Using Jena for OBDA to access non-relational datastores

[Benedikt Tröster <tr...@stud.uni-heidelberg.de>]

Hello everyone,

I know Jena is not exactly made for OBDA, but I guess you could achive
sth. that is very close to what I want to do:
I have an ontology for a process inside the medical domain. The idea is
to use the ontology and query for the instances of data (which are
located inside doc, xls files and a sharepoint). The access to this data
should happen via a GUI where (as an example) you would click on: create
a report. The system I want to create should then find out which data is
needed (as stated in the ontology) and where it is stored and finally
process it to show it in the GUI or wherever.

I see that Jena does support access to triple stores, but I don't
actually want to use the Extract-Transform-Load (ETL) approach, but
instead use mappings and a rewritingsystem to access the data. This is
due to the fact that the data in the sources will be changing constantly
and therefore ETL does not seem fitting for me.

Do you have any recommendations for me? Would that be possible to
implement with Jena? Does it support query rewriting to other sources
other than SQL?

Currently I'm thinking about implementing this with the Ontop framework.

Thanks so much for your valuable input! Happy new year and all the best! :)

Best,
Benedikt

Re: Using Jena for OBDA to access non-relational datastores

[Claude Warren <cl...@xenei.com>]

Benedikt,

There are a number of ways to achieve this.  It is possible to create an
instance of Graph.  Assuming you only want to read (not write) the
documents via the Jena interface this would be fairly straight forward as
you could extend the GraphBase class (org.apache.jena.graph.impl.GraphBase)
which only requires implementation of one method to locate triples in your
store.

Once you have achieved that you can plug your implementation into the Jena
system.  This will get you started to show proof of concept.  Beyond that
you will probably want to do some work to optimize the queries.










There is a description of the Graph contract at
https://cwiki.apache.org/confluence/display/JENA/SPI+Contract+Details


Claude


On Fri, Jan 8, 2016 at 4:33 PM, Benedikt Tröster <
troester@stud.uni-heidelberg.de> wrote:

> Hello everyone,
>
> I know Jena is not exactly made for OBDA, but I guess you could achive
> sth. that is very close to what I want to do:
> I have an ontology for a process inside the medical domain. The idea is
> to use the ontology and query for the instances of data (which are
> located inside doc, xls files and a sharepoint). The access to this data
> should happen via a GUI where (as an example) you would click on: create
> a report. The system I want to create should then find out which data is
> needed (as stated in the ontology) and where it is stored and finally
> process it to show it in the GUI or wherever.
>
> I see that Jena does support access to triple stores, but I don't
> actually want to use the Extract-Transform-Load (ETL) approach, but
> instead use mappings and a rewritingsystem to access the data. This is
> due to the fact that the data in the sources will be changing constantly
> and therefore ETL does not seem fitting for me.
>
> Do you have any recommendations for me? Would that be possible to
> implement with Jena? Does it support query rewriting to other sources
> other than SQL?
>
> Currently I'm thinking about implementing this with the Ontop framework.
>
> Thanks so much for your valuable input! Happy new year and all the best! :)
>
> Best,
> Benedikt
>



-- 
I like: Like Like - The likeliest place on the web
<http://like-like.xenei.com>
LinkedIn: http://www.linkedin.com/in/claudewarren

Re: Using Jena for OBDA to access non-relational datastores

[kumar rohit <ku...@gmail.com>]

I am new to Ontop but I think it is suitable here in this scenario and you
can then use owl API implemented as maven.

Best regards

On Fri, Jan 8, 2016 at 4:33 PM, Benedikt Tröster <
troester@stud.uni-heidelberg.de> wrote:

> Hello everyone,
>
> I know Jena is not exactly made for OBDA, but I guess you could achive
> sth. that is very close to what I want to do:
> I have an ontology for a process inside the medical domain. The idea is
> to use the ontology and query for the instances of data (which are
> located inside doc, xls files and a sharepoint). The access to this data
> should happen via a GUI where (as an example) you would click on: create
> a report. The system I want to create should then find out which data is
> needed (as stated in the ontology) and where it is stored and finally
> process it to show it in the GUI or wherever.
>
> I see that Jena does support access to triple stores, but I don't
> actually want to use the Extract-Transform-Load (ETL) approach, but
> instead use mappings and a rewritingsystem to access the data. This is
> due to the fact that the data in the sources will be changing constantly
> and therefore ETL does not seem fitting for me.
>
> Do you have any recommendations for me? Would that be possible to
> implement with Jena? Does it support query rewriting to other sources
> other than SQL?
>
> Currently I'm thinking about implementing this with the Ontop framework.
>
> Thanks so much for your valuable input! Happy new year and all the best! :)
>
> Best,
> Benedikt
>

Re: Using Jena for OBDA to access non-relational datastores

[Andy Seaborne <an...@apache.org>]

http;//d2rq.org/ is dynamic and translates at query time.

On 08/01/16 16:33, Benedikt Tröster wrote:
> Hello everyone,
>
> I know Jena is not exactly made for OBDA, but I guess you could achive
> sth. that is very close to what I want to do:
> I have an ontology for a process inside the medical domain. The idea is
> to use the ontology and query for the instances of data (which are
> located inside doc, xls files and a sharepoint). The access to this data
> should happen via a GUI where (as an example) you would click on: create
> a report. The system I want to create should then find out which data is
> needed (as stated in the ontology) and where it is stored and finally
> process it to show it in the GUI or wherever.
>
> I see that Jena does support access to triple stores, but I don't
> actually want to use the Extract-Transform-Load (ETL) approach, but
> instead use mappings and a rewritingsystem to access the data. This is
> due to the fact that the data in the sources will be changing constantly
> and therefore ETL does not seem fitting for me.
>
> Do you have any recommendations for me? Would that be possible to
> implement with Jena? Does it support query rewriting to other sources
> other than SQL?
>
> Currently I'm thinking about implementing this with the Ontop framework.
>
> Thanks so much for your valuable input! Happy new year and all the best! :)
>
> Best,
> Benedikt
>