Punk n Love Texts · posted by vaibhav bhawsar Feb 15, 04:39
Here are some results of training the Bayesian analysis on a bunch of love/mush song lyrics and another bunch of Punk lyrics.
I then ran the filter on texts like President Bush Delivers State of the Union Address, THE ORIGINAL “QUIT INDIA” RESOLUTION, Koran, a couple of manifestos (yes including the communist manifesto again!!). Some funny and some obvious and some not so obvious results follow.
The Bayesian analysis is binary? only outputs something like “this is good” and “this is bad”??
CONSIDER:
PersonA style of writing is Atype
PersonB style of writing is Btype
PersonC style of writing is Ctype
PersonD style of writing is Dtype
Now that we can say we know how each person writes(ATYPE, BTYPE, CTYPE ETC) is it possible to input a text and compute which style it is closest to?
As of now in the Spam filter exercise we know if a given text is spam or not. So its a boolean result. True or false. But with the above example we have four possible outcomes (excluding the result “this text does not match any of the four styles”). So how do we accommodate a logic to compute more than two outcomes using bayesian analysis?
MSC:
this message was not spam according to the filter!! while gmail says its spam! I say its spam! Maybe the filter needs more training set/data.
————————————————————————————————————-
Hi,
Save over 50% on your medication
http://www.ledrx .com
Remove space in the above link
The rain was still drumming heavily against the high, dark glass.
Another clap of thunder shook the windows, and the stormy ceiling
flashed, illuminating the golden plates as the remains of the first
————————————————————————————————————-
Commenting is closed for this article.