Permissions and Obligations Expression Working Group Teleconference

23 Sep 2016
See also 22 Sep 2016


benw, bob, ivan, paulj, renato, sabrina, simonstey, victor_, victor, phila, michaelS
paulj, victor, Serena
Flipchart Images
Accept Later today/tomorrow Must but Satisfied Future Req Recs & Notes WG selfie Time line/F2Fs


<renato> scribe: paulj

POE.R.V.10 Add terms to the Roles of a Party Vocabulary

renato: Last ten vocab items to process now

looking at POE-RV10

victor_: need to specify both ends of compensation payment

<simonstey> +q

<ivan> trackbot, start telcon

RESOLUTION: POE.R.V.10 accepted

<renato> chair: Ben

POE.R.V.11 New Party Category Vocabulary

victor_: As discussed yesterday, need to associate constraints with a party.
... this is in hand so the requirement is covered

RESOLUTION: POE.R.V.11 satisfied

POE.R.V.12 New Asset Category Vocabulary

RESOLUTION: POE.R.V.12 Satisfied

POE.R.V.13 Temporal Constraint

<simonstey> scribe: paulj

POE.R.V.14 Using time spans in Temporal Constraints

POE.R.V.15 Reference to Source License

victor_: There are terms such as Dublin Core source
... Could be used here

renato: Many means of implementation not specified. Could be included.
... ...as examples.

RESOLUTION: inlcude Dublin Core Source as example

<simonstey> +q

simonstey: Is this an annotation or something else?

benws: An annotation

POE.R.E.01 Referring to external resources for defining payment fees

<simonstey> http://w3c.github.io/poe/vocab/#term-status ?

victor_: Include not the fee but a reference to a fee
... as a URI for example

benws: Is this core standard or an implementation issue?

<simonstey> -q

victor_: Semantics differ slightly

renato: party has attribute "scope"
... can we can specify that something needs to be dereferenced before use
... so datatype isnt the amount but a reference to something external

benws: Risk of creating two ways of doing the same thing

victor_: can just change text to admit references as well
... MPEG-21 REL has this capability

<simonstey> +q

simonstey: should avoid having multiple ways of expressing the same thing

<victor_> +q

ben: accept scope and new operator to indicate reference

victor_: Idea: as we now have template we could include there whether things are resources of references?

renato: Not needed

RESOLUTION: accept that scope should have a new operator to indicate reference

POE.R.E.02 Prove that a list of existing licenses can be expressed/encoded

RESOLUTION: To be included as part of a Note

Define rules for matching permissions/prohibitions against specific use cases

benws: Including minimum viable policy and ensure consistency

sabrina: To be discussed as part of semantic checking

<simonstey> which is usually the empty policy?

<victor_> :)

sabrina: Project forthcoming to crawl over multiple licences and return the least compatible constraint meeting all

RESOLUTION: discuss later

Define a language for controlling processing policies

benws: Out of scope?

Define processing rules for versioned policies

benws: there are means of dealign with versioning

sabrina: for interop, partners will need to agree version to be used

<victor_> paulj: if things are different they cannot share the same identifier

ivan: there are good techniques to handle this and it doesnt need to be normative in this dpcument

benws: Useful to have a URI pointing to a reference to a policy which is the latest version

RESOLUTION: included best practice in a non nomative note


<victor_> +1

<victor_> -q

<victor_> q

Support to auto-generate a policy from a template plus provided parameter values

RESOLUTION: not needed now

Guidance on Rights Assignments through Aggregation and Derivation


RESOLUTION: satisfied

Guidance on Specifying Subsets of Assets

RESOLUTION: satisfied

Guidance on Conflicting Permissions

benws: Does this actually make sense?

sabrina: exceptions and overrides subject to rules of expression precedence

ivan: Can become very complicated...

benws: This is a request for guidance not normative treatment of conflict

renato: There is already guidance on conflict


Make policies accessible by URL

RESOLUTION: include best practice in a non nomative note

<renato> https://www.w3.org/2016/poe/wiki/Requirements#POE.R.DM.06_Support_relative_time_constraints

<simonstey> https://www.w3.org/TR/owl-time/

<simonstey> +q

POE.R.DM.06 Support Relative Time Constraint


simonstey: Can restrict timing by using time ontology

benws: Need both event and time offset.OWL time will not suffice here.
... time itself isnt the problem

ivan: Can we place a constraint on the whole ODRL graph?

<simonstey> https://www.w3.org/TR/owl-time/#scheduling

renato: need to be able to make constraints dependent on each other

<simonstey> :meeting a :Interval ; :hasBeginning :meetingStart ; :hasDurationDescription :meetingDuration .

<simonstey> :meetingStart a :Instant ; :inXSDDateTime "2006-11-05T14:00:00-8:00"^^xsd:dateTime .

<simonstey> :meetingDuration a :DurationDescription ; :minutes 45 .


renato: can make scope of a constraint refer to an other constraint

ivan: that is reification and something we are seeking to avoid
... this is a known recurring RDF problem

<simonstey> https://www.w3.org/TR/owl-time/#time:TimePosition

simonstey: Note complexity that will result from addressing this

ivan: Note that Provenance WG hit same problem.
... uses relationship including qualified version

<renato> https://www.w3.org/TR/2013/REC-prov-o-20130430/#qualifiedAttribution

ivan: Note complexity and that this issue isnt in the primer!

<renato> 3 options -

<renato> 1 - prov option

<renato> 2 - constraint on constrain

<renato> 3 - extend model (eg "embargo")


Digital Publishing IG

<renato> Guests: Tzviya Siegman ,Wiley

<renato> Guests: Bill Kasdorf

<simonstey> could you share the link also on irc?

<Sabrina> Renato: Digital Publishing IG - Use Cases (Tzviya Siegman)

<renato> link coming...

<simonstey> thx

<Sabrina> Bill: Book publishing industry worked on rights vocab and payments scheme however it never went anywhere

<Sabrina> what they were trying to express was complex and the big trade partners kept adding complexity

<tzviya> GoogleDoc with use cases https://docs.google.com/a/bisg.org/document/d/15nbqGY20IIGbTQOzKxzw59TLzwfPpRZu-1KKA97phKg/edit?usp=sharing

<renato> Simon - does that link work?

<Sabrina> ...They are interested in the rights associated with granular items (e.g. image in an article, postcast, excerpt from a book all in the one book)

<simonstey> You need permission -> I requested access

<tzviya> editable link https://docs.google.com/a/bisg.org/document/d/15nbqGY20IIGbTQOzKxzw59TLzwfPpRZu-1KKA97phKg/edit?usp=sharing

<Sabrina> ... all coming from different sectors that all express their metadata differently

<renato> Simon - I just emailed you the document as well

<Sabrina> textbook publisher understands all of the metadata and all of the vocabs

<simonstey> got it, thx

<Sabrina> ...ODRL could be the common denominator

<Sabrina> ..AP create a 1/4 of a million assets a day therefore you need machine readabiility

<victor> (can anybody type the standards he just mentioned?)

<renato> ideaAlliance PRISM

<tzviya> JATS Journal Article Tag Suite

<tzviya> JATS has a metadata header

<tzviya> XMP metadata with images


<Sabrina> ... plus image data, XMP also used

<Sabrina> ... the need for the rights expression to travel with the asset

<Sabrina> ... these various sectors need to understand that they need a standard model and vocab for specifying obligations and permissions

<Sabrina> benws2: they can develop a profile

<Sabrina> BISG are not usually technical they are contracts and rights people

<Sabrina> tzviya: we should look at these use cases

<Sabrina> benws2: author royalties are very important to TR

<Sabrina> bill: who owns the rights are relevant and does result in the determination of who gets payment

<Sabrina> tzviya: We should take a look at the use cases

POE.UC.22: Enhance discovery of library collection materials

<renato> note: UC numbers will be updated ;-)

<Sabrina> an important community that are often ignored are the library community

<Sabrina> They live and die by semantics

<Sabrina> Bill: This use case is about consistent metadata describing copyright

<Sabrina> ... Rights data associated with a digital resource work should be able to convey the whether or notthat is included in a library collection should always include permissions and obligations for library-to-library sublicensing inter-library loan are granted.

<Sabrina> ... interlibrary loan which relates to the digital resource as opposed to the physical book (where they own the book)

<Sabrina> ... in the case of digital assets they license the book

<Sabrina> benws2: Very similar to TR lots of assets and no idea how to license them

<Sabrina> ... ODRL has a very limited way to express copyright it would be good to go through the resquirements to see how well copyright is adressed in ODRL

<Sabrina> ... the idea to be able to reference items within items. As a standard we will not tell you how to identify your assets

<Sabrina> bill: the publishing industry would really like inheritance

<Sabrina> ... if you pull out an asset you need to pull out all the rights and sent them with the asset

<Sabrina> benws2: are you talking about aggregation?

<Sabrina> tzviya: Lets look at the later requirements e.g. who owns the copyright for chapter 2 along with all its subitems

<Sabrina> Bill: another example is embargo - an image has no idea of time however it cannot be displayed until a particular date

<Sabrina> ... another example is to create a video that they want to license a year later someone else wants to reuse that video but there is no way to know that they can't use it

<Sabrina> tzviya: POE.UC.24: Rights licensing data for e-book subscription services

POE.UC.24: Rights licensing data for e-book subscription services

<Sabrina> A number of opportunities exist for book publishers to sell works through subscription services, such as Scribd, Amazon Kindle Unlimited, and Playster, but they are unable to take advantage of these services because a consistent method for collecting and communicating subscription rights data has not been adopted within the industry

POE.UC.25: Improve internal rights management systems (large book publishers)

<Sabrina> Same use case just a different application

<Sabrina> Bill: I don't need to use ODRL internally but I will use it for exchange

<Sabrina> benws2: TR use it for text editing fields... It is a pull down of vocab.

POE.UC.26: Improve efficiency of foreign rights transactions (University Press)

<Sabrina> China wants to do Spanish for Dummies (language restrictions, internationalisation) same use case again just a different application

<Sabrina> ... the expressions need to be language agnostic

<Sabrina> tzviya: for publishing licensing in multiple areas is a very big area

<Sabrina> Bill: model and vocab needs to be language agnostic

<Sabrina> benws2: We will provide an extension model

<Sabrina> Bill: It would be good to have language codes such as THEMA or ONIX

<Sabrina> ... ONIX have a code, english word and a description

<Sabrina> benws2: Wiley in New York have a UI in english, send the policy to China and they should be able to see it in Chinese

<Sabrina> Bill: Books are priced differently depending on the country

<Sabrina> ... its not just the rights to license it could also be is the purchaser allowed to by it in a particular country

POE.UC.27: Disambiguate access permission from copyright permission (University Press)

<Sabrina> tzviya: Difference between license and copyright

<Sabrina> benws2: We need much more detail on this...

<Sabrina> paulj: Rights expression languages are not really suitable for rights

<Sabrina> tzviya: Can you provide more details on what you require?

<Sabrina> renato: fair use is not something that we are going to express

<Sabrina> Bill: Looking for the ability to specify rights but they could be superseded by copyright

POE.UC.28: Library collection management and access

<Sabrina> Bill: Looking for a translation into ODRL

<Sabrina> renato: We support this by promoting ODRL and marketing it so that there is an awareness of the outcomes of our work among these communities

POE.UC.29: Rights licensing for custom textbook publishing (higher education publishers)

<Sabrina> tzviya: Ability to build their own book from chapters from existing books and possible add their own content or something from the web

<Sabrina> renato: can you give some more information on the subscription model?

<Sabrina> tzviya: similar to the textbook use case... I subscribe to the service, they have books on HTML and I am only interested in 1 chapter...

<Sabrina> At the moment this is not automated

<Sabrina> Bill: You need a profile of the person, they are a student of a university and the university has a subscription

<Sabrina> ivan: I try to look way ahead ... candidate recommendation . way the technology is proberly tested consistency and usability... you can not start thinking about it early enough. We have use cases, and a real community, in the future it would be good to ask that community to verify what we have done

<Sabrina> ... The question is would BISG be willing to play that role when the time comes

<Sabrina> Bill: In general yes, however it's hard to know if we will be able to get people to do all that you require

<Sabrina> ... We should be able to get OCLC involved

<Sabrina> renato: Do they have to show implementations or just confirmation that the model and vocab are ok

<simonstey> https://www.w3.org/TR/2013/NOTE-prov-implementations-20130430/

<Sabrina> ivan: each working group defines their own criteria and asks the director if they are happy with our proposal

<simonstey> prov-o exit criteria -> https://www.w3.org/2011/prov/wiki/ProvCRExitCriteria

<Sabrina> Bill: It would be useful to know what the gaps are

<Sabrina> tzviya: Journal articles are the most important to us

<renato> POE thanks Tzviya and Bill

<simonstey> break? or do we continue?

<renato> 5 mins

<simonstey> kk

<victor> scribe: victor

Formal Semantics Note

<simonstey> https://www.w3.org/TR/prov-sem/

simon: We can take this specification as a reference for our spec
... second paragraph in section 1.1 is very clarifying
... formal semantics in ODRL can be stated as a set of description logic axioms, as well.
... this would naturally solve policy aggregation, conflict detection, etc. Enforcing is explicitly excluded.

<simonstey> https://www.w3.org/TR/2013/NOTE-prov-sem-20130430/#additional-axioms

simonstey: some conflicts can be avoided by declaring appropriate axioms.
... there are, for example, conflicts related to the duties or the constrains

benws2: the work is around the question of validity.

Sabrina: validity and conflicts.

<renato> new member: Konstantopoulos, Stasinos

Sabrina: a new ODRL member has much experience in formalizing. (who?)

<renato> who is also interested

ivan: how is the current charter?

renato: only two documents to be Recommendation

ivan: but you have also in mind about 4 notes. There may not be enough resources to actually edit the six documents.
... it is not tenable that one person can edit the documents in their own. also, renato, as chair, may not be that much involved.

Sabrina: a newly funded project will provide working force

ivan: only members of the WG can vote on the publication of the note

renato: changes needed to write the two recommendations are limited, as we don't start from the scratch.

victor: doing the formalization effort will improve the recommendations anyway

ivan: i see other more urgent tasks, like moving from JSON to JSON-LD
... this group has a strong bias towards RDF, and we do not want to disband the web developers (JSON lovers)

benws2: the value of ODRL for Thomson-Reuters is easing the task of validating compliance. For this regard, having a formal semantics would be a great value.

ivan: no doubt about it. but can everything be accomplished?

simonstey: (on additional logical implications, and how profiles can be improved if the note is made among other benefits)

renato: in the whiteboard, lists the 6 documents

(actually seven)

ivan: there is much to be improved in the two recommendations as they are. for newcomers, it may not be so immediatly understandable
... they need a lot of editorial work to make them sellable

benws2: when I first approached, it took me much time understanding ODRL.

ivan: they had the same problem in the Web Annotations group, and they had to illustrate it with plentiful of examples

benws2: the best practices may help at making things understandable

renato: we are moving in this direction, moving examples up

benws2: Renato, you should leave aside your years-experience and describe everything with new eyes
... 24 years of experience in this business are a heavy bag

ivan: the annotations document is now an example of good quality

<simonstey> lost you

<simonstey> no one on webex

<simonstey> back

ivan: owl and xml is less important

Serena: Renato should leave the lead on the model document re-engineering to me, as I have a fresher view than Renato.

ivan: which are the other notes?

renato: best practices would be examples

ivan: why not in the github as naked examples?

benws2: there are patterns in the practical expressions. The document would give context to the problem: "if this is your business model, this is the pattern"

<renato> ACTION: serena review info model to support "annotation model" style examples [recorded in http://www.w3.org/2016/09/23-poe-minutes.html#action01]

<trackbot> Created ACTION-27 - Review info model to support "annotation model" style examples [on Serena Villata - due 2016-09-30].

benws2: there are many types of ODRL users. Some are 24 year old techies, some are oldies who should understand.

victor: github repo may be useful for the ODRL mapping licenses

ivan: what is linked data profile?

victor: profile FOR linked data

ivan: i personally find it very nice, but indeed of least priority

benws2: I voluneer to edit the "Best practices".

paulj: I volunteer to edit "Best practices" as well

<simonstey> if we provide them, wouldn't they need to be audited/checked by a legal expert?

victor: legal experts may differ, also. it is a matter of interpretatoin

renato: we need a github repo of licenses

Sabrina: we will contribute along our project

RESOLUTION: We remove the ODRL Mapping Licences to a github repo referenced from the "Best Practices".

Sabrina: (and victor and serena) We cannot work on everything at the same time, so we can postpone this activity

<simonstey> break?

<renato> yes

<renato> 1 hr

<renato> We are back

<Serena> scribe: Serena

general data protection regulation - Sabrina

Sabrina: I recently launched a new lab in Vienna
... general data protection regulation

H2020 project accepted together with Ben

<scribe> … new general data protection regulations to be represented and model them ODRL to built on top of them

UNKNOWN_SPEAKER: two scenarios: (1) policies and transparency: who does what with my data
... (2) risk assessment for companies to address compliance checkinh
... examples of rules to have e.g., explicit consent
... who is doing what with the data
... model quite close to ODRL: constraints, actions, etc are the same
... you have obligations, and if you satisfy them then you're compliant
... parties: data subject -> party, other people like data protection officers, etc
... asset is personal data
... action is all regarding processing in the EU and outside the EU
... sometimes you have dispensation e.g., unless it is in the child interest etc
... the policy is the general data protection regulation, but we want the link to the articles, in each article there are many rules
... we will provide examples in n-triples

<simonstey> +q

UNKNOWN_SPEAKER: if you hava a duty, and you fulfill the duty then you have the permission to process the data

Sabrina: asset is personal data

<simonstey> there is a privacy type

renato: is this another kind of policy type?

Sabrina: this could be seen as a new policy type
... the H2020 project aims at designing a system to check compliance
... we will have legal guidance

renato: do you have inheritance among policies?

<simonstey> odrl doesn't allow for multi-inheritance

Sabrina: regulations contains articles, paragraphs and then duties

<renato> odrl can do "One Parent Policy to one or more Child Policy entities"

renato: are there actions from this project?

<victor> ...

<simonstey> which doesn't mean that all of those subpolicies belong together

benws2: every information with personal information can be the "input" for this project

renato: new policy type -> regulation

<simonstey> how does the privacy type relate to all of that?

<simonstey> http://w3c.github.io/poe/vocab/#term-Privacy

victor: for these policies it is important to keep provenance

<simonstey> https://www.w3.org/community/odrl/vocab/2.1/#section-21


<renato> action sabrina define a new Regulation Policy Type

<trackbot> Created ACTION-28 - Define a new regulation policy type [on Sabrina Kirrane - due 2016-09-30].

<simonstey> do we need the privacy type then actually?

<renato> yes, perhaps for more personal privacy polices

<simonstey> you dropped from webex

renato: use cases and requirements
... R.DM 04


scribe: what is the action here?
... do we want to be back to the group?
... any other comment or question about UCR?
... jason-LD as undelines by Ivan has to be considered

ivan: having both jason and jason-ld encodings is superflous
... json-ld should be enough
... we should forget xml/html
... is there a fresh market for pure xml?

* thanks simonstey *

ivan: what we do now for the annotation, we actually have a test suite with all examples converted from json-ld into turtle

benws2: do we have to show that there are implementations using xml?
... we can ask in the community group

<renato> ACTION: renato ask WG/CG - who is using plain XML? [recorded in http://www.w3.org/2016/09/23-poe-minutes.html#action02]

<trackbot> Created ACTION-29 - Ask wg/cg - who is using plain xml? [on Renato Iannella - due 2016-09-30].

ivan: I think that json and json-ld are the same
... and we all have to be careful to the discussion about using the @id or not, etc

<renato> action stuart Can we only have a JSON-LD serialisation? Will it impact RighstML?

<trackbot> Created ACTION-30 - Can we only have a json-ld serialisation? will it impact righstml? [on Stuart Myles - due 2016-09-30].

ivan: each example needs to be encoded in json-ld in the information model
... is there a plain to have the vocabulary defined in owl?
... what's in the current model?
... it says RDF/OWL encoding
... somebody did make the RDF/OWL triple ontology
... don't show examples with rdf/xml

<simonstey> +1 to ivan's point of view

ivan: we should not include rdf/xml in w3c recommendations

benws2: digital publishing people are claiming about the translation (e.g. in japanese), then this should be in the owl ontology

victor: who is going to do the translations?

ivan: we should not spend time on that

victor: question about the ontology connected to the formal semantics, will it have the same set of axioms or a superset?

<simonstey> agreed

ivan: owl ontology normative or informative?

Sabrina: Phil suggested to have two recommendations, with many of the items are not normative and other are

ivan: if we say "yes" we need to have consistency proved by at least two tools
... which means that all the statements in all the examples would be consistent with the ontology etc

victor: in favor of making it normaive

ivan: we have to keep in mind that if we decide to go for a normative owl ontology than we will have to follow a certain process to prove its consistency

Sabrina: then we will keep the non normative items right?

ivan: on the one hand, for end users it seems strange not to have an owl ontology, on the other side having a normative ontology with non normative items…
... let's go for the normative but complete

<renato> 15 min break

<renato> we've back

<benws2> We're about to restart.

<benws2> nick/benws

* :) *

renato: dates for the next working draft?
... any other issue to be discussed in the last hour?
... profile seems a good way to show odrl is used
... there is a whole section about the profile

<simonstey> +q

benws2: there are different levels of profiles
... we can automate the validation of licenses, for that we need a formal semantics, these are different levels of profile (profile in a profike)

simonstey: we have to be careful about the way profile works

ivan: we have to define what this MUST means?
... in normative terms

renato: there is no machine readable representation

ivan: I should have the right to ignore it
... I would put all the statements in stronger terms, e.g., "some requirements…" -> "we must document…" otherwise it is not normative

<simonstey> a core set of concepts that MUST be provided/used by all profiles?

ivan: why having two URIs?
... we're talking about an rdf model
... xml is a possible serialization, but the model itself is in rdf

victor: the ontology plus the text is the proposal

ivan: but the heart of it is just rdf
... don't use "deprecated" in the profile text

renato: what about a profile being machine readable?

benws2: it is optional, some profiles are incredible lightweight
... we should allow people to come to the level they prefer
... the best would be to point to some examples of profiles

<simonstey> I dont think you should

ivan: will SHACL become a rec?

simonstey: next year probably

renato: can I use it for reasoning?

simonstey: non actually for reasoning
... you can use it to check the profiles

Horizontal Reviews

ivan: horizontal review
... we have to ask the experts to review the document
... I'm not sure about security, but being a vocabulary I don't think it's an issue
... privacy issues
... internationalization
... in the annotation group, we made a mistake and we contacted them too late
... ideally we should have a model document by January to have a review in January
... not a final version but a reviewable one
... what we can do now is to look at the new version of the information model to avoid internationalization issues

<renato> action phila propose date/time for the virtual meeting in Nov/Dec

<trackbot> Created ACTION-31 - Propose date/time for the virtual meeting in nov/dec [on Phil Archer - due 2016-09-30].

<renato> Proposed next F2F in March 2017

<simonstey> lost you

<renato> yes

<simonstey> see you guys!

<simonstey> bye bye

<renato> thanks Simon!

Summary of Action Items

[NEW] ACTION: renato ask WG/CG - who is using plain XML? [recorded in http://www.w3.org/2016/09/23-poe-minutes.html#action02]
[NEW] ACTION: serena review info model to support "annotation model" style examples [recorded in http://www.w3.org/2016/09/23-poe-minutes.html#action01]

Summary of Resolutions

  1. POE.R.V.10 accepted
  2. POE.R.V.11 satisfied
  3. POE.R.V.12 Satisfied
  4. inlcude Dublin Core Source as example
  5. accept that scope should have a new operator to indicate reference
  6. POE.R.E.02 Prove that a list of existing licenses can be expressed/encoded To be included as part of a Note
  7. Rules for matching permissions/prohibitions against specific use cases - discuss later
  8. Define a language for controlling processing policies - included best practice in a non nomative note
  9. Support to auto-generate a policy from a template plus provided parameter values - not needed now
  10. Guidance on Rights Assignments through Aggregation and Derivation - satisfied
  11. Guidance on Specifying Subsets of Assets - satisfied
  12. Guidance on Conflicting Permissions - No action
  13. Make policies accessible by URL - include best practice in a non nomative note
  14. We remove the ODRL Mapping Licences to a github repo referenced from the "Best Practices".
[End of minutes]