Machine-readable Rights OLE/ODRL - TPAC 2015 Breakout -- 28 Oct 2015

Ira: there's been an ODRL CG

-> https://www.w3.org/community/odrl/ ODRL CG

scribe: there's a draft charter-in-progress

-> http://w3c.github.io/ole/charter.html Open License Expression Working Group Charter (Draft)

Benedict Wittam Smith (Thomson Reuters): people want to be able to combine datasets

scribe: we need a solution to be able to do this at scale
... to know what permissions we have
... we need a machine-readable language to know what permissions we have, what constraints exist
... we also have a discoverability problem
... if I mix a number of datasets, what can I do with the results?

DaveCramer (Hachette): the trade book industry has b2b exchange languages

scribe: I think there's been an attempt to do something with rights in this area, don't know how well it's succeeded

TzviyaSeigman(Wiley): Wiley publishes professional textbooks and thousands of journals

scribe: Ben expressed the issues concisely
... a lot of the licensing right now is in PDF or paper form
... there's not a good way to express all the complexities of rights
... if we translate a book to another language we need to express all the rights someone in that other country has
... different regions, different languages, parts of books, with different provenance for components of the books
... I represent Wiley in many industry organizations too, including W3C
... in BISG who primarily represent trade publishers
... BISG has a rights vocabulary which is currently a list of words
... they would like to be able to feed this into a system
... e.g. express "this book comes with a t-shirt"
... BISG has a consensus vocabulary for US trade publishers

IvanHerman: BIC plays a similar role in the UK

Ira: we need to identify a useful area of overlap
... that will benefit a wide group of people

IvanHerman(W3C): I am the staff contact for the Digital Publishing Interest Group

scribe: we've had some contact with BISG while drafting the OLE charter
... I also was active in Data on the Web for a long time and know that the expression of rights in data has been an interest for a long time
... e.g. scientific data
... need the rights attached to the data combined with the rights associated with the scientific publication
... and then the rights attached to software; what cna we use, reuse, etc.
... e.g. are you allowed to reuse [software] libraries
... I think there are lots of commonalites in these three areas

<dauwhe> Ralph: I'm w3c staff as well

<dauwhe> ... we want a common vocabulary around rights to help define scope

<dauwhe> ... I expect the participants will have diffferent connotations of common terms

<dauwhe> ... building a shared language

<dauwhe> ... that's one advantage of starting with ODRL

<dauwhe> ... then decide what changes we need

<dauwhe> ... but use ODRL definitions of terms

AxelPolleress(Vienna): we've been monitoring the ODRL CG

scribe: our perspective comes from Open Data
... integrating Open Data with different licenses and rights attached to the data
... we're interested in mechanisms that allow automatic aligment or combination of the licenses
... we'd like to at least partially automate the combination
... we've looked at how far you can describe, for example, the CC licenses in ODRL

FelixSasaki(W3C&DFKI): I work in natural language processing, including machine translation

scribe: in this area usage of data is common; for training corpora, lexicons, ...
... these data sources have been de3veloped outside the Web and are being moved into the Web under the "brand" of linguistic linked data
... looking at the European Data Portal you see the challenges of using open source corpora with in-house corpora
... some corpora have portions that are free and portions that are not free
... quality assurance processes can have a licensing aspect
... also relates to having licenses in various (natural) languages
... applying licenses to translation workflows

OlivierThereault(BBC): some of my colleagues have been in the ODRL CG

scribe: as well as the Linking Cultural Data CG
... BBC has a goal to help lots of cultural organizations expose their data on the Web
... and licensing comes into this

Ira: are there different restrictions on different parts of different archives?

Olivier: it's complex; we have our own archives we want to expose
... we also serve other groups' archives for them
... so we are both a producer and consumer

Kazue: I am working on personal data stores
... I want to attach a policy to my data to tell others what to do with my data
... also, if people offer me terms and conditions for use of my data I want the software for my data store to be able to determine whether I want to use the data

DanBrickley(Google): Google is likely to be skeptical about the complexity of these systems to scale

scribe: e.g. even robots.txt is super-simple but it gets screwed up

<shevski> http://w3c.github.io/ole/charter.html

scribe: it would be amazing if we can do something more than robots.txt but I haven't seen it

OLE draft charter

Ira: let's take a look at this charter
... does it seem to have the right scope?
... is there enough overlap?

Ivan: given a vocabulary, it can be serialized in several ways
... what worries me is whether there can be an automatic system that merges two license expressions in any meaningful way
... in the Semantic Web we have tried to come up with reasoning engines
... some of these are deployed on an acceptable scale; e.g. OWL
... some, like rule languages, have not been deployed on a significant scale
... for those that have been deployed I am not sure of the size of the data they can reasonably handle
... I would guess that these sizes are low relative to the needs of Thomson-Reuters
... it may still be manageable at the level of rights on each component of a book
... beyond that I am skeptical
... does ODRL handle this?
... what can we do in a Working Group without prior experience

Ben: we intend to run proofs of concept on scalability of content derivations
... at the first level if we can jsut describe our licenses in a common language this is a huge step forward
... the next level is to be able to trace through provenance what license belongs to a particular dataset
... then we have the issue of whether we can automatically generate a policy when we combine datasets
... we may be able to use an inference engine to do this but it can also be done with a very simple rule engine
... finally there's the issue of generating permissions for these combinations of contracts
... we might not use ODRL to do this; we might compile down to a more performant representation

Tzviya: my colleague says his dream world describes policy relationships
... he thinks allowing the declaration of multiple policies on a resource would move the focus
... he thinks some simple extensions to ODRL can accomplish this

<AxelPolleres> Are we talking about https://www.w3.org/community/odrl/vocab/2-1/ ?

Ben: ODRL does have some strange directions in its relationships; these would be easy to swap

Axel: I wouldn't say that OWL or a specific rule dialect would necessarily be the right reasoning formalism

<danbri> ralph, the rrsagent logs seem to be empty

Axel: looking at the ODRL terms I see implicit connections
... e.g. 'reproduce'
... implies 'read'
... some violations of combinability can be discovered with simple mechanism

ivan: I want to understand what this working group is aiming at
... Ben noted a useful step one
... is it resonable for this WG to do more than that?
... if not, we should say so in the charter
... if we intend to do a simple rules system this should be carefully scoped in the charter
... there are limits to how far this might go; it may not scale

Ben: I don't think there's any prior art for combination of policies
... there's some academic work
... I think we will start getting answers over the next couple of months

Ivan: if there's no prior art then W3C might not want a Working Group to do fundamental research
... we would exclude from the charter to standardize a specialized rule language

Axel: no one proposed that

Ivan: the use cases require reasoning over data
... we have to be clear that we won't solve this

<fsasaki> http://www.w3.org/TR/ws-policy/

Felix: in Web Services Policy there's a way to combine service descriptions
... there's no inferencing at all
... this is widely deployed

<AxelPolleres> AxelPolleres: My goal would be at least to take https://www.w3.org/community/odrl/vocab/2-1/ and resolv, clarify disambiguation/semantics of the terms used in the vocabulary, but not standardizing some formalism to reason about it. I’t just about definint implicit relations among this terms and compatibility … along the lines what felix says.

Ben: derivation is intersection

Felix: it can be described as set operations

Axel: right

Ira: it might be straightforward in some cases and not in others

Ivan: we should not promise that we would do these rules in this WG

Ira: in the open data space it would be useful to have a way to combine and make explicit machine-readable restrictions
... I think some of this should be in scope

Ivan: I'm playing devil's advocate
... I propose that it should be out of scope for the OLE WG to do reasoning on the data

Axel: even set operations?

Ivan: we need to be slightly more flexible that than
... I don't want us to get stuck in research
... we've done that in the past and that's why I'm cautious

DanBri: ODRL has been around for 15 years

<Zakim> danbri, you wanted to ask about ODRL standards

DanBri: what lessons have we learned from its deployment?

Ira: I asked that too

Ben: AP is using ODRL
... describing wire photos
... .photo editors go through lots of photographs looking at restrictions; geographic, etc.
... AP is adding ODRL to allow the CMS to prefilter for the subeditor
... this is saving a lot of time for the editors to check

<ivan> http://www.w3.org/TR/dpub-metadata/

Ben: if we can automate 80% of this it will be a big advance

Ivan: the Digital Publishing IG published a Note on metadata in the publishing industry

<danbri> http://virtualgoods.org/2013/ reports the "9th W3C ODRL Community Group" meeting — there may be records from those events that can help

Ivan: this was based on a series of interviews done by the IG
... one thing came up
... IDEAlliance is also working on rights expressions for magazines, based on ODRL
... IPTC was involved
... so ODRL wasn't done in a vacuum

Ben: "license" cuts the scope far too thin
... that covers copyright
... other things that concern use policy are privacy, @@

Olivier: could we do a first iteration that only does licensing?
... and broaden to policy later

Alex: it's a terminology issue
... starting with "license" might be harder to broaden later

Ben: we're looking at this as a means to help us stay compliant
... with data residency requirements and privacy requirements
... we want a way to manage all these concerns
... limiting to just license limits the value to us

<danbri> re ODRL see http://virtualgoods.org/2013/ODRL_2_0_Revisited_VirtualGoods2013.pdf "Improving ODRL" sections

Felix: I think we may not need an additional mechanism, just another use of the same mechanism

Ira: issues/next steps/concerns?

Ralph: who's willing to submit comments on this draft charter?

Tzviya: Wiley will submit some

Ben: Thomson will submit some too

Ira: mailto:public-ole-comment@w3.org
... end of November?

Ralph: we want to fix the charter with your comments *before* sending it for Member review

[adjourned]

- DRAFT -

Machine-readable Rights OLE/ODRL - TPAC 2015 Breakout

28 Oct 2015

Attendees

Contents

OLE draft charter

Summary of Action Items

Scribe.perl diagnostic output