Hi all,
as promised, I try to summarize the feedback that Johanna got at her
SWPM talk, as well as the discussion in our telcon from November 17th.
Sorry for the long delay!
First of all the important links:
- SWPM Workshop: http://wiki.knoesis.org/index.php/SWPM-2010#Program
- Presentation:
http://www.slideshare.net/kaiec/towards-interoperable-metadata-provenance-5712788
- Minutes of the Telco:
http://wiki.bib.uni-mannheim.de/dc-provenance/doku.php?id=protocol_2010_11_17
In this summary, I leave out all names of people who gave the feedback
to avoid mistakes due to Chinese Whispers. If you have further comments
or want to correct something, please answer this mail.
*RDF Developments:*
1. Named graphs are on top of list for rdf extensions, most widely
requested feature.
2. Named graph support will be part of the next RDF specs. Reification
will be deprecated. A new syntax that is not XML is perhaps going to be
recommended (Named Graphs are not compatible with RDF-XML).
3. A first working draft should be available in May 2011; Final is
planned for August 2012.
*Metametadata:*
1. Approach could be extended to arbitrary levels of metadata.
2. How do semantics change when making assertions about triples in the
context of named graphs?
3. We should implement a use case sooner rather than later, so we can
find out if this scales, how it works, etc.
4. The level of description set has to be specified (as or beyond named
graphs).
5. There is a high demand for standartiziation of vocabularies for
talking about named graphs
6. Is Dublin Core suitable as a vocabulary for Metadata Provenance or
are there deficiencies?
7. Complexity: a lot more complexity is introduced by this type of
layering, can this be handled by applications? Answer: There's no way
around additional complexity if you want to associate provenance
information with rdf data, we are just shifting it away from the data
level, our approach makes handling provenance information optional,
applications can just deal with the data and process metametadata if
necessary, e.g., to filter out statements from a particular source.
*Provenance and Reasoning, OWL:*
1. Some people were interested in reasoning. Some were intersting in
making provenance a first-class citizen in OWL. Requirement meta*data is
important, any input from our side would be highly welcome,
possibly send email to w3c discussion groups that we need named graphs
2. We need reasoning over provenance information, ideally description
logic reasoning, inference supported?
*DC-Provenance Group:*
1. We were encouraged to take part in other standardization efforts
(e.g., W3C Provenance WG).
2. There is no standard solution for all different application scenarios.
3. How does our approach relate to the concept web alliance, nano, drug
discovery, paul groth is involved, they use named graphs in a similar
way? We will need to look.
4. An encouraging comment: During this workshop we have seen proposals
for languages and architectures, named graphs are probably the only way
to do this, you're on the right track, very nice presentation.
*Provenance Requirements:*
1. Requirements for provenance standardization
- vocabulary, best practices, standards for query and access
- provenance protocol, point people to associated information
2. context in which certain statements about statements were made
-> information extraction: sets of statements
- federated sparql: repositories should describe themselves
3. minimal model which is extensible, or modular model that can be
adapted to different scenarios. temptation to model every edge case
instead of just main case
4. bottom-up, start with local solutions before scaling to the web,
until we get provenance we cannot push for more, chicken and egg
problem, people reluctant to provide provenance information
5. humans and machines as different types of agents, scientific
discourse and argumentation, ontology design decisions
- automatic updating of provenance as data changes
6. where are the consumers of the provenance information?
7. we need a layered approach, evidence and contextual weighting.
I know, quite brief sometimes, as this is directly taken from Johanna's
notes, but maybe some not so clear points are nevertheless inspiring for
further thoughts.
Cheers,
Kai
--
=============================================
Kai Eckert
KR & KM Research Group
Universität Mannheim
B6, 23-29; Building B; Room B 1.15
D-68159 Mannheim
Tel.: +49 621 181 2332
Fax: +49 621 181 2682
WWW: http://ki.informatik.uni-mannheim.de
---------------------------------------------
########################################################################
To unsubscribe from the DC-PROVENANCE list, click the following link:
http://jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=DC-PROVENANCE&A=1
|