You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@jena.apache.org by Cong Wang <jo...@gmail.com> on 2014/03/09 23:24:41 UTC
Gsoc 2014 Apply
Hi everybody,
My name is Cong Wang, currently a Ph.D student at Wright State University
with Pascal Hitzler. I am very interested two ideas, "Generate JSON from
SPARQL directly" and "Define SPARQL commands in Jena rules". I have several
questions regarding the two topics.
For "Generate JSON from SPARQL directly",
Is that you want translate sparql query into json format, such that we can
use it to query json data? Since under W3C recommendation, there's way to
translate sparql results into json [1]. The only missing part is converting
sparql query with json query. But even if so, I'm still not sure the
significance. What's the motivation here?
For "Define SPARQL commands in Jena rules",
I remember there's a paper by Axel Polleres [2]. It's about translating
SPARQL into datalog. Although datalog is arbitrary variables and Jena rules
is only 3 variables, there should be a way to convert.
But I don't quite understand how it can "increase the expressiveness of
Jena". In principle, SPARQL is just indexing, and Jena rules actually have
already indexed the triples while grounding variables.
[1]. http://www.w3.org/TR/sparql11-results-json/
[2]. Axel Polleres: From SPARQL to rules (and back). WWW 2007
Best regards.
--
Cong Wang
Ph.D Candidate,
Kno.e.sis Center,
Wright State University.
Re: Gsoc 2014 Apply
Posted by Cong Wang <jo...@gmail.com>.
Hi Miguel,
I can understand it's very useful if we can integrate some rules into
SPARQL, i.e., enrich inference ability in SPARQL (that's SPIN). But
integrate SPARQL into Jena rules is unclear, That's my understanding.
Best.
On Sun, Mar 9, 2014 at 6:51 PM, Miguel Bento Alves <mb...@gmail.com>wrote:
> Hi Cong,
>
> In introduction of the paper that you referred [2]:
> "Still, SQL is often viewed to be more powerful in several respects² (I
> think that this idea is also valid for SPARQL in the context)
> "On the other hand, aggregates ... are missing in pure Datalog".
>
> Related to this point "increase the expressiveness of Jena², I think that
> is very useful when you can represent your knowledge either through rules
> or through SPARQL and commands. Read about the motivations of spin rules
> [3][4].
>
> [3] http://www.w3.org/Submission/2011/SUBM-spin-overview-20110222/
> [4] http://www.w3.org/Submission/2011/SUBM-spin-sparql-20110222/
>
> Miguel
>
>
>
> On 09/03/14 22:24, "Cong Wang" <jo...@gmail.com> wrote:
>
> >Hi everybody,
> >
> >My name is Cong Wang, currently a Ph.D student at Wright State University
> >with Pascal Hitzler. I am very interested two ideas, "Generate JSON from
> >SPARQL directly" and "Define SPARQL commands in Jena rules". I have
> >several
> >questions regarding the two topics.
> >
> >For "Generate JSON from SPARQL directly",
> >
> >Is that you want translate sparql query into json format, such that we can
> >use it to query json data? Since under W3C recommendation, there's way to
> >translate sparql results into json [1]. The only missing part is
> >converting
> >sparql query with json query. But even if so, I'm still not sure the
> >significance. What's the motivation here?
> >
> >For "Define SPARQL commands in Jena rules",
> >
> >I remember there's a paper by Axel Polleres [2]. It's about translating
> >SPARQL into datalog. Although datalog is arbitrary variables and Jena
> >rules
> >is only 3 variables, there should be a way to convert.
> >
> >But I don't quite understand how it can "increase the expressiveness of
> >Jena". In principle, SPARQL is just indexing, and Jena rules actually have
> >already indexed the triples while grounding variables.
> >
> >[1]. http://www.w3.org/TR/sparql11-results-json/
> >[2]. Axel Polleres: From SPARQL to rules (and back). WWW 2007
> >
> >
> >Best regards.
> >
> >--
> >Cong Wang
> >Ph.D Candidate,
> >Kno.e.sis Center,
> >Wright State University.
>
>
>
--
Cong Wang
Ph.D Candidate,
Kno.e.sis Center,
Wright State University.
Re: Gsoc 2014 Apply
Posted by Miguel Bento Alves <mb...@gmail.com>.
Hi Cong,
In introduction of the paper that you referred [2]:
"Still, SQL is often viewed to be more powerful in several respects² (I
think that this idea is also valid for SPARQL in the context)
"On the other hand, aggregates ... are missing in pure Datalog".
Related to this point "increase the expressiveness of Jena², I think that
is very useful when you can represent your knowledge either through rules
or through SPARQL and commands. Read about the motivations of spin rules
[3][4].
[3] http://www.w3.org/Submission/2011/SUBM-spin-overview-20110222/
[4] http://www.w3.org/Submission/2011/SUBM-spin-sparql-20110222/
Miguel
On 09/03/14 22:24, "Cong Wang" <jo...@gmail.com> wrote:
>Hi everybody,
>
>My name is Cong Wang, currently a Ph.D student at Wright State University
>with Pascal Hitzler. I am very interested two ideas, "Generate JSON from
>SPARQL directly" and "Define SPARQL commands in Jena rules". I have
>several
>questions regarding the two topics.
>
>For "Generate JSON from SPARQL directly",
>
>Is that you want translate sparql query into json format, such that we can
>use it to query json data? Since under W3C recommendation, there's way to
>translate sparql results into json [1]. The only missing part is
>converting
>sparql query with json query. But even if so, I'm still not sure the
>significance. What's the motivation here?
>
>For "Define SPARQL commands in Jena rules",
>
>I remember there's a paper by Axel Polleres [2]. It's about translating
>SPARQL into datalog. Although datalog is arbitrary variables and Jena
>rules
>is only 3 variables, there should be a way to convert.
>
>But I don't quite understand how it can "increase the expressiveness of
>Jena". In principle, SPARQL is just indexing, and Jena rules actually have
>already indexed the triples while grounding variables.
>
>[1]. http://www.w3.org/TR/sparql11-results-json/
>[2]. Axel Polleres: From SPARQL to rules (and back). WWW 2007
>
>
>Best regards.
>
>--
>Cong Wang
>Ph.D Candidate,
>Kno.e.sis Center,
>Wright State University.
Re: Gsoc 2014 Apply
Posted by Andy Seaborne <an...@apache.org>.
On 09/03/14 22:24, Cong Wang wrote:
> Hi everybody,
>
> My name is Cong Wang, currently a Ph.D student at Wright State University
> with Pascal Hitzler. I am very interested two ideas, "Generate JSON from
> SPARQL directly" and "Define SPARQL commands in Jena rules". I have several
> questions regarding the two topics.
>
> For "Generate JSON from SPARQL directly",
>
> Is that you want translate sparql query into json format, such that we can
> use it to query json data? Since under W3C recommendation, there's way to
> translate sparql results into json [1]. The only missing part is converting
> sparql query with json query. But even if so, I'm still not sure the
> significance. What's the motivation here?
As things currently stand, there isn't a volunteer mentor for this project.
The idea is to produce application-specific JSON under the control of
some template system, whether that's application-specific idiomatic JSON
or JSON-LD. The standard result format is already supported and well
used - it encodes a result set in JSON with all the RDF details showing.
>
> For "Define SPARQL commands in Jena rules",
>
> I remember there's a paper by Axel Polleres [2]. It's about translating
> SPARQL into datalog. Although datalog is arbitrary variables and Jena rules
> is only 3 variables, there should be a way to convert.
>
> But I don't quite understand how it can "increase the expressiveness of
> Jena". In principle, SPARQL is just indexing, and Jena rules actually have
> already indexed the triples while grounding variables.
On a technical level, SPARQL can say things the rules can't, or can only
in a complex way. GROUP BY and aggregation for example.
On a deployment level, users know SPARQL syntax so being able to connect
that knowledge to the rules would be good. The SPARQL engine also has a
quite extensive collection of XSD operations.
Andy
>
> [1]. http://www.w3.org/TR/sparql11-results-json/
> [2]. Axel Polleres: From SPARQL to rules (and back). WWW 2007
>
>
> Best regards.
>