[kepler-users] The most impressive thing you've done with Kepler

[christopher tuot]

Dear Ilkay,
 
Here some feedback about the Kepler experience we made out of our projects:
 
- The context
- The beginning
- Kepler installation
- Current functionalities/actors in Kepler:
- Implementing new actors:
- Performance:
- Thin Versus Rich client
- Conclusion
 
- The context:
I am currently working at the German Research Center for Artificial
Intelligence on different projects where we are using Kepler to process
Spatial Information (Maps in Vector Information). More information on my
Blog: http://ctuot.twoday.net/stories/4133335/
 
Before going into some Kepler technical specific details, I would like
to explain why we choose the Kepler platform.
 
- The beginning:
Everything started with the RAPR project. RAPR aims to develop a
rule-based prototype to process space-oriented agricultural knowledge in
order to economically grow organic commodities.
Experts in the agricultural domain needed a platform to develop and test
their scientific models. An example of model if the Biomass Yield Model
(bym) developed by Prof. Piorr and Sybille Brozio at the FH Eberswalde
in Germany. The original model has been implemented with the ESRI
Product ArcGIS. Developing a model with ArcGIS requires have some strong
knowledge in the GIS (Geographic Information System) besides domain
specific knowledge, in our case agriculture.
 
The visual programming capabilities of Kepler seamed to be a good
approach to allow non-GIS experts to develop and test their models.
Therefore we have implemented new actors in Kepler capable of handling
most of the standards GIS operations.
 
Now, let’s go into some more technical details:
 
- Kepler installation:
Installing Kepler is very easy. The couple of examples delivered with
the standard installation is enough to get quickly started with your own
Workflows.
 
- Current functionalities/actors in Kepler:
The standard library already offers quite a lot of already implemented
actors which could suffice for some “general” problems. For really
domain specific usage, you need to implement your own actors.
 
- Implementing new actors:
So honestly, implementing new actors is not something you can call easy.
There is a documentation about how to check out and configure fisrt
Ptolemy (the core of Kepler) and then Kepler itself. Usually, after you
checked out from cvs, you need to correct some pieces of code before the
whole thing can even compile. Moreover, the number of Jars in the
project is amazing and makes it really difficult to keep an overview
about the different versions used. I would rather have something like
maven and all my jars organized in a nice repository.
 
- Performance:
Our workflows usually have up to 15 actors, not more. However, we have
processed Gigabytes of data though the workflows. If the actors are
implemented correctly, you should not encounter any speed or memory
problems. Some of the workflows have ran for days (huge amount of data),
without any problems.
 
- Thin Versus Rich client
I was bothered by the fact you needed to install the whole Kepler just
to be able to edit and execute a workflow. Therefore we have deployed
Kepler as a client-server application. The server is a Web-Application
in Tomcat and the client is implemented with Adobe Flex 3 (Flash). This
allows workflows to be edited and executed in your brother.
 
- Conclusion
In my opinion, it is really worth using the Kepler platform. Even though
we encountered some problems in the first implementation phase, we
quickly managed to get over the few issues and now we are really
concentrating on implementing further functionalities. I have already
contacted Dr. Ludaescher to see how we can best contribute to the project.
 
If you have more questions, please do not hesitate to contact me.
 
Best regards,
 
Christopher Tuot
 
 
-- 
______________________________________________________________
Dipl.-Ing. Christopher Tuot
DFKI GmbH (German Research Center for Artificial Intelligence)
Knowledge Management Department
POBox 2080, 67608 Kaiserslautern, Germany
fon: +49(0)631 / 20575 - 127
fax: +49(0)631 / 20575 - 103
mail: christopher.tuot at dfki.de
web: http://www.dfki.de
______________________________________________________________
Deutsches Forschungszentrum fuer Kuenstliche Intelligenz GmbH
Firmensitz: Trippstadter Strasse 122, D-67663 Kaiserslautern

Geschaeftsfuehrung:
Prof. Dr. Dr. h.c. mult. Wolfgang Wahlster (Vorsitzender) Dr. Walter Olthoff

Vorsitzender des Aufsichtsrats:
Prof. Dr. h.c. Hans A. Aukes

Amtsgericht Kaiserslautern, HRB 2313
______________________________________________________________
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mercury.nceas.ucsb.edu/ecoinformatics/pipermail/kepler-users/attachments/20071019/9d5a8c2c/attachment.htm