We are back on track now to delivering Weka as a web service. I finally got a freelancer Java developer to help me implement the required front-end. Everyone wanted to volunteer, noone had any time.
Rather than use a scattergun approach I am interested in getting some reader feedback (yes you the reader) on what specific part of Weka people would like to use first: classifiers, clustering or attribute selection. First come first served.
The basic service will also include the standard file handling in weka, so you can pass ARFF formatted file or just a plain CSV or just a URL to a file. You will know it is ready when I post an example of using the web service.
Once we are happy with a simple web service we can start to ramp up the functionality. I envisage that this service will be blended within other web services or as part of a Mashup.
Once we have demonstrated what the raw service is capable of, we can open the compute floodgates and use EC2 to ramp up and down as required.
Have Fun
Paul
Wednesday, June 25, 2008
Weka Web Service Prototype: Reader choice
Subscribe to:
Post Comments (Atom)
2 comments:
Paul:
I got into your blog late so I have to be educated on what Weka is. Once I know what you are trying to do I would be happy to provide feedback.
I found your blog while researching data mining tools on EC-2.
Theo
I'd like to see a basic, but full-lifecycle classifier application working: train, test, predict.
Post a Comment