[NLP2RDF] NIF + SADI (The Semantic Automated Discovery and Integration)

[Artjom Klein]

Hi Sebastian,

thank you very much for your reply.


On Mon, Apr 16, 2012 at 6:17 PM, Sebastian Hellmann <
hellmann at informatik.uni-leipzig.de> wrote:

> **
> Dear Artjom,
> awesome to hear, that you are using NIF!  Would you like to write a blog
> entry about it?  It would be really helpful as input for NIF 2.0,
> especially, if you present your use case.
> (Short update: we are still sorting out the plentiful feedback we received
> on NIF since November, soon we will form an advisory board and then we will
> start work towards NIF 2.0 as a community).
>
>
I will think about it. The use case is very small and does not show all the
power of NIF.


> Regarding the SADI Framework: It seems to be difficult to answer your
> questions here, as we are not the SADI experts.
> Maybe we can cross-post to their dev list ( sadi-dev at googlegroups.com)?
> The framework sounds alright and it would be very good, if we can reuse
> something for NIF, especially the Web service specification and validation
> and the error handling.
>
> I was actually wondering, however,  if NIF services aren't already SADI
> services: http://sadiframework.org/content/how-sadi-works/  , if you just
> describe NIF services as something that takes as input and output a
> str:String .
>
> I think you are right here. The main difference (that I can see now) is
that SADI services have semantic descriptions. This makes possible service
discovery, orchestration and automatic composition of services.
I am trying to find a use case that will show the power of combination
NIF+SADI. Service discovery? Automatic service composition?
This works very well in the data integration and knowledge discovery
scenarios. Since SADI can be also used as tool integration framework, it
would be interesting to leverage it. NIF can be useful here.
I would appreciate any ideas in this direction.


> Did you have any actual benefits from using SADI? i.e. any
> tools/vocabularies/infrastructure they provide?
>

I will mention that I've used:
1. Service template/sceleton generation command line tool (There is a
Protege plugin to generate service skeleton code in either JAVA or Perl,
and enable the testing of SADI services with input taken from the currently
opened ontology or from files, but I've never used it.)
2. Registry to register services
3. CardioSHARE SPARQL Client (http://biordf.net/cardioSHARE/) to query
distributed data. It can use registry of SADI services and call them if
required due query planning and execution.
4. I also tried Taverna SADI plugin to compose pipelines of SADI services.
I can't say much because I tried only one "HelloWorld" example.


> By the way, your service seems to be down:
> http://unbsj.biordf.net/ie-sadi/extractDrugNamesFromTextV4
>
> All the best,
> Sebastian
>
>
>
> On 04/16/2012 09:39 PM, Artjom Klein wrote:
>
> I would like to know what would be a benefit of wrapping NLP tools/services
> as SADI services?
>
> SADI services are semantic web services which consume and produce instances
> of OWL Classes. Describing input and and output of services in terms of
> some reference ontology gives possibility to make services interoperopable,
> interchangeable. The SADI allows automated discovery of distributed data
> resources and the automated orchestration of chains of Web Services into
> complex workflows.
> More details can be found herehttp://sadiframework.org/content/links-and-docs/
>
> Recently I developed a couple of text mining SADI services using NIF-1.0
> Spec. (One of themhttp://unbsj.biordf.net/ie-sadi/extractDrugNamesFromTextV4
> ). They work in a
> workflow with another SADI service. Basically it is a very basic pipeline
> consisting of one NLP service to extract drug names from text and a service
> which checks dangerous drug-drug interactions between drugs found in the
> text and drugs in the personal drug consumption database (it simulates a
> personal medical health record). Actually I implemented two alternative
> services to extract drugs. Since they are interchangeable, it was easy to
> switch between them and choose one which performs the best.
>
>
> Now I am trying to figure out if there is *a motivation to continue work on
> building SADI NLP services*???
>
> Pro:
> - easily interchangeable, easy to build pipelines
> - fits well in scenarios of real-time data discovery
> - fits well in data integration scenarios
>
> - Artjom Klein
>
> Computer Science Researcher at University of New Brunswick Canadahttp://www.unbsj.ca/sase/csas/facultyPage.php?member=Artjom%20Kleinl
>
>
> _______________________________________________
> NLP2RDF mailing listNLP2RDF at lists.informatik.uni-leipzig.dehttp://lists.informatik.uni-leipzig.de/mailman/listinfo/nlp2rdf
>
>
>
> --
> Dipl. Inf. Sebastian Hellmann
> Department of Computer Science, University of Leipzig
> Projects: http://nlp2rdf.org , http://dbpedia.org
> Homepage: http://bis.informatik.uni-leipzig.de/SebastianHellmann
> Research Group: http://aksw.org
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.informatik.uni-leipzig.de/pipermail/nlp2rdf/attachments/20120417/b32b44dd/attachment-0001.html>