DCMI RDF AP Task Group
Meeting date: June 4, 2015
Meeting link: https://meetings.webex.com/collabs/#/meetings/detail?uuid=MDAL0I2WSSY4CQBPV7OKXPJGYL-JV0D&rnd=144925.55834
Attendees: Karen, Corey, TomJ, Valentine, Hugo, Stefanie, Antoine
1. De-briefing from last W3C F2F
Decision: use http://w3c.github.io/data-shapes/shacl/ as base for SHACL
still needs simple syntax...
Karen: This is the most advanced document
... There are some decisions being made
... very technical
....It would be great to have input
....I can alter you for specific questions.
... Right now, no "friendly" language
... Is it ok to share links to our list?
Antoine: use the public or close list at W3C?
Karen: I'll send a link to DC thread to the WG list, and suggest that we take the discussion to the W3C public list.
Corey: not sure I'll have the bandwidth
Karen: I can't answer at the very technical level the discussion is at now
1a: DC Architecture discussion -
Corey: is this out of scope for RDF validation?
-- or is it more appropriate for an application profile? (mixing validation on content and validation on data)
Karen: does your validation indicate what you're going to be able to do with the object of the triple?
Antoine: RDF validation can pass to another validator
... I think it could be alright to have the RDF AP says which validator is called with say, the objects of a specific properties
.... it's worth raising the issue with W3C
Karen: Tom is it what you have in mind?
Tom: I think it's useful. SHACL could be the RDF side of the AP validation
... I think it's worth exploring as well.
Karen: they're aiming at people to create (named) templates
... bundles of commands.
... they may be in SPARQL but also in other languages. Allows to something that's more extensive than the base validation.
Corey: I was thinking of validating metadata, but rather the technical metadata where things have to be related to content.
All: it would be good if Karen can link the conversations.
Tom: I'm also interested in which graphs should be in scope for validation.
Corey: it connects to what Antoine and I were discussin on the list earlier
... and the work in Hydra group.
... where things could be in a side car
... is it the validator's responsibility to get the data before validation.
Antoine: this really should go the W3C discussion
recursion issue: https://lists.w3.org/Archives/Public/public-data-shapes-wg/2015Mar/0461.html
they also talk about closing graphs.
Tom: what to include in validation graphs.
Is this already a user story? http://www.w3.org/2014/data-shapes/wiki/User_Stories
Use of classes conflates two issues:
1. selecting set of triples that validation process applies to - function over graph g
2. specify a shape, says which entities the shape applies to (focus node)
Corey: you can build the graph on the fly using a sparql construct query - could that be the
mechanism that fits?
Tom: there is a formalism that is needed outside of sparql; construct queries assume that
you have a sparql end point that provides access to some dataset;
but Corey wants to pull data from different sources,
Corey: question: how important is it to us that this get folded into shacl -- or
can we assume something like DSP to construct our set, then outsource
actual validation to shacl
Tom: shacl needs a clear model of what it is validating; there may be models
other than the "function over graph g" approach. Then we need to look at it; it may
not meet our use cases. Relates to concept of "linked data fragments".
ACTION: Corey and Tom to write short position paper.
2. Alignment between DCMI and W3C Requirements
DCMI Req Database: http://lelystad.informatik.uni-mannheim.de/rdf-validation/
Requirement discussion pad: https://etherpad.wikimedia.org/p/requirements_analysis
New cleaned list: http://wiki.dublincore.org/index.php/RDF_Application_Profiles/Requirements
W3C Use Cases and Requirements: http://www.w3.org/TR/shacl-ucr/
Comparison of W3C and DCMI requirements:
Current W3C requirements list, reorganized: http://w3c.github.io/data-shapes/data-shapes-ucr/
Hugo's email with suggestions: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=DC-ARCHITECTURE;41aa27ca.1505
ACTION: group to tackle by email:
- requirements taken from W3C which apparently don’t have a match in DC
- suggestions on the mapping from DC to W3C
3. Core DCMI Requirements
ACTION: Karen to send an email re-organization reqs, to see if the notion of 'core' help us to make things clearer
4. Other requirement actions
ACTION: Corey and Karen to write up cases of validation with de-referencing or local caches, to be sent to W3C
Corey: I thought we had done. I thought it had been sent and they were not very interested. I will ask Karen for confirmation.
Hugo's second email with suggestions:
ACTION: Antoine to look at unclear requirements in Hugo's email
--ONGOING: will try a first discussion offline
R-208: should be more precise
Scope could be broader - POSTPONED until later stage (i.e we can name the scope)?
R-115 needs definition.
R-1 is highly unclear
"Depending on the property semantics, there are cases where two different literal values must have a specific ordering with respect to an operator"
"Some constraints or property semantics require to compare literal values, based on order and/or (in)equality"
Suggest to replace
"Limit number of namespaces in a profile"
"Define allowed namespaces in a profile"
(the idea of a number limit is weird and not really found in cases)
Suggest to replace
"Define valid properties"
"Define allowed properties"
(statement with an allowed property may be invalid)
Suggest change: "Properties may be required based on patterns, such as" -> "Descriptions and constraints may employ boolean combinations of properties, such as"
ACTION: Hugo to try to organize reqs in abstract requirements, trying to re-use Thomas' classification in the DB
--ONGOING: Thomas' classification was still ongoing work.
Coming back to Requirement report
5. Update UCR document?
Evelyn and Thomas' version
6. Development of test cases
- Europeana examples:
ACTION: Karen & Valentine: Develop one or more test cases based on Europeana data
--DONE: Two samples have been prepared: one valid EDM, one with errors
- Stefanie's examples:
- Corey's examples (based on Hydra work)
7. AOB, next calls
============= Remaining items, for next calls
X. Publishing and accessing LD profiles
Thread started by Lars on profiles on the LOD list:
https://lists.w3.org/Archives/Public/public-lod/2015May/thread.html ("Profiles in Linked Data")
ACTION: Antoine to ask Lars and send it to the DC-A list.
--DONE: discussion to happen on June 18
X. Coordination with Hydra
Hydra Metadata Working Group: https://wiki.duraspace.org/display/hydra/Hydra+Metadata+Working+Group
Portland Common Data Model: https://wiki.duraspace.org/display/FF/Portland+Common+Data+Model
ACTION: Karen: post this link to W3C group - on hold until W3C group gets to the right point
Sprint on Linked Data Fragment and Fedora. June 8-June 19. Right now we're unsure whether there will be validation-related requirements
Fedora4's using linked data fragments - http://linkeddatafragments.org/ ?
X. Express requirements as RDF, mapping between DSP and our requirements [on hold for W3C first specs]
X. Coordination with BIBFRAME
Thomas' evaluation of BIBFRAME AP
X. Glossary - postponed until we have time!
X. Possible Next Steps: http://etherpad.wikimedia.org/p/requirements_next_steps
Agreed: defined what an RDF AP is, after agreeing on Core