[odc-discuss] [ODC-Discuss] machine-recognizable labelling

Jonathan Rochkind rochkind at jhu.edu
Tue Jan 27 15:25:14 UTC 2009


The scenario we're talking about is also complicated in that a record 
can exist in multiple databases. Institution A originated some records 
that they dedicated with the PDDL. These records may live in a database 
with other records not so dedicated.

Institution B then downloads a subset of instution A's records that are 
PDDL dedicated into their own database (where they may also exist 
co-mingled with non-PDDL records). Institution B then may make 
modifications to these records that they also dedicate under the PDDL 
(which they might share back with Institution A in an automated 
fashion--or might not).

Then institutions C, D, and E-Z get into the picture. :)

It's in fact to support this scenario that the PDDL is so important. But 
it also leaves me confused as to how to apply it appropriately, 
especially since your explanation of the PDDL seems more applied to 
_databases_ than to _records_.   I understand in broad strokes the legal 
reasons for what Jordan is saying---but I'm still confused as to how to 
apply it appropriately.

I think it's this kind of scenario that in large part motivated Talis's 
funding of the PDDL, so in my mind the project isn't really useable 
unless we can explain how to legally apply it in this situation.  In 
fact, biblios.net is _already_ claiming that their using the PDDL; who 
knows if they are doing so appropriately.

It would be awfully unfortunate if it turned out that the PDDL wasn't 
appropriate for this situation; it's the whole reason that many of us 
are interested in the PDDL in the first place.

Jonathan

Rob Styles wrote:
> Jordan,
>
> Could you give some background on the first two points - that the URI is not
> enough and the text would need to be in. That would help us get clearer on
> what the appropriate mechanisms might be.
>
> The scenario Jonathan and I are considering ate the use of the license for
> web-based Linked Data. In this case the response from a server may be only a
> small part of the db, just the response to the query. It's this case that
> we're trying to understand, as we feel there is a need to make the license
> clear even in the small extract, yet attaching the entire license text to
> each response would be a substantial cost.
>
> rob
>
> On Tue, Jan 27, 2009 at 12:57 PM, Jordan S Hatcher <
> jordan at opencontentlawyer.com> wrote:
>
>   
>> On 13 Jan 2009, at 17:33, Jonathan Rochkind wrote:
>>
>>  It looks like the other part we need though, according to Jordan, is
>>     
>>> instructions for how you legally release your data under ODC PDDL.
>>> Jordan suggests that just putting a URI on your data is not sufficient,
>>> you need to make a legal release via some other means. That's what we
>>> need a template/example for.
>>>
>>> The technical stuff Rob talks about below, I (and Rob) are capable of
>>> coming up with examples. It's the legal stuff for what you need to do to
>>> ensure your data really has been released according to your intentions,
>>> that we don't neccesarily have competence for without some legal aid.
>>>
>>> Rob, does Talis want to pay for a bit more lawyer to get that done? :)
>>>
>>>       
>> Thanks for the referral Jonathan!
>>
>> But seriously -- there really isn't a "legal" bit here. Just a need to
>> create a document on how best to use the PDDL with the following parameters:
>>
>> -- Simply putting a URI in the metadata isn't enough; and involves
>> -- Putting the text of the PDDL somewhere in the database
>>
>> The key is that:
>>
>> -- users know *exactly what is covered by the PDDL
>> -- users can easily find out the text of the PDDL
>>
>> That's really a job for everyone that works with databases everyday and
>> knows what would work best as a practical matter.
>>
>> Thanks
>>
>> ~Jordan
>>
>>
>>
>>
>>
>>
>>     
>>> Jonathan
>>>
>>> Rob Styles wrote:
>>>
>>>       
>>>> I am now...
>>>>
>>>> I agree with Jonathan, it seems like a great opportunity to make ODC PDDL
>>>> more usable - by providing
>>>>
>>>>  - A URI that associates specific versions of the PDDL with a release of
>>>>  data
>>>>  - A template for adding the license to RDF specifically (that may have
>>>> to
>>>>  include the specific text).
>>>>
>>>> Hello everyone.
>>>>
>>>> rob
>>>>
>>>>
>>>> On Tue, Jan 13, 2009 at 4:37 PM, Jonathan Rochkind <rochkind at jhu.edu>
>>>> wrote:
>>>>
>>>>
>>>>  Okay, thanks. If there's anything I can do to help with this, let me
>>>>         
>>>>> know. Unfortuantely, I'm not a lawyer, and don't know the answer to this
>>>>> question myself, or I'd contribute an example right now!  But it is very
>>>>> important to me that we get this worked out asap--there is a need for it
>>>>> right now, people are wanting to release open data using the ODDC PDL
>>>>> right now.
>>>>>
>>>>> I do however have some familiarity with IP law, from several courses I
>>>>> took in school and a personal interest.  If there's anything I can do to
>>>>> help, let me know.
>>>>>
>>>>> Jonathan
>>>>>
>>>>> Jordan S Hatcher wrote:
>>>>>
>>>>>  On Jan 13, 2009, at 3:24 PM, Jonathan Rochkind wrote:
>>>>>           
>>>>>>  Thanks, that's helpful.
>>>>>>             
>>>>>>> So you think it's okay to link to that URI, as long as you do
>>>>>>> something _else_ to actually release your data. Can you supply a
>>>>>>> sample procedure you would use to make sure the data is actually
>>>>>>> released properly, and then use that consistent URI as an identifier
>>>>>>> in the records themselves?  Or does that approach legal advice you
>>>>>>> can't give us?
>>>>>>>
>>>>>>>  It doesn't -- basically you just have to associate the PDDL with what
>>>>>>>               
>>>>>> you are dedicating and be clear about your dedication and what it
>>>>>> contains.  Examples are something that should go into the licence /
>>>>>> project development -- and will.  At the moment it is a matter of time
>>>>>> and number of contributors to the project.
>>>>>>
>>>>>>
>>>>>>
>>>>>>  Any ideas as to timeline for the "transition to hosting at OKF",
>>>>>>             
>>>>>>> which I hadn't previously heard about? (Not even sure what OKF is).
>>>>>>>
>>>>>>>  Open Knowledge Foundation, which was hinted at on the blog but not
>>>>>>>               
>>>>>> formally announced I believe on the ODC site (though I did send an
>>>>>> email around about it):
>>>>>>
>>>>>> <
>>>>>>
>>>>>>
>>>>>>             
>>>>> http://www.opendatacommons.org/2008/11/26/updates-on-the-open-data-commons-project/
>>>>>
>>>>>
>>>>>           
>>>>>> <http://www.okfn.org/>
>>>>>>
>>>>>> And should be added to:
>>>>>>
>>>>>> <http://www.okfn.org/projects>
>>>>>>
>>>>>> (Rufus or Jonathan Gray?)
>>>>>>
>>>>>>
>>>>>>  Jonathan
>>>>>>             
>>>>>>> Jordan S Hatcher wrote:
>>>>>>>
>>>>>>>  Hi Jonathan,
>>>>>>>               
>>>>>>>> Rob and Talis people are (I htink) on this list but since it is so
>>>>>>>> low traffic I will have to check.
>>>>>>>>
>>>>>>>> There is a difference between using a link to the PDDL as the only
>>>>>>>> way of licensing your work and linking to the URI to make that
>>>>>>>> choice clearer to machines.
>>>>>>>>
>>>>>>>> The only thing that I think should NOT happen is that someone links
>>>>>>>> to the PDDL using licence metadata and does NOT do anything else to
>>>>>>>> license the data.  That is an incorrect way of using it IMO for the
>>>>>>>> reasons I mentioned.
>>>>>>>>
>>>>>>>> The current URI should remain stable and could be used in this
>>>>>>>> manner.  I leave it up to volunteers such as yourself and others
>>>>>>>> more technically inclined to take up this section of the work once
>>>>>>>> the site has fully transitioned to hosting at OKF.
>>>>>>>>
>>>>>>>> Thanks
>>>>>>>>
>>>>>>>> ~Jordan
>>>>>>>>
>>>>>>>>
>>>>>>>> On Dec 30, 2008, at 2:52 PM, Jonathan Rochkind wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>  I note that the current CC public domain dedication makes you go
>>>>>>>>                 
>>>>>>>>> through an interactive process to register your public domain
>>>>>>>>> dedication, but once are done, it does give you a standard uniform
>>>>>>>>> hosted-by-CC URI to link to:
>>>>>>>>> http://creativecommons.org/licenses/publicdomain/
>>>>>>>>>
>>>>>>>>> I'd assume CC0 will do similar. There is a LOT of utility to having
>>>>>>>>> this standard uniform URI.
>>>>>>>>>
>>>>>>>>> Rob Styles, are you on this list? Or anyone from Talis? Any ideas?
>>>>>>>>> It's really a handicap to the use of the ODC PDDL to not have this
>>>>>>>>> uniform URI that can be used.
>>>>>>>>>
>>>>>>>>> Jonathan
>>>>>>>>>
>>>>>>>>> Jonathan Rochkind wrote:
>>>>>>>>>
>>>>>>>>>  Hi Jordan, thanks.
>>>>>>>>>                   
>>>>>>>>>> Using my own local URI doesn't quite work. The issue is having a
>>>>>>>>>> _standard_ machine recognizable URL on the internet in general.
>>>>>>>>>>
>>>>>>>>>> To compare to CC, a standard machine recognizable URL for CC
>>>>>>>>>> licenses means that Google can spider the net and know what things
>>>>>>>>>> are CC licensed. Anything that has a <link rel="license"
>>>>>>>>>> href="http://creativecommons.org/licenses/by/3.0/"> has a CC-BY
>>>>>>>>>> license, and their software can automatically include it in
>>>>>>>>>> Google's special index of freely licensed stuff.  If every
>>>>>>>>>> licensor was using their own local URI for their CC-BY license,
>>>>>>>>>> that wouldn't work.
>>>>>>>>>>
>>>>>>>>>> There is a lot of utility to this with the ODC-PDDL too,
>>>>>>>>>> especially since our data is often aggregated in giant databases.
>>>>>>>>>> I might want to collect a subset of "all of this bibliographic
>>>>>>>>>> data I have access to" limited to ODC-PDDL stuff.  The most
>>>>>>>>>> straightforward way to do this is if all of the data has a
>>>>>>>>>> _standard_ URI signifying ODC-PDDL in a standard place.
>>>>>>>>>>
>>>>>>>>>> Does this make sense?  I understand where you're coming from
>>>>>>>>>> too--putting your data in the public domain is indeed a big deal.
>>>>>>>>>> But your approach seems potentially incompatible with that
>>>>>>>>>> important purpose.
>>>>>>>>>>
>>>>>>>>>> I would assume that once CC0 is finished, there will be a standard
>>>>>>>>>> URI for that, like there is for all CC licenses, that people can
>>>>>>>>>> link to?  Even though CC0 too will be a public domain dedication?
>>>>>>>>>>
>>>>>>>>>> Or wait, now I realize I have another question, there may be
>>>>>>>>>> bigger problems. The ODC PDDL is clearly intended to apply to an
>>>>>>>>>> entire database. What if I have a database where _some_ of the
>>>>>>>>>> records were imported from someone who applied the ODC PDDL to
>>>>>>>>>> them, but others of the records were imported from someone who did
>>>>>>>>>> not. Is this a problem?
>>>>>>>>>>
>>>>>>>>>> Jonathan
>>>>>>>>>>
>>>>>>>>>> Jordan S Hatcher wrote:
>>>>>>>>>>
>>>>>>>>>>  Hi Jonathan,
>>>>>>>>>>                     
>>>>>>>>>>> Apologies for not getting back to you sooner.  This is a very low
>>>>>>>>>>> population list and traffic list, and at the moment I'm the main
>>>>>>>>>>> driving force for work on Open Data Commons (and I'm very busy!).
>>>>>>>>>>>
>>>>>>>>>>> The short answer is that dedicating your work to the public
>>>>>>>>>>> domain is a pretty big deal, especially in comparison to just
>>>>>>>>>>> licensing it.  That's why as a project I decided that users
>>>>>>>>>>> should use a copy of the PDDL directly on what they do.  Linking
>>>>>>>>>>> to a document somewhere else on the net (such as CC does) means
>>>>>>>>>>> that:
>>>>>>>>>>>
>>>>>>>>>>> -- you (the user/linker of the PDDL) are less likely to read the
>>>>>>>>>>> document
>>>>>>>>>>> -- your users (those receiving databases/data under the PDDL) are
>>>>>>>>>>> less likely to read the document
>>>>>>>>>>>
>>>>>>>>>>> As a legal matter, it is clearer that you have an intent to
>>>>>>>>>>> dedicate the work to the public domain when you use the text
>>>>>>>>>>> itself on something, which means that it is less likely that you
>>>>>>>>>>> could turn around and try to take something back out of the
>>>>>>>>>>> public domain (by arguing that you didn't know what you were
>>>>>>>>>>> doing). Also, there is likely stricter legal requirements for
>>>>>>>>>>> public domain dedications than for licences, and using the text
>>>>>>>>>>> directly can help meet those requirements.
>>>>>>>>>>>
>>>>>>>>>>> As a practical matter at this point you can:
>>>>>>>>>>>
>>>>>>>>>>> -- make your own URI with the text
>>>>>>>>>>> -- copy the PDDL into the database as the disclaimer intends.
>>>>>>>>>>>
>>>>>>>>>>> You could get in contact with EDINA at <http://edina.ac.uk/> to
>>>>>>>>>>> see how they use the PDDL for their project.
>>>>>>>>>>>
>>>>>>>>>>> Hope this helps.  I'm catching up with open data work this week.
>>>>>>>>>>> Also as the project progresses over the next few months, we'll be
>>>>>>>>>>> addressing things such as this.  If you'd like to be a part of
>>>>>>>>>>> helping out with the project, please let me know.
>>>>>>>>>>>
>>>>>>>>>>> Thanks
>>>>>>>>>>>
>>>>>>>>>>> ~Jordan
>>>>>>>>>>>
>>>>>>>>>>> On 29 Dec 2008, at 23:12, Jonathan Rochkind wrote:
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>  Um, so a month ago I said the below. Is anyone else on this list?
>>>>>>>>>>>                       
>>>>>>>>>>>> Anyone have any idea why the ODC PDDL web page says "Do not link
>>>>>>>>>>>> to this document as a means of using it for your content...
>>>>>>>>>>>> please host it on your own site or apply it directly to the
>>>>>>>>>>>> content you wish to be covered," what the rationale was for that?
>>>>>>>>>>>>
>>>>>>>>>>>> Knowing why that's there would be important for figuring out how
>>>>>>>>>>>> to work around it without making some kind of mistake. It's
>>>>>>>>>>>> important to me that there be some URI to identify ODC PDDL
>>>>>>>>>>>> data. Otherwise, it's of less use.
>>>>>>>>>>>>
>>>>>>>>>>>> Should I just use CC0 instead, which has such a URI?
>>>>>>>>>>>>
>>>>>>>>>>>> Jonathan
>>>>>>>>>>>>
>>>>>>>>>>>> Jonathan Rochkind wrote:
>>>>>>>>>>>>
>>>>>>>>>>>>  Hi. I have been thinking about how to label data records that
>>>>>>>>>>>>                         
>>>>>>>>>>>>> have been
>>>>>>>>>>>>> licensed/dedicated with the PDDL in a machine recognizable way.
>>>>>>>>>>>>>
>>>>>>>>>>>>> For instance, Creative Commons supports/reccommends, in an HTML
>>>>>>>>>>>>> document
>>>>>>>>>>>>> that is licensed under a CC license, including this HTML code:
>>>>>>>>>>>>>
>>>>>>>>>>>>> <a rel="license"
>>>>>>>>>>>>> href="http://creativecommons.org/licenses/by/3.0/us/">
>>>>>>>>>>>>> ... </a>
>>>>>>>>>>>>>
>>>>>>>>>>>>> The key thing there is that the URI in the href there is a
>>>>>>>>>>>>> persistent
>>>>>>>>>>>>> identifier for a particular license. In HTML, you can use that
>>>>>>>>>>>>> <a
>>>>>>>>>>>>> rel="license"> thing to say so, but in a non-HTML format, you
>>>>>>>>>>>>> could
>>>>>>>>>>>>> still find some other analagous appropriate way to advertise
>>>>>>>>>>>>> the license
>>>>>>>>>>>>> in a machine-readable way, and you'd use that same URI, and
>>>>>>>>>>>>> software can
>>>>>>>>>>>>> recognize that URI as identifying a particular license.
>>>>>>>>>>>>>
>>>>>>>>>>>>> Google, for instance, uses that to support, in their 'advanced
>>>>>>>>>>>>> search',
>>>>>>>>>>>>> limiting search results to only CC licensed materials.
>>>>>>>>>>>>>
>>>>>>>>>>>>> It would be useful to be able to have an analagous
>>>>>>>>>>>>> machine-readable
>>>>>>>>>>>>> notice of the PDDL in our data.  If I am investigating a giant
>>>>>>>>>>>>> corpus of
>>>>>>>>>>>>> data, maybe that I have locally or maybe even out on the web,
>>>>>>>>>>>>> some parts
>>>>>>>>>>>>> of records might be under the/a PDDL, and it would be useful
>>>>>>>>>>>>> for obvious
>>>>>>>>>>>>> reasons to filter out those parts.
>>>>>>>>>>>>>
>>>>>>>>>>>>> The obviously analagous thing to do is to use the ODC PDDL URL
>>>>>>>>>>>>> as an
>>>>>>>>>>>>> identifier, like CC does. But, the web page for the ODC PDDL
>>>>>>>>>>>>> seems to
>>>>>>>>>>>>> discourage this:
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>>                           
>>>>> http://www.opendatacommons.org/odc-public-domain-dedication-and-licence/
>>>>>
>>>>>  "Do not link to this document as a means of using it for your
>>>>>           
>>>>>>>>>>>>> content.
>>>>>>>>>>>>> If you choose, after consulting with an appropriate legal
>>>>>>>>>>>>> professional,
>>>>>>>>>>>>> to use this document, please host it on your own site or apply
>>>>>>>>>>>>> it
>>>>>>>>>>>>> directly to the content you wish to be covered."
>>>>>>>>>>>>>
>>>>>>>>>>>>> Now, embedding that URI in my records to indicate that it's
>>>>>>>>>>>>> licensed/dedicated under the PDDL would seem to be prohibited.
>>>>>>>>>>>>>
>>>>>>>>>>>>> I am curious as to why, what the motivations were for this. Can
>>>>>>>>>>>>> anyone
>>>>>>>>>>>>> answer that?  That might help us figure out a solution.
>>>>>>>>>>>>>
>>>>>>>>>>>>> Which could be hosting the PDDL at some other location that can
>>>>>>>>>>>>> be
>>>>>>>>>>>>> linked to. Talis? Maybe CC would do it?  Or, theoretically, it
>>>>>>>>>>>>> could be
>>>>>>>>>>>>> establishing an identifier that doesn't actually resolve to
>>>>>>>>>>>>> license/dedication text, there's no reason the identifier
>>>>>>>>>>>>> _needs_ to
>>>>>>>>>>>>> resolve to text in order to function as a machine-recognizable
>>>>>>>>>>>>> label.
>>>>>>>>>>>>> But it's awfully convenient when it does, in part because it
>>>>>>>>>>>>> allows you
>>>>>>>>>>>>> figure out what it actually means, what the text of the
>>>>>>>>>>>>> license/dedication that applies to the data so labelled is.
>>>>>>>>>>>>>
>>>>>>>>>>>>> Any feedback here?
>>>>>>>>>>>>>
>>>>>>>>>>>>> Jonathan
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>>  --
>>>>>>>>>>>>>                           
>>>>>>>>>>>> Jonathan Rochkind
>>>>>>>>>>>> Digital Services Software Engineer
>>>>>>>>>>>> The Sheridan Libraries
>>>>>>>>>>>> Johns Hopkins University
>>>>>>>>>>>> 410.516.8886 rochkind (at) jhu.edu
>>>>>>>>>>>> _______________________________________________
>>>>>>>>>>>> TCL-Discuss mailing list
>>>>>>>>>>>> TCL-Discuss at lists.opencontentlawyer.com
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>                         
>>>>> http://lists.opencontentlawyer.com/listinfo.cgi/tcl-discuss-opencontentlawyer.com
>>>>>
>>>>>  _____
>>>>>           
>>>>>>>>>>> Mr. Jordan S. Hatcher
>>>>>>>>>>> Head of Research
>>>>>>>>>>> ipVA
>>>>>>>>>>>
>>>>>>>>>>> e: jordan.hatcher at ipvalueadded.com
>>>>>>>>>>> m: +44 (0)7804 909 466
>>>>>>>>>>> NEW blog: <http://www.tangible-ip.com>
>>>>>>>>>>> skype: jshatcher_ipva
>>>>>>>>>>>
>>>>>>>>>>> Medius House | 2 Sheraton Street | London W1F 8BH IP Value Added
>>>>>>>>>>> Ltd. | Registered 05601817 in England & Wales
>>>>>>>>>>> =====
>>>>>>>>>>> CONFIDENTIALITY NOTICE.  This message may be privileged and
>>>>>>>>>>> confidential and is solely for the use of the intended recipient.
>>>>>>>>>>> Please notify the sender if you have received this email in error.
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>  --
>>>>>>>>>>>                       
>>>>>>>>> Jonathan Rochkind
>>>>>>>>> Digital Services Software Engineer
>>>>>>>>> The Sheridan Libraries
>>>>>>>>> Johns Hopkins University
>>>>>>>>> 410.516.8886 rochkind (at) jhu.edu
>>>>>>>>> _______________________________________________
>>>>>>>>> TCL-Discuss mailing list
>>>>>>>>> TCL-Discuss at lists.opencontentlawyer.com
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                   
>>>>> http://lists.opencontentlawyer.com/listinfo.cgi/tcl-discuss-opencontentlawyer.com
>>>>>
>>>>>  --
>>>>>           
>>>>>>> Jonathan Rochkind
>>>>>>> Digital Services Software Engineer
>>>>>>> The Sheridan Libraries
>>>>>>> Johns Hopkins University
>>>>>>> 410.516.8886 rochkind (at) jhu.edu
>>>>>>>
>>>>>>>  --
>>>>>>>               
>>>>> Jonathan Rochkind
>>>>> Digital Services Software Engineer
>>>>> The Sheridan Libraries
>>>>> Johns Hopkins University
>>>>> 410.516.8886
>>>>> rochkind (at) jhu.edu
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> odc-discuss mailing list
>>>>> odc-discuss at lists.okfn.org
>>>>> http://lists.okfn.org/cgi-bin/mailman/listinfo/odc-discuss
>>>>>
>>>>>
>>>>>
>>>>>           
>>>>         
>>> --
>>> Jonathan Rochkind
>>> Digital Services Software Engineer
>>> The Sheridan Libraries
>>> Johns Hopkins University
>>> 410.516.8886
>>> rochkind (at) jhu.edu
>>>
>>> _______________________________________________
>>> odc-discuss mailing list
>>> odc-discuss at lists.okfn.org
>>> http://lists.okfn.org/cgi-bin/mailman/listinfo/odc-discuss
>>>
>>>       
>>     
>
>   

-- 
Jonathan Rochkind
Digital Services Software Engineer
The Sheridan Libraries
Johns Hopkins University
410.516.8886 
rochkind (at) jhu.edu




More information about the odc-discuss mailing list