Bayesian filters are actually pretty good at "sentiment analysis" - of
course the quality of the classification is based upon the corpus of
information fed into the filters.

I implemented this for a while with http://flixpulse.com for movie
review tweet analysis.  It wasn't perfect, but the filters got
"smarter" really quickly. The hardest part, of course, was spending
the time to train them initially by classifying tweets manually.

I had to somewhat abandon FlixPulse to move on to other things, but I
hope to revive it at some point.
-Chad

On Sat, Jul 18, 2009 at 6:57 PM, Nick Arnett<nick.arn...@gmail.com> wrote:
>
>
> On Sat, Jul 18, 2009 at 3:53 PM, M. Edward (Ed) Borasky <zzn...@gmail.com>
> wrote:
>>
>>
>>  And there is
>> http://twittersentiment.appspot.com/, a research project by some grad
>> students at Stanford. Perhaps you've heard of two other Stanford grad
>> students, Sergey Brin and Larry Page?
>
> Gilt by association?
>
> Nick
>

Reply via email to