W3C home > Mailing lists > Public > public-lld@w3.org > July 2011

Re: Review of Relevant Technologies section

From: Antoine Isaac <aisaac@few.vu.nl>
Date: Thu, 07 Jul 2011 15:56:02 +0200
Message-ID: <4E15BAF2.2050606@few.vu.nl>
To: public-lld@w3.org
Hi Jon,

Don't take me wrong, my answer was not intended as a general conclusion to put in the report. That was in the specific context of answering Karen's question, on whether there were cases of LD that was not in RDF out already there.

And, yes, even if I wouldn't like it to appear in the report, I believe that the more of the LD methodology you would want to implement, the more useful something that *look like* RDF will be.
I'm not saying that it should be RDF, but just that it could be likened to RDF.
In particular, if you want to *link* entities within the data, it will be difficult to escape the notion of typed relation. If a connection is made in the raw data it will usually come with some form of semantics. If MARC records contain URIs, I'd expect it would be really easy to generate some RDF from it, no?

But well, otherwise I think I agree with you on the fundamental points that you make--I mean, much more fundamental than my devising on comparison between data models.

Antoine


> Antoine,
> re:
>
>     ...RDF is so closely matching the principles of linked data (data about things expressed using links) that it will be probably difficult to find something radically different.
>
>     Antoine
>
>
> I'm concerned that this is one area where the report draws an unsupportable conclusion.
>
> It's not the purpose of an XG group, as I understand it at least, to draw this kind of conclusion. For instance I personally think that it's an open question, worthy of serious investigation, of whether marc21 can be distributed as linked data in more or less its current form. What are the pros and cons? Is it technically feasible and what are the limitations? Do the potential benefits outweigh the fairly severe limitations? Is it useful as a transition to a more broadly useful technology?
>
> We collectively may think that's nuts, but that doesn't mean that under the broader definition of linked data it's not feasible or desirable or of benefit to the community. It certainly fits with TBL's "raw data now" message and the definition of Linked Data in the two most seminal Linked Open Data papers:
> http://tomheath.com/papers/bizer-heath-berners-lee-ijswis-linked-data.pdf
> http://www.w3.org/DesignIssues/LinkedData.html
>
> I think we should be far more careful than we have been so far about drawing conclusions about library data and how it could best be expressed, distributed, and aggregated in a linked data (not an exclusively RDF) environment. If our sole message to the 'Library Community' is that in order to participate in the 'Linked Data Community' they have to effectively abandon their long-held notions of cataloging and librarianship and radically change their entire data infrastructure, we run the serious risk of driving that community away.
>
> I believe that it's possible, and highly desirable, to present Linked Data as a methodology, a way of thinking about data distribution and resource description, rather than a specific technology, regardless of how appropriate that single technology is to the task. We may collectively view the data with some distaste, but the Library Community already has methods in place for distributed maintenance, publishing, even identification of data. Authority control, ISBN, ISSN, DOI, and many other methods of collective, shared and even (textually) linking data have been enthusiastically embraced by libraries for years. MARBI has made recent efforts to provide for the inclusion of URIs in marc21 records. How is any of this not Linked Data if the resources are identified by 'cool' URIs?
>
> We need not be radicals to promote a Linked Data agenda. We shouldn't let our enthusiasm for a particularly relevant technology blind us to the potential benefits to the Library Community of at least proposing a serious investigation into how that community can transition into a more open environment without radically altering or discarding the concepts with which they are most comfortable.
>
> Let's get the data out of the library silos first, let people outside the community "do interesting things" with it, and see what happens. Then lets talk about RDF, because very shortly (certainly within 20 years ;-) ) the benefits of Linked Open Data will be obvious (or not) and the Library Community will enthusiastically embrace it (or not).
> --
> Jon
>
Received on Thursday, 7 July 2011 13:55:16 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 7 July 2011 13:55:17 GMT