- From: Antoine Isaac <aisaac@few.vu.nl>
- Date: Thu, 18 Sep 2014 21:53:18 +0200
- To: "public-dwbp-wg@w3.org" <public-dwbp-wg@w3.org>
Hi Deirdre,
Thanks a lot. The table view really makes it clearer!
I do not remember that we had agreed that metadata may not be open. But perhaps I'm unconsciously blanking out a good argument that someone made, because I'm biased ;-)
Otherwise this looks very good.
As said in another mail, I would just prefer if the requirement on metadata was defined in a way that hints at the different types of metadata - not something formal, just a hint...
Cheers,
Antoine
On 9/18/14 6:03 PM, Lee, Deirdre wrote:
> Hi,
> Yes, this has been addressed in the latest version of the UCR available here http://htmlpreview.github.io/?https://github.com/w3c/dwbp/blob/gh-pages/usecasesv1.html
>
> The following requirements have been removed:
> R-MetadataInteroperable
> R-LicenseInteroperable
>
> And the requirements we agreed to are in the attached matrix.
>
> Cheers,
> Deirdre
>
> Deirdre Lee, Research Associate, eGovernment Group
> Insight Center for Data Analytics, NUI Galway, Ireland
> Twitter: @deirdrelee, Skype: deirdrelee
> Linkedin: ie.linkedin.com/in/leedeirdre/
>
> ________________________________________
> From: Antoine Isaac [aisaac@few.vu.nl]
> Sent: 16 September 2014 14:17
> To: public-dwbp-wg@w3.org
> Subject: Re: Does standardisation assume interoperability?
>
> Hi,
>
> I think we had agreed last time on a global solution that would roughly halfs the number of requirements, while making them a nice system. Hopefully Deirdre will send this around soon :-)
>
> Antoine
>
> On 9/16/14 2:56 PM, Mark Harrison wrote:
>> Hi Phil,
>>
>> I think we discussed R-MetadataInteroperable a couple of weeks ago and favoured replacing it with R-MetadataMachineRead and R-MetadataStandardized.
>>
>> I remember that I commented on that call (within the last month) to say that we could not ensure Interoperability - that was a goal that was enabled through the availability of standardised machine-readable metadata. I think that was captured in the IRC transcript at the time, so it should be fairly easy for you to find that discussion and try to close that one on this Friday's call.
>>
>> Personally, I agree with replacing R-LicenseInteroperable with R-LicenceAvailable, since I think that is as much as we can recommend at this time, for the same reasons. Note that in British English, the noun 'licence' is spelled with '…nce', whereas the verb 'to license' is spelled with '…nee' (just like advice vs advise, etc.). R-LicenceAvailable (or R-LicenceInteroperable) feels like we're talking about a noun ('licence') rather than a verb ('to license').
>>
>> - Mark (the pedant)
>>
>>
>>
>>
>> On 16 Sep 2014, at 12:22, Phil Archer <phila@w3.org>
>> wrote:
>>
>>> Looking at the tracker, I don't think was resolved?
>>>
>>> Others have given ample evidence that standards are not always interoperable. Personally, I favour removing
>>>
>>> • R-MetadataInteroperable
>>> • R-LicenseInteroperable
>>>
>>> And offer these definitions (which build on the existing ones in the UCR)
>>>
>>> • R-MetadataMachineRead
>>>
>>> Metadata should be machine-readable, i.e. structured and available primarily for consumption by machines rather than in natural language text.
>>>
>>> • R-MetadataStandardized
>>>
>>> Metadata should be standardized, i.e. follow a recognized approach that maximizes usefulness and interoperability.
>>>
>>> In other words, you've got more chance of matching up X and Y if X and Y both follow standards then if they don't, but there's no guarantee that they will match 1 - 1.
>>>
>>> On licenses I'm more hesitant as I'm wary of the WG committing to more than it can deliver. This time next week I may know whether a new project is going ahead that should help to increase the capacity of the group but for now, I rather like the existing definition which says:
>>>
>>> R-LicenseAvailable
>>>
>>> Data should be associated with a license. License is a type of metadata, so all metadata requirements also apply here.
>>>
>>> (I'd reword that slightly to: Data should be associated with a license. A license is a type of metadata so the requirements concerning machine readability and standardization of metadata also apply here.)
>>>
>>> Phil.
>>>
>>>
>>>
>>> On 29/08/2014 11:38, Lee, Deirdre wrote:
>>>> Hi all,
>>>>
>>>> In today's call I would like if possible to make a decision on this...
>>>>
>>>> Should we keep the following requirements:
>>>>
>>>> •R-MetadataInteroperable
>>>> •R-LicenseInteroperable
>>>>
>>>> In addition to the following requirements are defined:
>>>>
>>>> •R-MetadataMachineRead
>>>> •R-MetadataStandardized
>>>>
>>>> •R-LicenseMachineRead
>>>> •R-LicenseStandardized
>>>>
>>>> Discussion at https://www.w3.org/2013/dwbp/track/issues/23
>>>>
>>>>
>>>> -----Original Message-----
>>>> From: Antoine Isaac [mailto:aisaac@few.vu.nl]
>>>> Sent: 22 August 2014 13:05
>>>> To: public-dwbp-wg@w3.org
>>>> Subject: Re: Does standardisation assume interoperability?
>>>>
>>>> Hi,
>>>>
>>>> I'm not sure that discussing too much interoperability will help us get a quick solution. In fact in what Deirdre put below, item (c) implies that (a) and (b) are not necessary (they're just 'standard-focused' rewriting of (c) )
>>>>
>>>> I don't understand what Joao Paulo said. He proposed to get rid of the the *Standardized requirements but he suggests to add them back, in practice.
>>>>
>>>> In the end I would just keep
>>>> R-MetadataStandardized, making the point that standard should address the model, documentation and interoperability (as it should, see the intro at http://en.wikipedia.org/wiki/Standardization)
>>>> R-MetadataMachineReadable, assuming this includes format concerns (which with the first requirement gives you standardized formats).
>>>>
>>>> Best,
>>>>
>>>> Antoine
>>>>
>>>> On 8/11/14 4:40 PM, Lee, Deirdre wrote:
>>>>> Thanks for your input Giancarlo and Laufer. Is it fair to summarise as follows:
>>>>>
>>>>> Two datasets are interoperable if:
>>>>>
>>>>> a.They are modelled according to the same standard.
>>>>>
>>>>> b.They are modelled according to standards that are interoperable
>>>>>
>>>>> c.They are modelled according to the same formal representation (not necessarily standardised)
>>>>>
>>>>> (b) corresponds to a case where we know how the meaning of the elements in each standard relate to the elements in the other standard.
>>>>>
>>>>> (c) corresponds to the case where you have used a reference ontology to facilitate the interoperability of ECG data.
>>>>>
>>>>> -----------------------------------------------------------------------
>>>>>
>>>>> Deirdre Lee
>>>>>
>>>>> Research Associate
>>>>>
>>>>> eGovernment Domain (DEG)
>>>>>
>>>>> Insight-NUIG
>>>>>
>>>>> IDA BusinessPark, Lower Dangan,
>>>>>
>>>>> Galway, Ireland
>>>>>
>>>>> deirdre.lee@deri.org <mailto:deirdre.lee@deri.org>
>>>>> skype: deirdrelee
>>>>>
>>>>> twitter: @deirdrelee
>>>>>
>>>>> linkedin: ie.linkedin.com/in/leedeirdre/ <http://ie.linkedin.com/in/leedeirdre/>
>>>>>
>>>>> -----------------------------------------------------------------------
>>>>>
>>>>> *From:*Giancarlo Guizzardi [mailto:gguizzardi@gmail.com]
>>>>> *Sent:* 11 August 2014 09:00
>>>>> *To:* Laufer
>>>>> *Cc:* Lee, Deirdre; Data on the Web Best Practices Working Group
>>>>> *Subject:* Re: Does standardisation assume interoperability?
>>>>>
>>>>> Folks,
>>>>>
>>>>> Regarding the two questions:
>>>>>
>>>>> (a) Does standardization entail interoperability?
>>>>>
>>>>> and (b) Can something be interoperable without being standardized?
>>>>>
>>>>> Let us first settled what we mean by interoperability.
>>>>>
>>>>> I will start with the following working definition of semantic interoperability:
>>>>>
>>>>> A model X is semantically interoperable with model Y if we know
>>>>>
>>>>> how the meaning of the elements in X relate to the meaning of elements in Y.
>>>>>
>>>>> If we take "meaning" in the sense of referential semantics, this means that
>>>>>
>>>>> we can always relate in the correct manner
>>>>>
>>>>> the referents of the model elements in X with the referents
>>>>>
>>>>> of the model elements in Y.
>>>>>
>>>>> With that in mind (and answering (b)), we can achieve interoperability whenever we are able
>>>>>
>>>>> to establish and fully understand the relation between the referents of X and Y.
>>>>>
>>>>> Good standards certainly facilitate that.
>>>>>
>>>>> However, if we put the question as "standardization ENTAILS interoperability" (question a),
>>>>>
>>>>> the answer is clearly no. To put it simply, this is because (among other reasons, including non-technical ones...)
>>>>>
>>>>> there are bad standards. Bad standards in the sense that they are not sufficiently expressive and clear in helping
>>>>>
>>>>> users to express their world views in terms of the standard.
>>>>>
>>>>> There are many examples of domains with multiple standards
>>>>>
>>>>> that it is far from obvious how to relate the meaning of things
>>>>>
>>>>> in standard X and with the meaning of things in standard Y.
>>>>>
>>>>> This is even the case in the so-called hard science domains.
>>>>>
>>>>> To cite on example in heart Electrophysiogy.
>>>>>
>>>>> In the following paper
>>>>>
>>>>> http://www.j-biomed-inform.com/article/S1532-0464(10)00118-8/pdf
>>>>>
>>>>> we show the difficulties in relating multiple existing ECG standards.
>>>>>
>>>>> The paper is actually both an example of lack interoperability with
>>>>>
>>>>> the presence of multiple standards as well as an example of
>>>>>
>>>>> interoperability achieved with a reference ontology that is
>>>>>
>>>>> not a standard, i.e., "interoperability without a standard".
>>>>>
>>>>> Best,
>>>>>
>>>>> Giancarlo
>>>>>
>>>>> On Sat, Aug 9, 2014 at 3:04 AM, Laufer <laufer@globo.com <mailto:laufer@globo.com>> wrote:
>>>>>
>>>>> Hi,
>>>>>
>>>>> I don´t know if the following example makes sense to this discussion. But it comes to my mind.
>>>>>
>>>>> Analog television has some different standards as, for example, PAL and NTSC, and they are not interoperable.
>>>>>
>>>>> I don´t know if, in this case, the term should be compatible.
>>>>>
>>>>> Best,
>>>>> Laufer
>>>>>
>>>>> 2014-08-08 11:43 GMT-03:00 Lee, Deirdre <Deirdre.Lee@deri.org <mailto:Deirdre.Lee@deri.org>>:
>>>>>
>>>>> Hi,
>>>>>
>>>>> We had a good call today, addressing many of the raised issues for UCR. One issue that we felt warranted more discussion was /ISSUE-23: Review definition of interoperability/ https://www.w3.org/2013/dwbp/track/issues/23
>>>>>
>>>>> The following questions were raised: Does standardisation imply interoperability? Are there cases where data uses standards, but is not interoperable? Can interoperability be achieved independent of standardisation? In DWBP, should we have a specific definition of interoperability?
>>>>>
>>>>> In the UCR, there are the following two requirements:
>>>>>
>>>>> ·R-MetadataInteroperable
>>>>>
>>>>> ·R-LicenseInteroperable
>>>>>
>>>>> And also
>>>>>
>>>>> ·R-MetadataMachineRead
>>>>>
>>>>> ·R-MetadataStandardized
>>>>>
>>>>> ·R-LicenseMachineRead
>>>>>
>>>>> ·R-LicenseStandardized
>>>>>
>>>>> Possible resolutions for the UCR could be:
>>>>>
>>>>> a.Remove R–MetadataInteroperable and R-LicenseInteroperable from UCR because they’re redundant
>>>>>
>>>>> b.Improve description of R-MetadataStandardized and R-LicenseStandardized
>>>>>
>>>>> c.Other?
>>>>>
>>>>> Cheers,
>>>>>
>>>>> Deirdre
>>>>>
>>>>> -----------------------------------------------------------------------
>>>>>
>>>>> Deirdre Lee
>>>>>
>>>>> Research Associate
>>>>>
>>>>> eGovernment Domain (DEG)
>>>>>
>>>>> Insight-NUIG
>>>>>
>>>>> IDA BusinessPark, Lower Dangan,
>>>>>
>>>>> Galway, Ireland
>>>>>
>>>>> deirdre.lee@deri.org <mailto:deirdre.lee@deri.org>
>>>>> skype: deirdrelee
>>>>>
>>>>> twitter: @deirdrelee
>>>>>
>>>>> linkedin: ie.linkedin.com/in/leedeirdre/ <http://ie.linkedin.com/in/leedeirdre/>
>>>>>
>>>>> -----------------------------------------------------------------------
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> . . . .. . .
>>>>> . . . ..
>>>>> . .. .
>>>>>
>>>>
>>>
>>> --
>>>
>>>
>>> Phil Archer
>>> W3C Data Activity Lead
>>> http://www.w3.org/2013/data/
>>>
>>> http://philarcher.org
>>> +44 (0)7887 767755
>>> @philarcher1
>>>
>>
>> CONFIDENTIALITY / DISCLAIMER: The contents of this e-mail are confidential and are not to be regarded as a contractual offer or acceptance from GS1 (registered in Belgium).
>> If you are not the addressee, or if this has been copied or sent to you in error, you must not use data herein for any purpose, you must delete it, and should inform the sender.
>> GS1 disclaims liability for accuracy or completeness, and opinions expressed are those of the author alone.
>> GS1 may monitor communications.
>> Third party rights acknowledged.
>> (c) 2012.
>> </a>
>>
>>
>>
>
Received on Thursday, 18 September 2014 19:53:48 UTC