Academic Blog

The pdfs of VLDB 2009 papers are already available in DBLP. Let me immediately draw your attention to the section on Experiences and Lessons, wherein the last paper is about DBLP, by Michael Ley. It’s really great to read about something so popular among my academic colleagues. I hope you’ll take a look at it.

Pretty soon there will be one year since we have released the first ICE. There will be surely a lot of noise around this first anniversary. In the meantime, let’s look around. Of course everyone knows MySQL and Pentaho (and some might hear about Weka - now also a part of Pentaho). Some may also know other open source columnar stores, such as MonetDB and LucidDB. However, it’s even more important to realize that there are newcomers to the Open Source World practically every day.

A couple of days ago we received the following message from one of our friends, who had helped us in making Infobright’s technology successful and then decided to go back to the Realm of Data Mining. Of course we wish him success also with this new undertaking and we hope to be able to do something cool together in the nearest future. Here is the message:

=======

I am pleased to announce that TunedIT system has been publicly released and is available at:

http://tunedit.org/

TunedIT facilitates evaluation and comparison of data mining and machine learning algorithms. Its website contains exhaustive data on performance of nearly 100 algorithms from Weka and Rseslib libraries, tested on tens of different datasets from UCI Repository. You can extend this information with your own test results, as well as upload new algorithms and datasets.

I invite you to visit and explore TunedIT and to contribute new algorithms, datasets and results. I hope TunedIT will prove useful in your research. If you find our system interesting, please help us improve it by sharing your comments.

=======

So I’ve just looked at TunedIT and I guess I’ll be a frequent visitor there. (Apart from Infobright, working on new data mining algorithms is a kind of my hobby; moreover, as I wrote before, there may be a place for data mining algorithms even inside Infobright engine.) Surely, there are such websites as that of the UCI Repository but - as the name suggests - it’s just a repository of data mining / machine learning data benchmarks, without such a convenient and sophisticated testbed framework that TunedIT provides.

So, maybe TunedIT should be rather compared to tpc in the world of databases?

And, by the way, there is also a mention about Rseslib - again, something for the fans of rough sets!

Best greetings,

Dominik