23 Mar
2009
23 Mar
'09
4:48 p.m.
Anamaria Stoica wrote:
Hi,
Just to quickly introduce myself. My name is Anamaria Stoica, and I'm
studying Computer Science at University "Politehnica" of Bucharest. I'm
currently in the 5th year. (graduation year)
I have a 2 year experience working with Java SE & EE on the ITerating.com
platform (http://www.iterating.com) (also with JS, AJAX, HTML&CSS). The team
(less than 10 members) was spread out in several countries (but mostly in
Romania) and worked remotely.
My working experience also includes a summer internship (in 2008) at Cresta
Technology, Inc. (http://www.crestatech.com) in Santa Clara, CA. There I
worked as a Software Engineer Intern.
Regarding the "Anti Vandalism Filters" project, before today it had as a
developer requirement Machine Learning knowledge.
I loved the idea of the project since the first time I saw it, not only
because of the Machine Learning component in which I have recently developed
a lot of interest, but also of the project as a whole.
So, my question is: does the project still imply some work on ML Algorithms,
just that it isn't a strong requirement as before or has that change ?
Also, will the project use some other components already developed towards
this end ?
The 'Anti-vandalism Filters' project description has changed because we
thought it would be easier and safer to make the project more
"technical", and have it use external services or third party tools for
vandal detection, rather than design and implement a mechanism from scratch.
The main idea that triggered the description change was that for the
automatic spam detection based on content there are third party tools,
which are tested and optimized by extensive usage. Starting from scratch
with a new algorithm implementation would only waste time in the wrong
place. However, there are other places that still need intelligent data
processing, like detecting what is a spam bot and what is a productive
user, and I don't know if there is a tool for checking content updates;
a third party spam detection tool is good for checking added comments,
but it will fail to detect evil edits, like adding cursing to pages.
Note that the objective is to have an effective vandalism detection and
recovery system in XWiki. How we do this is still a subject open to
discussions.
If you do have some ideas and you want to use your machine learning
knowledge to make a smart tool for this purpose, we'd be happy to hear
and discuss your vision on this project. So, please let us know how you
would like to approach the problem. This is one of the more open
projects, where lots of ideas are expected from the student.
Looking forward to hearing your proposal,
--
Sergiu Dumitriu
http://purl.org/net/sergiu/