/01/18 18:48:51 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) No ScribeNick specified. Guessing ScribeNick: rfrost Inferring Scribes: rfrost WARNING: No "Present: ... " found! Possibly Present: AdrianP Jun Kei_Cheung P11 P15 P19 Pubby aabb adrian eric ericP inserted kei matthais matthiassamwald mattias mscottm rfrost rob scott You can indicate people for the Present list like this: <dbooth> Present: dbooth jonathan mary <dbooth> Present+ amy WARNING: No meeting title found! You should specify the meeting title like this: <dbooth> Meeting: Weekly Baking Club Meeting WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth Got date from IRC log name: 15 Dec 2008 Guessing minutes URL: http://www.w3.org/2008/12/15-BioRDF-minutes.html People with action items:[End of scribe.perl diagnostic output]