GSoC 2012

[Gabor Makrai <ma...@gmail.com>]

Hi,

Last year, I was so happy when I realized that Mahout was offering some
task for students who wanted to participate in GSoC, but that time I
haven't got enough time for apply it. So this is the time when I want
to create something new for Mahout and experience how can I work together
with an international community :)
I know the application deadline is super close, but I believe that we can
found a proper task for me!

Please, let me introduce myself shortly! I got my Computer Science bachelor
degree at Budapest Technical University and Economics in Hungary. After my
graduation, I continued my studies at Budapest, and I decided to deepen my
knowledge in distributed systems and data mining. Meanwhile, I had a chance
to study abroad, so I spend a semester at University of Bradford, UK in
2009 and a semester at University of New Hamsphire, NH, US in 2011. I
studied some advanced Artificial Intelligence (especially AI in computer
games) at University of Bradford, and I studied algorithm theory and
algorithm design at UNH. From September 2010, I am working on Radoop (
http://blog.radoop.eu/) which will provide an user-friendly graphical
interface for Hadoop. Radoop work with Hive (it uses the Hive JDBC Driver),
but I have great experience in base Hadoop (HDFS and MapReduce), because we
have some unique MR job to preprocess data. Moreover, we managed to
integrate a part of Mahout to Radoop, so I have knowledge about the base
architecture of Mahout! :)

I founded a closed issue (https://issues.apache.org/jira/browse/MAHOUT-364)
which propose an implementation of Neural Network. I'm really interested to
make something similar (especially a classifier for Mahout and I chose NN,
because I have some experience in using and implementing it). Unfortunately
I don't have a unique dataset for this task, but I can use dataset from UCI
Machine Learning Repository. I don't see exactly the whole situation of
Mahout classifier's, so I like to ask your help to find an appropriate task
which can be useful for the entire Mahout community and with this work, I
can become a value member of this community!

Thank you,
Gabor Makrai