[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-devel
Subject:    Re: KDE Naive Bayes (or other ML algorithm)
From:       Nathan Toone <nathan () toonetown ! com>
Date:       2005-05-24 23:59:23
Message-ID: 200505241759.23534.nathan () toonetown ! com
[Download RAW message or body]

[Attachment #2 (multipart/signed)]


On Tuesday 24 May 2005 05:50 pm, Dirk Mueller wrote:
> What is your target ? predict user interaction?

I think that there are lots of applications for machine learning in an 
operating system.  Automatic learning of mail filtering rules, better music 
playlist randomization, and guessing which news stories a user is more likely 
to want to read in an RSS reader are just a few applications.

I work with machine learning algorithms all day long, and there are nearly 
limitless possibilities for application.  I was thinking of some sort of core 
library (maybe part of kdelibs, but it doesn't have to be) that will take 
calls (either "train" calls or "analyze" calls) via DCOP and handle all the 
algorithm stuff under the covers.  You could even create user interfaces for 
classifier creation for all types of filtering applications - web page 
content classification, spam filtering, phishing recognition, virus 
detection, etc.

It's really just a pet project I'd like to adopt - because ML is where my 
interests lie, and it seems that having a "smart" operating system (or 
desktop manager, as the case may be) would be a useful thing...

Sorry if that sounds too much like an advertisement - I'm pretty passionate 
about it.  :)

-Nathan
 

[Attachment #5 (application/pgp-signature)]

>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<


[prev in list] [next in list] [prev in thread] [next in thread]