Hi Pantelis,

I also support your plan of dividing the session in two parts.
In the beginning of the first part I would give a (short) technical overview what is neccessary for an OT-webservice (RESTLET, Jena, OWL,...). 
To keep part 1 short, I would concentrate on few points (download and parse 1 OpenTox resource like dataset (for other resources it is technically similar) and learn 1 model with it), since I think the second part is the more important one which also will be very time-consuming.

I think building an OpenTox web service with some guidance is that what is expected from the audience from us. I am very fine with the procedure you proposed in part 2. What do you think of showing and explaining the participants at the begining of part 2 some small code fragments that are neccessary for developing their webservice (related to Nina, I second to use as much of the existing code for time and efficiency constraints).


Best regards,

Fabian
-------- Original-Nachricht --------
> Datum: Mon, 06 Sep 2010 18:22:04 +0300
> Von: chung <chvng at mail.ntua.gr>
> An: development at opentox.org
> Betreff: [OTDev] Techie Table

> Hello All,
>    I'm sending out some notes/thoughts about the Techie Table 
> 
> @All:
>    I'm thinking of organizing the discussion and presentation in two
> parts. In the first we are going to show how a developer can consume
> OpenTox web services from inside a Java application using some tools of
> ours. This will include some A&A also. Using web clients and parsers,
> the participants will download and parse OpenTox resources such as
> compounds, features, datasets and models and inspect them from within
> their Java application. Exercises will include parsing of datasets and
> conversion into weka data objects (Instances), training of various
> models using NTUA, TUM and AMBIT algorithms (where necessary using also
> the A&A API). So this will familiarize the participants with the direct
> consumption of OT web services on a more advanced level than  using web
> interfaces and a more programmatic way than using plain curl. This will
> take about 45 minutes including discussion/questions.
>    In the second part, I'm thinking of having the audience building an
> OpenTox web service with some guidance. We will design and deploy a
> clustering algorithm for the sake of getting in touch with web services
> and having a better insight into the OpenTox API. First we will study
> the requirements of the OT API and will formulate hypothetical curl
> commands for how a request would look like (In a top-down approach).
> Then we will proceed formulating an RDF representation of the algorithm
> discussing at the same time about the structure of such a document and
> the information that should be found therein. Jena will be used as an
> RDF editor/parser, which will be quite useful to developers that want to
> get involved into any contemporary project (web semantics etc). We will
> not get into much detail as far as the storing of models is concerned
> but briefly we'll present the underlying database tables structure. Weka
> will be used to materialize the XMeans algorithm. The whole effort will
> boil down to just about 500 lines of source code including the use of
> various Java libraries such as Restlet, Weka, Jena and DeciBell.  This
> will take about 75' and the rest of the time we will discuss about
> various challenges a developer can take.
> 
>   I'll be sending you a list of what should a participant have installed
> by the end of the day. If any participants experience problems
> installing any of the libraries/tools/programs in the list, we'll do it
> together on the table.
> 
> Best regards,
> Pantelis
> 
> _______________________________________________
> Development mailing list
> Development at opentox.org
> http://www.opentox.org/mailman/listinfo/development

-- 
GMX DSL SOMMER-SPECIAL: Surf & Phone Flat 16.000 für nur 19,99 Euro/mtl.!*
http://portal.gmx.net/de/go/dsl