Results
We think http://www.myalpinepath.blogspot.com is written by a woman (65%).
Since my research involves quite a bit of neural networks and their applications, I am really curious to see all the attributes that the NN behind this site used. Also, I plugged in some of the blogs of my friends and it gave an accuracy of 85%(among 15 to 20 people) and that's not bad! (Ok! Getting out of the report writing mode...). Try it out, its fun!
7 comments:
You know i can't answer that question....ur identity is anonymous! How can we know it?
well..one more interesting fact..i tried this gender analyzer on other blogs that i know of...and guess what?? it failed 3/3 times...howz that?
vissu, lol! Guess the NN couldn't predict it accurately enough. Or the writers were very gender neutral in their writings. I'm suspecting that the software should scan for particular words as one attribute. There was one guy's blog which the software said was a girl's....and with 90% accuracy :D But its fun :)
Not 90%... 78% :P
Madhan, Really? Seems the NN in the website is not as good as touted. Or maybe it didn't have enough Indian bloggers in its training set... Will check your post on it.
count the word counts of all posts and sort it by decreasing order of frequency after removing common words like the,a etc.
if words like programming,bar,girl,bro occurs with higher frequency than shopping,guy,cute etc then its a male otherwise female.
This should have a better prediction ratethan this:)( which says its a woman for the few sites I tried.)
actually this shud also work-
return doesBlogMentionPornSomewhere() ? "male":"female" ;
Arun, lol! Though girls do talk about shopping and cute guys, I'm not sure how many would actually blog about it. And porn?? I've seen very less guy friends talk about porn, much less than the female-author-blogs that I've seen. But yeah, these general guidelines are sometimes predictable, sometimes surprising and all the time, interesting! Guess that's what makes NN fun!
Post a Comment