See also: IRC log
<renato> victor/linda - will u join the teleconf call?
I can scribe if nobody does
<scribe> scribe: victor
<renato> https://www.w3.org/2017/08/07-poe-minutes
decision: approve last meeting minutes
<renato> https://github.com/w3c/poe/projects/1
RESOLUTION: approve last meeting minutes
<renato> https://github.com/w3c/poe/issues/209
subtopic: Consequence/Remedy, issue 209
renato: there has been a first proposal (2
weeks ago) and a second one (1 week ago)
... https://www.w3.org/2016/poe/wiki/Duty_Relations
simon: it is faster to talk than to speak
... failure is more general and there may be ambiguity
... the fact that there is a permission with a non-fulfilled constraint
does not imply that there is a prohibition on that permission
... it is a owa/cwa
... the verification of chain of implies requires verifying not only not
the existence of prohibitions but also the existence of permissions
whose constraints are not satisfied
sabrina sees a difference between remedy and consequence
<renato> victor: spec worded as follows:
Constraints must be checked by processings system at the time of executing the relevant action. If the constraint is satisfied (ie the logical expression is true) then the action becomes valid for the enclosing Rule. If the constraint is not satisfied, then the action is not valid for the enclosing Rule.
<simonstey> +q
<renato> victor: agree on diff between conseq/remedy
<simonstey> "For a Rule to become effective, all of it's Constraints MUST be satisfied and any applicable relationship to a Duty MUST be fulfilled."
simon: the term "valid" is confusing to me.
<renato> https://w3c.github.io/poe/model/
renato: failure as a superproperty of consequence and remedy
victor: if a permission has a constraint a failure, then the rule fire upon failure becomes in effect.
simon: this means an addition verification to be done
victor: yes, an additional verification is needed. we need generality
simon: yes, this is not incompatible
benws: the current wording of (??) is now unclear
<renato> https://w3c.github.io/poe/model/#duty-policy
victor: I wished I could represent Prohibition(failed)-->permission, Prohibition(failed)-->obligation, Prohibition(failed)-->Prohibition2, every possible combination
benws: the difference between remedies and consequences can be better stated in that section
victor: (example with matrix of 3x3)
simon: i agree with a superproperty, but we
leave it to custom implementations.
... we can leave failure as rule-to-rule property, whereas remedy and
consequence as kept more precise.
renato: consequence and remedy remain as they are, failure is introduced as a property from rule to rule.
linda_b: this has to be well explained to non-technical users
RESOLUTION: integrate the agreement reached today
renato: this was the last issue before closing a version
benws: (shows screen with a matrix) The evaluator needs to make checks as those shown in the screen
simon: (explains an example)
... someone fulfills a duty (payment), having permission to download a
dataset during 2017
... in 2018, the duty is fulfilled but the constraint is not
benws: interpretation is not so evident: is the rule in effect?
simon: it is not
benws: ok
... Example 16
simon: the consequence is not triggered in this example
benws: but the evaluator does not understand the constraints in the duty
ping
benws: Example 17
<renato> (we are still speaking?)
benws: we have now two duties
... etc. etc.
... IMPORTANT: is this what we expect from an evaluation?
victor: yes
simon: +1
... in addition, evaluators have to deal with conflicts, etc.
benws: we have (or will have) examples with
conflicts
... given the set of columns (marks a column in the screen)... is the
permission in effect or not?
... If you believe I am doing well, I will continue with this job.
renato: any comment?
(silence)
renato: please Ben, go ahead
... what else would be needed to go towards CR?
ivan: are we clear on how implementations will be done?
renato: every implementor will say "this is conformant because it passes A, B, C..."
<simonstey> earl!
ivan: there is no practical support at W3C to help with the conformance tests.
simon: the benefit of earl is manifested
when there are many implementors
... not practical for a handful of implementors
ivan: we need implementations not coming from academic institutions
<renato> https://w3c.github.io/test-results/annotation-model/all.html
<simonstey> +q
<simonstey> http://w3c.github.io/data-shapes/data-shapes-test-suite/
renato: is that table created manually or automatically?
ivan: somebody must be responsible for compiling that
<simonstey> -q
benws: we have to focus on the ODRL interoperability layer, whereas the actual interpretation of constraints is domain specific.
<simonstey> +q
ivan: understand. some partners produce their own policies that others should be able to undestand. I wonder about INRIA (for example) producing policies ... till what extend do they have to be understood?
benws: the core model is slim
ivan: even if using a profile, the core of the policy should be understood (if it is valid)
benws: the disctinction between validating and evaluating is there: anybody can validate anything, evaluate requires knowing the meaning of constraints and duties
simonstey: we can check validity, yes.
... we want to find inconsistencies, etc.
<simonstey> https://github.com/simonstey/ODRL-SHACL-Shapes/wiki/SHACL-Shapes-for-validating-ODRL-Policies
ivan: you will have SHACL validators for S+
simonste: the wiki shows some examples of shapes
ivan: this should be part of the CR
renato: ben will have a lot of work to have
the tables ready
... comments in github will be appreciated
benws: I will write updates on the wiki
renato: AOB?
... i hope we have next week an almost closed version