Hi Juan, All,
We are actually working on some solutions that can help people to
explore what is inside a “dataset” deployed in an endpoint, to have a
better and more exhaustive view about the information and structure of
the data.
The proposed SPARQL queries are excellent, however most of them are not
feasible in practice we are faced with timeouts and endpoint
limitations, especially with the one below (although it is quite
informative).
|SELECT ?type1 ?pred ?type2
WHERE {
?subj ?pred ?obj.
?subj a ?type1.
?obj a ?type2.
}
|
The question that you raised Juan is exactly one of the purpose of the
graph summary [1]. It provides information about the predicates,
classes, and how classes relate with each other.
In [2] is a SPARQL auto-completion tool which you can use to get
suggestions about predicates/classes in an endpoint by typing
|CTRL+SPACE|. You can have a look at [3] for more details on how to use it.
This tool uses the SPARQL endpoint to get theses recommendations.
However, a summary could just as well be used. So that in addition, you
would get better performance and useful information, since the summary
is smaller than the original data and contain info about the graph
structure.
We are working on an official specification draft that we would like to
submit to W3c.
Also, we are currently working on a *repository* similar to datahub for
providing summaries that would help end-user that have your problem, and
a solution for provider on how to provide these graph summaries
alongside their original datasets.
We think that “graph summaries” coupled with VoID and Service
Description information could help people to have a better and immediate
view about the data available for a better use of them, especially if we
really want to exploit the potentialities of Linked Open Data.
Just throwing this out there to get early feedbacks…
Cheers,
[1] http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6327436&tag=1
[2] http://scampi.github.io/gosparqled/
[3] https://github.com/scampi/gosparqled
On 22/01/15 14:09, Juan Sequeda wrote:
Assume you are given a URL for a SPARQL endpoint. You have no idea
what data is being exposed.
What do you do to explore that endpoint? What queries do you write?
Juan Sequeda
+1-575-SEQ-UEDA
www.juansequeda.com <http://www.juansequeda.com>
--
Stephane Campinas