See also: IRC log
<renato> https://www.w3.org/2016/poe/wiki/Meetings:Telecon20170619
RESOLUTION: minutes are approved
<renato> https://www.w3.org/2016/poe/wiki/Policy_Inference
renato: checks if test cases are defined -
not yet
... simonstey has created some examples
... it is important to define how ODRL should be tested by users - and
we should create test cases accordingly
simonstey: we have a lot of discussions on
Github - victor brought up details which need test cases
... before we can start drafting test cases we (this group) has to
define what is allowed and what not
<benws> Another expansion?
renato: we have added to ODRL the feature of
shortcuts and this makes some evaluations harder
... does this group want to support shortcuts?
<simonstey> odrl:permission odrl:use
<simonstey> odrl:perm [ a odrl:Perm..
simonstey: showed examples of unclear modeling
renato: asked simonstey if odrl:permission odrl:use is wrong
simonstey: this model raises the assumption that odrl:use is a subclass of Rule
renato: ok, then we need to stop doing that
simonstey: but the ontology allows that - we need to modify/expand the ontology
renato: ok, then it is required to fix such
issues
... suggested to have a look at
https://www.w3.org/2016/poe/wiki/Policy_Inference
benws: suggested that it should be possible
to have a local evaluation at the site of an implementer
... as testing the evaluation requires human resources
simonstey: this policy inference page should be the starting point for evaluation.
<benws> ... implementation resource
simonstey: should check for conflicts. and
then check if permissions and prohibitions become effective
... details of that could be driven by local interpretations
<benws> +1
ivan: about the planning of the CR phase:
who are implementers?
... want to get a clearer picture of where we are currently?
benws: take the standard and check what
needs to be implemented.
... this will apply to the evaluation of the "black boxes"
simonstey: the validator will be based on
SHACL
... for the evaluation: given a policy the output should be a set of
rules which are effective
ivan: asked simonstey : is the evaluator an
existing tool?
... understands simonstey as "part of the ODRL specification needs
implementation"
<renato> https://www.w3.org/2016/poe/wiki/Policy_Inference
renato: went over the Policy Inference document - finally a lot of expanded rules will be created
simonstey: agreed
... a tricky detail is if a generic use is permitted and only at a lower
level permissions and prohibitions "fight" against each other
renato: asked benws if his TR black boxes will create so many expanded rules
benws: the expanding creates only a temporary object, only the compressed policy will be stored
<simonstey> "If a Policy has the conflict property of prohibit then any conflicting Prohibition Rule MUST override the Permission Rule and continue with the Policy as valid."
renato: asked CarolineB as implementer
CarolineB: she does not have the technical knowledge
CarolineB: we create policies and other
people have to read and understand it
... no evaluation will be done by the Copyright Hub
simonstey: re the Expanded stage of the inference page: it is only required to figure out the rules which are in effect and if conflicts come up
benws: important for him is if the rules
associated with an asset are in a conflict
... even if they are in different policies
renato: we have not discussed that so far
benws: could be added as test case
ivan: raises a basic question: the semantics
of the common vocabulary will be a separate document in future
... the model document is describing an RDF vocabulary without any
semantics going beyond RDF and OWL features
renato: the information model document is more narrative than an ontology
ivan: what document(s) will be part of the
CR - and what are the exit criteria
... if it is only a vocab - is this what the community wants?
... is not happy about testing thing which are not part of the formal
specs
benws: ODRL comes with a processing model -
this is very important to TR
... TR would create a new policy based on policies received from
suppliers - and this will be more than just a vocab
ivan: who will create that, who will test that?
renato: e.g. the hierarchy of actions which
may create conflicts is not defined somewhere explicitly
... the test case must be only based on the Information Model document
ivan: agreed to that approach
... if something which is required in a test case does not exist in the
IM then the IM is not complete
... a CR should prove that a specification is complete
renato: would it help us to check the activities of testers and if something is missing
<renato> Activity Streams test cases
<renato> https://github.com/w3c/activitystreams/tree/master/test
<renato> simon?
simonstey you have disappeard on audio
<simonstey> oh
benws: if all the specs are in OWL then the IM would be sufficient. If the semantics go beyond
<renato> @simon nope
benws: then the IM cannot define everything
ivan: sees the need of adding SHACL to the CR documents
<simonstey> without the evaluator yes
<simonstey> OWL can't tell you whether a rule is in effect or not
simonstey we still can't hear you
renato: we need a better plan for test cases
... invited all to join in Github and the issues discussed there
<simonstey> internet was breaking up again
renato: this test case discussion should be continued at the next call
<ivan> trackbot, end telcon