On 03/12/2014 12:13, Daniel Vila Suero wrote:
Ah, so you can do it for Shakespeare, but not for me. :-) My HTML
(see above) has a link to the work Museum Documentation Systems, but
the Turtle doesn't.
:-) We will work to bring your links up to date! At least, we hope it
was easy to find yourself in datos.bne.es :-)
Yes, it was very easy. I was sad not to see Presenting XML (SAMS Net,
1997) on there, but that's hardly the fault of your LD implementation.
Also, there seems to be some confusion in the URL pattern: the
subject URL within the RDF is
http://datos.bne.es/resource/XX1000054, i.e. "resource" replaces
"autor".
We have kept the canonical URIs for things that we published in
previous versions: http://datos.bne.es/resource/ which correspond to
non-information resources and use the 303 mechanism to provide
different representations, in this case changing to a typed URI
pattern + the file extension, this was motivated to support nicer
URIs for the human-oriented content of the portal and its similar to
patterns like the one used in DBpedia where they do /resource/ -->
/page/ or /resource/ --> /data/ + the file extension for RDF
representations.
Having recently finished "Linked Data for Libraries, Archives and
Museums" (van Hooland and Verborgh), I am newly enthused about the
application of a proper REST approach. To conform to this, I think,
the URLs quoted when a resource is dereferenced should be compatible
with the URL itself. Otherwise, won't machine agents attempting to
use these URLs to navigate become confused?
My understanding is that an agent should only use the Location to
retrieve more data (a representation of a resource). Given a canonical
URI to name the entity (http://datos.bne.es/resource/XX1000054), an
agent can try to dereference it, asking for a concrete representation
that he understands (e.g., turtle) and what it gets back is a LOCATION
with a Turtle representation of the entity in question, that it's
described in terms of canonical URIs (e.g.,
http://datos.bne.es/resource/XX1000054 a frbr:Person; rdfs:label
"Richard"). The agent then can follow his nose using these URIs to
retrieve more data (in the RDF representations we only use canonical
URIs not Locations).
We have tried to follow existing best practices (ISA guidelines [1]
mention for example the pattern /id/ --> /doc/), but we are open to
suggestions from the community on how to improve content-negotiation
mechanisms because our goal is to make the data useful for application
developers.
OK, so the content negotiation for both variants works in the same way?
That is certainly what I am finding. So even if an agent starts off
from the "autor" variant of a URL, it will end up being able to follow
its nose using the "resource" variants, which it will find in the
machine-processible variant of its choice. Works for me. :-)
Richard
Best wishes
Daniel
[1]
https://joinup.ec.europa.eu/sites/default/files/D7.1.3%20-%20Study%20on%20persistent%20URIs_0.pdf
Richard
Thank you very much for your comments, they help us to improve the
service.
Daniel
Richard
On 03/12/2014 10:11, Daniel Vila Suero wrote:
The *National Library of Spain* (BNE) (www.bne.es
<http://www.bne.es>) and the *Ontology Engineering Group*
(www.oeg-upm.net <http://www.oeg-upm.net>) are glad to announce
the new *datos.bne.es <http://datos.bne.es>
(http://datos.bne.es)* Linked Data service (in Spanish).
This new service represents a milestone of the Linked Data project
started by the end of 2011 and that already published*Linked Open
Data under a Public Domain license* (Creative Commons CC0). We
have been working to improve many aspects of the service and would
like to share with you some *key features*:
*A new way to search, discover and explore*
*-------------------------------------------------------------*
The new (beta) portal exploits linked data *to create better
experiences for the user*. A graph with millions of new
connections allows the user to explore the collections
comprehensively and across three core entities: authors, works and
topics. The search engine (e.g., http://datos.bne.es/find?s=joyce
<http://datos.bne.es/find?s=joyce>) also uses this graph to
retrieve and rank entities, presenting relevant information to the
user and allowing for simple, easy-to-use faceting.
Besides, we continue to offer a public SPARQL endpoint
(http://datos.bne.es/sparql) for people to query and use the data
for their own applications, content negotiation, and we also
provide schema.org <http://schema.org> descriptions of authors and
works using JSON-LD.
*More data, more links*
*-------------------------------*
We have published the full catalogue comprising *more than 9
million records and around 150.000 digitalized materials* that
generate more than 140 million RDF triples. These linked data
resources describe and give access to authors, organizations,
topics, modern and ancient books, photographs, cartographic
materials, drawings, manuscripts, or printed and manuscript music.
We provide around *1.4 million sameAs links* and add links to new
datasets such as ISNI, data.bnf.fr <http://data.bnf.fr>,
id.loc.gov <http://id.loc.gov>, and geo.linkeddata.es
<http://geo.linkeddata.es>. More importantly, we have
significantly increased the internal links between authors,
bibliographic resources and digital materials.
*The BNE data model*
*-----------------------------*
The BNE vocabulary, inspired by the FRBR data model, reuses and
integrates several vocabularies such as IFLA FRBR, ISBD, or RDA,
among others. The vocabulary is available for both humans and
machines at http://datos.bne.es/def/, it is documented in English
and Spanish. We will soon provide alignments to the aforementioned
vocabularies.
*Help us to improve*
*---------------------------*
We would very much appreciate receiving feedback from the
community. If you have any ideas/comments on how to improve the
service, you encounter issues/problems, you want to collaborate,
etc. please get in touch. But first of all, we invite you to visit:
http://datos.bne.es
Thanks and our best wishes.
Daniel Vila Suero, Asunción Gómez Pérez, Ricardo Santos and Ana
Manchado, on behalf of the OEG and BNE teams.
--
*Richard Light*
--
*Richard Light*
--
*Richard Light*
