W3C

Permissions and Obligations Expression Working Group Teleconference

22 Sep 2016
See also 23 Sep 2016

Agenda

See also: IRC log

Attendees

Present
bob_bailey, phila, paulj, renato, simonstey, sabrina, Nandana, BobBailey, Sabrina, Victor, Kerry, michaelS, kerry, ericP, victor, dsinger
Regrets
Chair
Renato, Ben
Scribe
phila, kerry, Sabrina, nandana, victor

Contents


<simonstey> present

<phila> agenda: https://www.w3.org/2016/poe/wiki/Meetings:TPAC2016

scribe duties

<phila> scribe: phila

<scribe> scribeNick: phila

renato: Starts with a tour de table
... I put in an overview history session to start with
... I'll go through those, which I went through with the DPub IG earlier in the week

<simonstey> slides -> https://www.w3.org/2016/09/w3c-poe-wg-overview.pdf

renato: Talks about history, battles between different standards bodies, odd use cases etc.
... On to the CG that created version 2

phila: How did the ExRML pan out in the end?

renato: We lost at MPEG, won at OMA. Went to IEEE etc. OASIS started a WG around this
... But those groups didn't go anywhere
... Only OMA and MPEG 21 that did something definite
... OGC developed a DRM reference model

paulj: I wasn't sure whether the MPEG process was real. Content Guard covered everything. So hard to dig into it.

renato: MPEG is a RAND organisation, ODRL is royalty free
... MPEG has a patent
... Moving on to scope
... CG specs are the baseline
... Want to look at UCRs today and tomorrow

benws: Complete rule language... I don't want a complete rule language, but I do want to be able to combine statements

renato: Describes OWL -> HTML page system (Mo did that)
... Talks about profiles. We don't expect complete implementations, different communities will create different profiles.
... Lots of new use cases
... Maybe we should set a deadline for new use cases?

simonstey: You mentioned the script that generates HTML from OWL, is there some kind of regulation wrt the correctness of the transformation.
... Is there some organisational/adamin step

renato: No, it's just a hand crafted bit of scripting that Mo did in the CG.
... We changed it recently for the SKOS collections and put that collection in an order

phila: We have a class called rule that isn't a Rule language

renato: This diagram is expected to be in the vocab spec
... Many terms were deprecated between 2 and 2.1
... greater use of Creative Commons
... Not using OMA URIs so there was some cleaning up
... On Expressions...
... Examples in all serialisations
... Need to discuss JSON-LD
... And we need to promote profiles
... IPTC RightsML is a good example
... Only use one policy type, only this subset of actions etc. They can narrow the semantics of terms, but can't broaden its semantics
... You can add new duties and constraints that are important for your community
... RightsML used by AP has an API
... All of the content published by AP you can hit the API and get the licence for that content

<simonstey> +q

Sabrina: Do you have details of how they're using it in blockchains?
... I'd be interested in that

renato: Monegraph is also interested in publishing music

simonstey: It's super interesting. I've created an account and uploaded a picture and created a sample licence
... They're storing it on the BitCoin block chain

Sabrina: Did it cost you anything?

simonstey: No, it's free. I can create as many licences as I wish.

<simonstey> maybe related to http://data.open.ac.uk/licence-picker/?controller=picker&action=index

<simonstey> +q

bob: Asks about implications of mixing lots of assets that each have their own licence.

Sabrina: Simon and I have a national project starting in November, on that topic. What licence would you give a compound asset? We're working on the reasoning engine

simonstey: I think that's where the formal semantics should guide you in telling you how to combine. If you're combining, how should you dead with conflicts that arise.
... As soon as there are some conflicting terms, you need some guidance.

renato: So that's the main background and history
... Today and tomorrow, we have the agenda. The primary idea is that today in the next two sessions we look at the requirements in ore detail.
... Also need to look at whether we want to adopt them.
... Maybe some that have greater impact than others.
... Asking for support for policy templates, temporal relations etc.
... So we'll go through them one by one.
... Paper on the wall to help us draw things out.
... After afternoon teas we look at the Notes we've been talking about. We've promised a formal semantics doc

Sabrina: What's the plan beyond the end of the WG?

renato: We have the ODRL CG that can continue to discuss things. Maybe new profiles.
... Can't produce more Notes/Recs but we can produce CG Reports

benws: Can Phil tell us... we write the docs... and then request they become Recs. What issues/hoops are there?

<Sabrina> the implementations are the key thing

<Sabrina> you need to have a wide review... reach a large audience

<Sabrina> the eventual outcome is a vocab

<Sabrina> we are not chartered to do AC

<simonstey> we could use https://www.w3.org/TR/2013/NOTE-prov-implementations-20130430/ as orientation

<Sabrina> In the case of a vocab you need to show implementations to prove that people are using it

<Sabrina> Each term should be used twice.....

<Sabrina> These can be spread across different profiles

<Sabrina> Timeline is designed to get you to finish the documents

<Sabrina> Then you work on the implementations and go to the director with evidence of adoption

<Sabrina> recursive process - going back to remove terms that are not used

<Sabrina> This takes time....

<Sabrina> Phil hopes for updated publications based on the discussions here and then 1 or 2 revisions afterwards

<Sabrina> A good spec has an extension mechanism

<Sabrina> You can use this to meet the requirements

<simonstey> +q

simonstey: Extension mechanisms, makes sense. Profiles make sense for thata
... You can extend the functionality. Think about fleshing our ODRL a little that way. Moving into the realm of extending cf. core.
... The profiles part is a good path.

benws: On Profiles - are serialisations profiles?

phila: No.

Sabrina: You said two implementations. Does a profile count as an implementation.

phila: Talks about profiles being part of something else which would be the focus.

renato: Tomorrow it's more on the specs, using GH issues. See if we can assign people to take on things.

[Coffee Break]

<scribe> chair: Ben

<scribe> scribe: kerry

<phila> scribeNick: kerry

Requirements

benws: go thryu requirements

ben: 1 and 3 are similar

benws: victor, number 1 comes from you -- [reads out] licence for an asset only people over 18 years
... a filter on the assignee

victtor: yes

benws: and re3 is similar e.g. only targets equititues in the dataset -- ie referencing just a subset of the data
... req 1 contrains the assignee and #3 on the data -- both could be handled as a "filter" constraint

<michaelS> simonstey could you mute too, please

benws: perhaps just defign the asset more sclosely os ODRL does not have to scope what the target is.
... but in my work you are given an identifier and you need to target and having to come up with a descrition is difficcult for out workflow
... have you looked at contraints outside odrl?

victor: question is whther this is in or out? It would be nice to have in but it could be ok to be out

paulj: what happens with odrl meets a contraint/entity it does not know?

benws: we would refer to its own namespace, eg a definition of database columns
... althgouhg "only 18 can watch this film" appeals to me to be inside -- it soulds odd outside

paulj: are we saying waht contraints could be written and how?

benws: we could be neutral

paulj: if we cannot do the verification (e.g has the authority, has the ticket) that is a problem

benws: it could go either way in drl - but it does not allow contraints on assignee or data target

paulj: it is the semantics of the experession, not the implementation

[agreement]

benws: does anyone else have any more comments on contraints?

renato: we would have to update the contraint model to be more flexible

phila: the implications are profound -- mentions "verifiable claims" task force that is in that space

<simonstey> fwiw, http://w3c.github.io/dwbp/vocab-dqg.html#ExpressConformanceWithPolicy

paulj: but it does not matter to us -- we just need the expression, not the validation mechanism

<Sabrina> Validation is outside of scope.... This is the work of the verifiable claims task group https://www.w3.org/Payments/IG/wiki/Main_Page/ProposalsQ42015/VerifiableClaimsTaskForce

phila: if the duty is on the asset owner to verify, then we need to be able to say that the obligation is on the receiver to verify

renato: saying the assignee must verify is different to saying the applicant must have the property

phila: introduces eric prudhommeaux

paulj: the more i think about this -- it is a big deal ( refers t o an example about media)

benws: a cheap way would be to create a duty like "end user is an 18 year old" but this overloads end-user

renato: we have a contraint on the recipient of the right -- as long as we can implement age over 18 as a uri
... victor wants to generalise the contraint model

benws: but can we do this using the url method you mention?

<simonstey> one could define a SHACL shape for the assignee

Sabrina: is there also contratints on assets?

renato: no
... there is a contraint type on the use of the assent, but not on the recipient at present

benws: with some balck magic somehow realtes to the assignee

renato: but they might be different parties. looks like we need to allo contraints on other parties not just end user

benws: by just using an existing constraint on permission will this do it all?

paulj: a list of exhaustive cases will be very long and we will wish we structured it better

benws: we have a simple model now -- will this become too complex?

Sabrina: you could potnetiall have contraints on party, assrt, everyone etc

paulj: yes, and on the system, a lot of ripple thru

victor: we have contexts like "industry" too, if we allow specific declaration

benws: i agree it would be good to pin them down

victor: this requires only onle more element in a constraint

Sabrina: agrees contraint needs more structure

renato: 2 things: updating constraint model and adding it to the other entities

<simonstey> constraint vs. scope?

benws: so do we need that all contraints say what kind they are? (e.g. part, asset).

ericP: [lost it]

renato: yes, and aslo adding duties to publishers and others up the chain

s/alo/also/

renato: clearer -- do we want to do it?

benws: benefit is expressivity and accuracy -- cost is more complex objets all over the place

s.objets/objects/

victor: in rdf this is not a complexity increase -- although it may be for others

phila: agrees -- easy in rdf

benws: yes painful in XML -- maybe we should pririties rdf as out main serialisation

s.our/our/

phila: flattening it out into an xml tree should not be too hard.

some agreement

renato: taking note of R.DM.01 and R.DM.02 on butcher's paper
... will extend constraint model and add constraints to party and to asset

RESOLUTION: extend constraint model and add constraints to party and to asset

R.DM.03 introduce policy template

renato: a template just says there is going to be an action but we do not know what it is yet

benws: odlr assumes you point from the policy to the asset,but we need to point from the asset to the policy
... you could do an inverse but this is implicit -- should we make this explicit?

renato: do we have a reuquirement?

benws: it is inhereent in several

victor: [ puts on screen]
... I want you to see this text..[reads] this example , see the asset, licence, policy. But policy makes no reference to the asset and it needs to

renato: but we can only define the policy -- the rest defencs on the application world -- not sure how to do it in say xml

phila: what is wring with this? using dct:licence

renato: licence could be a set, template,

benws: inverse of t arget we need
... it is just the inverse of the target relationship

phila: licence is in iana link registry -- this works

benws: but might not be a licence -- could be something else
... "istargeteby"?

phila: "targetedBY"
... but no -- no link from asset to policy

renato: agrees

benws: the last thing we want is to change the policy every time there is a new asset

phila: in POWDER it would be describedby

benws: we need to b e able to move from an asset to the policy that controls it -- need inverse of ordl:target relationship

ericP: powder did not have a link back from desribed object to description

s/poder/powder/

[scribe notes that speaker benws above should all be read as "benws"]

[phila and eric discuss how powder works]

phila: suggests that poder may have some useful bits n it for this

benws: this is one form of templated policy: a policy type?

Sabrina: agrees

benws: so does it need more tying to templating -- will there be other properties left open for later instantiation?

victor: [explains example]... these properties need additional triples
... people assigningpolicies to assets know what kind of policy they need and htey know they need to fill in some fields

benws: is this an implementation issue -- does not affect odrl?
... breaking it up into separate implementations for parts may not affect our model

victor: want a template allowing description of parties to b e added later

renato: [restates]

benws: the quation is whther it is a policy before it is filled out?

renato: yes -- doe we want an incomplete policy (a template) to be a policy, requiring template fields to be instantiated before finalised?

victor: e.g. language resources communiity portal, using templates with some bits to be filled in

benws: yes, that is a user interface but does it matter to us that an unfinished policy is part of our language?

ericP: question of interoperability

reanto: are there sytems being built tht want to exchange templates?

Sabrina: importance of validation -- should validation reject all incomplete policies?

<ericP> kerry: to me, this is useful if an incomplete policy is instantiated as a policy before it's completely filled in

<ericP> ... i would expect it to be legally binding on e.g. subcontractors who can conform to a piece of it

RESOLUTION: we need an inverse of target to point from an asset to the policy

benws: for the rest this sounds like a draft , can a template do this?

Sabrina: but we need versioning too

renato: we need to focus on what we need to share, not the internal process that software can manage

victor: aiming for contract negotitation lifecycle

ericP: could say we might get to this

Sabrina: or it could be an implementation detail

<simonstey> +q

phila: agrees
... by bot including it as a requirement we are not preventing the draft /template to be used

simonstey: [missed, on telecon]
... test cases could do this? define your own assignee to override certain property values

benws: i understand the direction -- may be complicated
... we do not have any use cases for this -- can we park it until we have better use cases that can be tested against using inheritence to implement templates

[agreement]

<simonstey> simonstey: was wondering whether ODRL's inheritance relation could be utilized for templating

<Zakim> phila, you wanted to talk about validation

RESOLUTION: requirement DM03 resolved as a call for improved use cases

phila: refers to work on shacl for validation against rdf

ericP: needs bnf grammar

renato: is on agenda

[ argument about time zones asnd lunch]

support versioning of policies

benws: requirement does not seem to match underlying use case to me
... but read requirement as it is... lets look at use case 04
... requirement seems to say we need to version policies which seems ok but maybe should refer to use case 04
... would like to take requirement at face value first and come back to req for referring from a policy to relevant version of regulation
... we iwll need to main history and replacement of policies but the question for me is whther that is any of our business -- can some other w3c thing be used?

phila: sucgets prov + hav and dublin core

eric: agrees

s/suggests/

<phila> PAV - Provenance, Authoring and Versioning

benws: but we need to check that it is valid

<victor> https://www.w3.org/TR/prov-o/

ericP: look ... maybe owl:time helps (on rec track in sdwwg now) .. think about whther you need anything more formal

Sabrina: temporal information came up on a number of calls

benws: do we need slighlty slack dublin core terms or do we need more?

victor: mayve w3c prov does this all

<nandana> https://www.w3.org/TR/prov-o/diagrams/starting-points.svg

[discussion about rpov terms isversionof, isreplacedby etc]

ericP: sometimes versioning breaks systems too easily
... might be cheaper to just use new identifiers

benws: for for thisour needs we do not have strong enough set of use cases -- so it would be better to look at other ontologies

phila: we did have some requirements for setting validity period

benws: we needs some use cases -- i will write them

<scribe> ACTION: benws to write use cases(s) for versioning needs [recorded in http://www.w3.org/2016/09/22-poe-minutes.html#action01]

<trackbot> Created ACTION-22 - Write use cases(s) for versioning needs [on Benedict Whittam Smith - due 2016-09-29].

[some discussuion about how versioning and valid time should work]

<simonstey> +q

Sabrina: is there a workaround for now?

benws: lets park until i find the use case

simonstey: my use case is to uniquely identify a version of a policy
... if you generate a policy you have a uri for it and i do not need a specific property fo the version nor that all the conditions expire at some time

benws: could be one whereby you want to know what price we sold this for 6 months ago -- to travel from hte asset back in time?

<Zakim> phila, you wanted to talk about versioning

renato: easily implmented in a system -- question is whether we need to include it for interoperability

phila: explains W3C method of 2 uris -- fixed and latest

benws: looks like validity is the issue -- maybe we should provide our own terms.. does valvidity apply o a policy or a permission?

renato: for a permision can use normal constraint mechanism

benws: no, we need that this permission to do x is only valid in april or may
... i don't thnk it is covered

renato: can say from the first of april to 2nd may

benws: we mean thsi permission is only valid to 21 dec 2015
... so you want to say that the permission that was there is no longer valid

Sabrina: why not do a new policy?

[discussion between ben and sabrina and that should this be an expiroy on the policy, not at the level of the permission?]

<simonstey> +q

simonstey: a policy may have different time permissions attached to different permissions in the policy, even though the policy itself remains valid

Sabrina: if there is no time beteen the permission being inactive and the policy being inactive -- they can be the same

benws: we do need to manage the notion of whther a poliscy is valid

RESOLUTION: we do need to manage the idea of validiy and park for later what is the realtion bete=ween validity of permissions and policy

simonstey: we do need that multiple permissaions can expire at differnt times and the policy remains active at least until all permissions expire and maybe even afterwards

benws: but we might end up with empty husks of policies with no permissions

come back to thsi when we have reviewd requirements relating permissions to policy

renato: all permissions and obligations at atomic level have dates atached

break for lunch!!!!

<michaelS> Does it make sense to put myself on the speaker queue ??????????????

<renato> hang on....

<renato> Webex is playing up...hang up

<renato> nope - webex here still thinks we have the wrong meeting number!

<simonstey> https://mit.webex.com/mit/e.php?MTID=m461e1094f20ee7cabdcc3cccf4544c8e

<simonstey> this link is different from the one in the agenda..

<nandana_> /nick nandana

<renato> Should work now!

<michaelS> I can hear Lisbon

<Sabrina> Ben next requirement: POE.R.DM.05 Set a global price for all permissions of a policy

<Sabrina> Renato: If we accept this requirement the bundling of permissions it is a common requirement

<Sabrina> Can duties on a policy conflict with duties on a permission

<Sabrina> What are the semantics

<Sabrina> This is one of the things that complicate the model

<Sabrina> Renato: technically it does not complicate the model

<phila> scribe: Sabrina

<phila> scribeNick: Sabrina

If we have a duty on the policy you will need to fulfil this to get access to the permissions

If you have a set of permissons and they all have the same duty (e.g. cost 1 euro), can I do this

If there are three permission ad each has the same duty that has the same identifier the do I have to pay 3 times

victor: Understands the opposite... You only need to fulfil the duty once

<simonstey> isn't verifying whether a duty is/was fulfilled part of access control/drm?

Think of it as identified duties ... with 3 different identifiers

<simonstey> +q

<simonstey> -q

renato: If the duties are linked then you only have to pay once

<simonstey> (was about to say what renato said)

You might have different actions associated with the same asset

If you want to apply a duty to three premissions you need to apply the duty to the policy

<simonstey> +q

simonstey:

In the end it is checking if the duty was fulfilled

Simon agrees with Renato . If there is 3 permissions and 1 linked duty then once the duty is fulfilled then you have access

<renato> http://w3c.github.io/poe/vocab/#sc-example3

If there are three different duties then you have 3 different identifiers and you need to pay three times

renato: goes through an example

victor: presents an alternative PermissionToDuty relation with a permission on one side and Duty on the other

Although he does not support it as it would complicate the model

renato: going back to the original usecase - a duty applying to the policy

you need to link to this new duty from all of the permissions

benws: it is a nice shortcut - however might be overly complicated given that policies will be machine readable

Renato: we can not verify duties

benws: when we write the spec we need to be clear about the duties

<scribe> ACTION: benws to please read section 4.5 of the information model to see if one duty applies to multiple permissions [recorded in http://www.w3.org/2016/09/22-poe-minutes.html#action02]

<trackbot> Created ACTION-23 - Please read section 4.5 of the information model to see if one duty applies to multiple permissions [on Benedict Whittam Smith - due 2016-09-29].

renato: this requirement is already satisfied by the existing model

POE.R.DM.06 Support relative time constraints

benws: The source UC is not named correctly should be relative time constraints

phila: Michael would you like to comment as this is your UC

benws: I guess we need a property for a time anchor and an offset
... it looks like a vocab rather than a model issue

phila: but we don't have an anchor time

benws: given them a property that they can slot their time into

phila: thats an event

renato: so you could have a constraint saying you can distribute photos to do with a football match 1 hour after the event
... and another constraint for 30 an hour after the event

michaelS: there are different ways to express a particular point in time
... you have to check when the game ends and start a ticker

phila: the trigger is an event and the event is the end of the match
... in vocab terms you need a separate class for relative time period with a property trigger and the values will be event

renato: The permission will have a constraint and the constraint will have an event

benws: a constraint on the constraint

phila: where does validity going on the policy or the permission?
... this should go in the same place

renato: there is an event on the permission

simonstey: it depends on what we want... the semantics should help determine if there are conflicts and how to resolve them
... i don't think ODRL should verify constraints

benws: usually there is one constraint but here we have a dependency between constrains
... how to we know that the 2nd property applies to the 1st property

michaelS: I disagree.... it might be necessary to have a date and a reference for the date.... it is necessary to define from which calendar

benws: when you need to set the constraint to refer to another constraint the semantics doesn't allow this

michaelS: This requires a chance request

s / chance / change /

<simonstey> what about recursion & loops?

phila: you can have something specific relative time instead of datetime
... or generalise to a complex constraint

(trigger and period)

benws: we need this for embargos

<simonstey> doesn't owl time offer various ways for expressing intervals/durations/etc.?

benws: for an embargo you need an event and a time interval

phila: you can have trigger + period OR period + trigger
... OR start and end
... you need starttrigger + endtrigger + period

kerry: the w3c time could help here

phila: better to have a separate class

renato: lets look at the model
... the constraint needs to be a complex model - with multiple names, operators etc...
... cardinality will be 1..*
... bundle constraints

benws: we have talked a lot about this and in summary there are 2 ways to go A) invent complex objects and B) set of constraints
... park it for now

<phila> The model

phila: there are real use cases that need this

benws: are there other use cases that we should consider before making a decision

victor: this does not appear in Linked Data or language resources

POE.R.DM.07 Define a category property for class Party

<simonstey> +q

<simonstey> it's the scope

satisfied (conditional) on adding a constraint to the party

simonstey: this is the scoping of constraints, where you have a policy and you are saying it does not apply to all people but just those that are over 18

POE.R.DM.08 Add a category property to Asset

covered under requirement 1 - add a constraint to asset

POE.R.DM.09 Complex Constraints

<phila> benws: Offers example of no use in the UK after 7 days, that's 2 constraints

<phila> ... but you can use it outside the UK forever

<michaelS> see UC example: http://dev.iptc.org/RightsML-Combined-Example-geographic-and-time-period

same problem as before we need relative time

benws: leaving aside the relative time you have to split this into 2 permissions

Sabrina: depends on closed word or open world

phila: assume open world
... same as R.DM.06

POE.R.DM.10 Extended Relations

ya

<renato> 15 mins

<renato> back now

<renato> can u hear us?

<simonstey> +1

<nandana> scribe: nandana

<michaelS> +1

<scribe> scribeNick: nandana

extended relations

simonstey: we don't want to go into details but we want to verify and provide general guidelines
... explains the details

<simonstey> +q

<simonstey> -q

victor: I have a strong opinion, with the proposal <on the board> one can support disjuction and conjuction

<victor> (AND and NOT would serve, though)

benws: should be provide a graceful way of non-compliance to a policy, i.e., remedy?

<victor> And in any case, having these extended relations is very nice syntactic sugar

s / be / we

benws: benws doen't feel strongly about it, victor neither.

<simonstey> it was more of a nice to have feature

sabrina: if someone cares about it, it could be handled at the implementation level.

<simonstey> we lost you

<michaelS> Hi, WebEx has lost the audio connection

renato: we need a nice use case for this.

benws: will look into a use case.

<simonstey> +1

use case 5 - grouping party entities

<simonstey> +q

sabrina: this would apply not only to parties but to asests and policies too

<simonstey> ?q

benws: my sense is this is covered by requirement 1.

renato: we define an asset as a collection of assets and refer to the "bucket" of assets

sabrina: I felt that the constraints need much structure, at the moment there's a huge list of properties put together

<renato> ok...

sabrina: an option would be to type the constraints

simonstey: currently in ODRL you can refer to groups. One can group multiple parties together adding a property to them all, e.g., groupMember

benws: is this more a vocabulary issue rather than a model issue?

simonstey: this will require adding additional properties to existing entities, e.g., partyMember.

renato: if you assign a right to a group, all individuals will have that right. Do we have requirements like only 5 group members will have that right?

sabrina: for example, if we take peo w3c wg, we don't have a way to assign who are the members of that group.

phila: we don't need to define how group membership is specified. It has to be done elsewhere. We just define what is a group and what's its id.

renato: we need to note that in the model for parties, assets, and policies

<renato> Refer to PROV for grouping/members of Assets

<renato> and FOAF for parties

benws: is there a ontology that defines groups?

phila: FOAF.

kerry: schema.org?

benws: that's all for ucr.

Vocab

victor: we are not following any convention for naming the entities
... for example, some of there single word lowercase, some of them are camelCase.

phila: the best practice is properties starting with lowercase letter with camel case, classes beginning capital case.

victor: my concern is about the properties. Some of them are two words and some of them are not.

phila: do you have classes and properties with the same name only differ from capital and lower case?

renato: yes. e.g., prohibition

phila: that would not work in an international context. May be it is better to rename it to hasProhibition.

victor: consentingParty doesn't look right?

<renato> ACTION: renato change name of properties with same name (with lowercase letter) of associated Class [recorded in http://www.w3.org/2016/09/22-poe-minutes.html#action03]

<trackbot> Created ACTION-24 - Change name of properties with same name (with lowercase letter) of associated class [on Renato Iannella - due 2016-09-29].

sabrina: seems alright.

victor: these terms are not arranged by domain of application.

sabrina: it is not possible because at the moment one term might be related to one domain of application but later associated with other domains.

renato: can you give an example?

sabrina: what's the benefit of this classification.

victor: someone reading the spec for the first time might wonder why these terms and not others.

benws: this classifications might fit into profiles, but the terms we have at the moment seem mostly generic.
... for people coming from a particular domain, terms in profile may make sense

phila: what do you have in mind when you say "profile"?
... they should be machine readable, ODRL expressions can be validated against them

victor: validator might check an ODRL expression and say its ok and also to what profiles it complies with.

<phila> POE.R.V.01 Add rights holder terms to Roles of a Party Vocabulary

POE.R.V.01 Add rights holder terms to Roles of a Party Vocabulary

RESOLUTION: POE.R.V.01 Add rights holder terms to Roles of a Party Vocabulary Not accepted

POE.R.V.02 Add term for redepositing to Duty Vocabulary

benws: what's the context this occur?

victor: use case, you are allowed to make modification but you have to upload the modifications.

RESOLUTION: POE.R.V.02 Add term for redepositing to Duty Vocabulary Not accepted

POE.R.V.03 Add concept of 'unit-of-count' to Duty Constraint Vocabulary

renato: is it a constraint on a duty?

benws: yes.
... at the moment you can define a fine for a duty, but you can't express something such as $5 for an ip address

<scribe> ACTION: victor provide an example of how good relations vocab support 'unit-of-count' [recorded in http://www.w3.org/2016/09/22-poe-minutes.html#action04]

<trackbot> Created ACTION-25 - Provide an example of how good relations vocab support 'unit-of-count' [on Víctor Rodríguez-Doncel - due 2016-09-29].

RESOLUTION: POE.R.V.03 Add concept of 'unit-of-count' to Duty Constraint Vocabulary Pending

<victor> You can see the definition at http://www.heppnetz.de/ontologies/goodrelations/v1.html#hasUnitOfMeasurement

benws: unit-of-measurement doesn't fit well for this use case.

<victor> nice to read also http://www.heppnetz.de/ontologies/goodrelations/v1.html#UnitPriceSpecification

benws: we should add it if we don't find a better alternative

sabrina: it doesn't look like a constraint, more a property.

<phila> nod from the sidelines, Goodrelations now part of schema.org, prob better to use the latter incarnation

sabrina: it's more a constraint with a property.

benws: may be you can use directly in the constraints similar to "units". Just as another property "unit-of-count"

sabrina: does it only apply to financial?

benws: no. it applies similarly in reporting.

sabrina: can we define it using two constraints? one constraint referring to another constraint.

RESOLUTION: POE.R.V.03 ACCEPTED

benws: can we take a break from the requirements and talk about the other remaining stuff?

<simonstey> I would actually prefer to talk about the fs tmr morning

<simonstey> and I've to leave at 6pm

POE.R.V.08 Add terms to the Action Vocabulary

benws: "acknowledge copyright" seems to be common in many occasions.

victor: I would like to add "extract" and "reutilize", because they are very related to database rights.

<renato> ACTION: renato See property naming suggestions on page 3087 in http://subs.emis.de/LNI/Proceedings/Proceedings220/3081.pdf [recorded in http://www.w3.org/2016/09/22-poe-minutes.html#action05]

<trackbot> Created ACTION-26 - See property naming suggestions on page 3087 in http://subs.emis.de/lni/proceedings/proceedings220/3081.pdf [on Renato Iannella - due 2016-09-29].

<renato> scribe: victor

<scribe> scribe: victor

the requirement is "Add the action contract, acknowledge copyright and acknowledge to the Action Vocabulary"

benws: acknowledge means that "I acknowledge that X is the rightsholder"

<phila> victor: The database directive defines extract and @@

<phila> ... When I converted instant licences to this... are suitable for use as data licences (CC4 cf. version 3)

<phila> victor: I suggest adding one term

victor: database law defines extract and re-utilize
... which can be summarized within one single action ("databaseAction")

renato: does not the ODRL "extract" suffice?
... we do not want legislation-dependant definition

?

victor: CreativeCommons 4.0 and CreativeCommons 3.0 differ also in a sentence that says "whereever it applies, database rights are also waived"
... without a specific term, CC3.0 and CC4.0 would be undistinguishable

phila: no explicit mention to laws can be made, this is a matter of profiles.

renato: perhaps re-utilize is secondaryUse

sabrina: secondaryUse is also interesting for the GDPR (data protection)

benws: this is a matter for a profile, even if tiny

RESOLUTION: database law terms are rejected
... POE.R.V.08 accepted

POE.R.V.09 Add Linked Data related actions to Action Vocabulary

renato: the criteria to declare terms is "to be of frequent use". We have a couple of dozens.

benws: I like at least "query"

phila: there exists the concept of "superconcept" like "access" which embraces "query" or "read", etc.
... is "query" that important?
... 28 are perhaps too many actions as to keep in mind

<renato> http://w3c.github.io/poe/vocab/

renato: some of them have been deprecated, as it can be read in the current version (just linked by renato)

phila: some of the existing terms fall under one single category
... perhaps we could have a few terms, and then in a github repo maintain a non-normative larger set

victor: besides the general/specific relation there is another relationship: implication (making this action implies making this other)

renato: We have two options. Either we review the terms, removing the seldom used actions (watermark, textToSpeech, etc.) or we take the majority of them into another document.

sabrina: what about leaving them in the document, but distinguishing between normative and non-normative
... categorizing would be an option, another one would be going to the bare bones

benws: subclassing is always useful

renato: there is also "inheritance" in ODRL
... We have 12 actions for duties.
... we might distinguish between actions for duties and actions for permissions

phila: the text specification is generated from the ontology automatically

renato: the text specification is generated from the ontology automatically

benws: time is getting over. What are we discussing tomorrow?

renato: we need a new organization of the terms in the vocabulary (normative, non-normative). normative being really small.

benws: we need debate on that.

sabrina: Risk: if we go into details, we shall be discussing the whole day on that list of terms

<renato> thanks simon!

<simonstey> see you guys tmrw

benws: we have about 10 terms to be discussed, they can be reviewed tomorrow again

Agenda for tomorrow is here: https://www.w3.org/2016/poe/wiki/Meetings:TPAC2016

<michaelS> Tomorrow I can join only after 15:30

<renato> bye

Summary of Action Items

[NEW] ACTION: benws to please read section 4.5 of the information model to see if one duty applies to multiple permissions [recorded in http://www.w3.org/2016/09/22-poe-minutes.html#action02]
[NEW] ACTION: benws to write use cases(s) for versioning needs [recorded in http://www.w3.org/2016/09/22-poe-minutes.html#action01]
[NEW] ACTION: renato change name of properties with same name (with lowercase letter) of associated Class [recorded in http://www.w3.org/2016/09/22-poe-minutes.html#action03]
[NEW] ACTION: renato See property naming suggestions on page 3087 in http://subs.emis.de/LNI/Proceedings/Proceedings220/3081.pdf [recorded in http://www.w3.org/2016/09/22-poe-minutes.html#action05]
[NEW] ACTION: victor provide an example of how good relations vocab support 'unit-of-count' [recorded in http://www.w3.org/2016/09/22-poe-minutes.html#action04]
 

Summary of Resolutions

  1. extend constraint model and add constraints to party and to asset
  2. we need an inverse of target to point from an asset to the policy
  3. requirement DM03 resolved as a call for improved use cases
  4. we do need to manage the idea of validiy and park for later what is the realtion bete=ween validity of permissions and policy
  5. POE.R.V.01 Add rights holder terms to Roles of a Party Vocabulary Not accepted
  6. POE.R.V.02 Add term for redepositing to Duty Vocabulary Not accepted
  7. POE.R.V.03 Add concept of 'unit-of-count' to Duty Constraint Vocabulary Pending
  8. POE.R.V.03 ACCEPTED
  9. database law terms are rejected
[End of minutes]