AW: LLD Web Services

Hi Jeff,

Thanks for adding and tweaking our text. However, the main idea of http://www.w3.org/2005/Incubator/lld/wiki/Web_services_on_LLD was that simple Web Services are important for the spreading the re-use of LLD, for two main reasons:


SPARQL backends and production environments

There are good reasons why organizations generally do not offer SQL access to their databases: In a general query language such as SQL or SPARQL, it is very easy to construct queries that force a server down by sheer load. This happens even if we assume that all users are good-willing - simple mistakes or missing knowledge about the physical data access paths (indexes defined) makes it very likely. My feeling is that the counter strategies implemented in SPARQL servers (cutting results after a certain number of triples and/or after a certain amount of CPU usage and/or restricting the types of queries) are far from mature and more like makeshifts. 

For a production implementation, the owners have to guarantee reliable access and a certain quality of service. So in my eyes powerful SPARQL interfaces are currently limited to labs environments, in order to discover useful access patterns and queries, or to drive other experimental implementations. The patterns discovered possibly could push the further evolution of production quality web services. So, in my point of view, SPARQL endpoints *are* useful and *should* be offered, but are at the same time very limited in scope. (Even for medium sized datasets such as authorities - as far as I know, neither DNB nor VIAF offer SPARQL access to their datasets).


Developer skills

As the authors of the Linked Data API (thanks, Richard, for the hint) put it: "Simple RESTful APIs are well supported and understood by a large community of web developers. Faced with Linked Data and SPARQL endpoints this community faces a steep learning curve before they are able to make use of the power provided by the underlying technologies. Put differently, SPARQL is a power tool whose sophistication is unnecessary for many users."

Our main goal should be to offer Linked Data URIs, to aid their proliferation and to demonstrate the enableing power of Linked Data - lowering the entry barrier by not requesting a complete new technologies skillset ("making it much less foreign and frightening", as Karen put it). 


I've tried to integrate these aspects into the wiki text. It would be great if we can agree so far - otherwise please feel free to tweak it again.

Cheers, Joachim




> -----Ursprüngliche Nachricht-----
> Von: public-lld-request@w3.org 
> [mailto:public-lld-request@w3.org] Im Auftrag von Young,Jeff (OR)
> Gesendet: Montag, 9. Mai 2011 18:29
> An: Ford, Kevin; public-lld
> Betreff: RE: LLD Web Services
> 
> I added Kevin and Joachim's text to the Draft Relevant 
> Technologies page and tweaked it a bit. I'm afraid I made it 
> less readable, but hopefully tied up a few loose ends in 
> exchange. If the changes are too radical, I can back them out. 
> 
> Suggestions and help for making the text readable again would 
> be very welcome. Also, a few questions/issues popped out that 
> could use some broader opinions. They appear in brackets in the text:
> 
> http://www.w3.org/2005/Incubator/lld/wiki/Draft_Relevant_Techn
> ologies#We
> b_Services_for_Library_Linked_Data
> 
> Jeff
> 
> > -----Original Message-----
> > From: public-lld-request@w3.org 
> [mailto:public-lld-request@w3.org] On 
> > Behalf Of Ford, Kevin
> > Sent: Thursday, May 05, 2011 9:43 AM
> > To: public-lld
> > Subject: LLD Web Services
> > 
> > Dear All,
> > 
> > Joachim contacted me and asked, based on discussion during 
> a telecon, 
> > if I could trim the Web Services text he and I authored for 
> inclusion 
> > into the final report.  After reviewing the minutes of that 
> telecon, I 
> > am operating under the assumption that it is to go in the Relevant 
> > Technologies section of the report (or, at least what appears to be
> the
> > current draft) [1].  I hope so: I've tried to tailor it to that 
> > section.  I see it going after "Linked Data front-ends to existing
> data
> > stores" and before "OWL and supporting tools".  I've halved 
> the text 
> > (at least).  I'm having a devil of a time signing in to the wiki 
> > currently, so I've pasted it below.  If someone wants to 
> paste it into 
> > the document, that would be great.
> > 
> > Warmly,
> > 
> > Kevin
> > 
> > [1]
> > 
> http://www.w3.org/2005/Incubator/lld/wiki/Draft_Relevant_Technologies
> > 
> > 
> > Web Services for LLD
> > 
> > Many LD implementations, for a variety of reasons, can not 
> or have not 
> > provided SPARQL endpoints (or bulk downloads).  Some LD
> implementations
> > might not use a triplestore in the back-end, which is seen as a
> natural
> > precursor for a SPARQL endpoint; for others, security or robustness 
> > considerations preclude such a feature in production use.  Not
> offering
> > these options can hinder further resource discovery.  
> Furthermore, it 
> > may also not be feasible to layer a Linked Data front-end on to an 
> > existing back-end.
> > 
> > Therefore LLD efforts should encourage the development of LD Web 
> > Services to facilitate greater access to the data offered by a LD 
> > Implementation.  Web Services can be offered in the absence of a
> SPARQL
> > endpoint or in conjunction with one.   Web Services should be fully
> > documented.
> > 
> > A few LD implementations have endeavored to implement Web 
> Services to 
> > enhance discovery and use of resources, often by providing some form
> of
> > an application programming interface (API).  Agrovoc and STW provide
> an
> > API to discover resources based on relationships in the data, among 
> > many more web services.  VIAF, LC's ID, and STW offer autosuggest 
> > services for resources, delivering JSON responses ready for
> consumption
> > in AJAX browser applications.  Agrovoc and STITCH/CATCH include
> support
> > for pure RDF responses.  Some services provide full-fledged 
> SOAP APIs, 
> > while others support a REST approach.
> > 
> > By focusing on method parameters and response formats to provide 
> > enhanced discovery, LD Web Services diminish, if not eliminate, the 
> > requirement that data be stored in a triplestore.  And, because web 
> > service APIs are common, web services can lower the barrier 
> to entry.
> > 
> > 
> 
> 
> 
>

Received on Tuesday, 10 May 2011 15:05:17 UTC