PubChem Re: SPARQL lite?

[jerven Bolleman <je...@sib.swiss>]

Hi Steve,

Indeed the NCBI PubChem team is really worried about having a resilient 
service, at least that is what I recall of the conversation over sake.
But with the lower cost of hosting a sparql endpoint they may change 
their minds.

Also of interest is.
https://jcheminf.biomedcentral.com/articles/10.1186/s13321-021-00515-1

If you want a pubchem rdf sparql endpoint you should definitely ask them 
for it.

Regards,
Jerven

On 18/05/2021 20:59, Steve Vestal wrote:
> Re use cases, the OSLC folks say they want to enable light-weight 
> addition of OSLC support to existing servers of specialized data (they 
> mention that nothing prevents SPARQL if vendors want to). It would be my 
> own speculation, just a guess, that the US National Center for 
> Biotechnology Information does not want to host full SPARQL query 
> services for huge numbers of people.
> 
> The BGP comment is informative, thanks.  That would explain why these 
> other solutions are also fairly simple single-triple-pattern sorts of 
> things that can be encoded in the query part of a URL.
> 
> On 5/18/2021 1:36 PM, Andy Seaborne wrote:
>>
>>
>>> On Tue, May 18, 2021 at 3:59 PM Steve Vestal
>>> <st...@adventiumlabs.com> wrote:
>>>>
>>>> PubChemRDF, OSLC, and probably others define their own light-weight
>>>> query protocol, conceptually similar to the Jena Graph interface.  I 
>>>> was
>>>> curious if there exists, or there are moves afoot to propose, a 
>>>> standard
>>>> "SPARQL lite" or BGP query protocol, closer to the Jena Graph or
>>>> StageGenerator interfaces than full SPARQL?
>>>>
>> On 18/05/2021 15:06, Martynas Jusevičius wrote:
>>> Triple Pattern Fragments?
>>> https://linkeddatafragments.org/specification/triple-pattern-fragments/
>>>
>>
>> Hi Steve,
>>
>> What's the use cases behind those those light-weight query protocols?
>> There's a big difference between Graph.find and BGP here - Graph.find 
>> (and TPF) are low impact on the storage/.
>>
>> That said, I think the bulding block is "triple pattern + filter". The 
>> filter could test slots of the triple pattern maybe in a limited 
>> fashion e.g. range
>>
>> { ?s ?p ?o . FILTER (?o > 10 && ?o < 500) }
>>
>> not FILTER(?x/2 - FLOOR(?x/2) != 0) ## Test ?x is odd
>>
>> and if a range, it's easy to make the query string parameters.
>>
>>
>> BGPs can be server-side expensive.
>>
>>     Andy
> 

-- 

	*Jerven Tjalling Bolleman*
Principal Software Developer
*SIB | Swiss Institute of Bioinformatics*
1, rue Michel Servet - CH 1211 Geneva 4 - Switzerland
t +41 22 379 58 85
Jerven.Bolleman@sib.swiss - www.sib.swiss