Philosophy, computers, plumage.
I'm in ur toobz, hidin' ur spam.
Posted 10-24-2008 at 03:39 PM by Dromiceius
I know I'm hardly the only person who hates Youtube comments. I am, however, the first person I know of to implement the particular solution I wanted to try.
I wrote a GreaseMonkey script, mainly to see how well I could isolate the signal from the noise. My method was to filter out internet-isms, lol-speak, and a few words that tend to occur in e-penis duels.
The code itself isn't complicated. It matches every comment against a bunch of regular expressions, and hides the comment when it finds a match...
Edit: After more experimentation, I'm finding that it worked better before I fixed all the false positives in the regexes.
I think a simple random filter would improve things. Maybe by judging "bogosity" as a function of comment length and any evidence signifying a basic understanding of English.
I'm aware of a Firefox addon called "Youtube Snob" which seems to do a subset of my filters, and which is a pain in the ass to extend. Obviously, I don't care about that.
I wrote a GreaseMonkey script, mainly to see how well I could isolate the signal from the noise. My method was to filter out internet-isms, lol-speak, and a few words that tend to occur in e-penis duels.
The code itself isn't complicated. It matches every comment against a bunch of regular expressions, and hides the comment when it finds a match...
Edit: After more experimentation, I'm finding that it worked better before I fixed all the false positives in the regexes.
I think a simple random filter would improve things. Maybe by judging "bogosity" as a function of comment length and any evidence signifying a basic understanding of English.
I'm aware of a Firefox addon called "Youtube Snob" which seems to do a subset of my filters, and which is a pain in the ass to extend. Obviously, I don't care about that.
Code:
/* ==UserScript==
@name Youtube Comment Sanitizer
@include http://www.youtube.com/watch*
@description Filters out the incoherent, leaving the almost-coherent.
==/UserScript==*/
var pileoshit = [/L+O+L/i, /than[kx]/i, /thx/i, /[!\?][!\?][!\?]+/, /z*om.g/i, /ro[l|f]/i, /[^a-z]ty/i, /fu+ck/i, /shit/i, /wtf/i, /wo+t/i, /lmf*ao/i, /stfu/i, /[\<]+[\3]/i, /teh/i, /:[pPdD\)]/, /ga[yi]/i, /x[dp]/i, /[ \t][\?]/i, /[ \t]r[ \t]/i, /[op]wn[z|ed]*/i, /jap[ \t|s]/i, /p[r|o|0]+n/i, /epic/i, /fail/i, /boob/i, /fap/i, /nuff/i, /\^\^/i, /[\.\t]COM/i];
// /cunt/, /jew/, /fag/, /nigger/, /queer/, /homo/ //matches too seldom, surprisingly
// /^[^A-Z]/, /.*[^\.]$/, /[\.]+[a-zA-Z]/, //always matches
(function() {
var comments = document.getElementsByTagName('DIV');
if(comments.length < 1) return;
for(i = 0; i < comments.length; i++) {
var match = 0;
for(j = 0; j < pileoshit.length; j++) {
if((comments[i].className=='watch-comment-body') && comments[i].innerHTML.match(pileoshit[j])) {
comments[i].parentNode.style.display='none';
match = 1;
break;
}
}
if(match) continue;
}
})();
Total Comments 1
Comments
|
LOL dude.
|
Posted 10-26-2008 at 02:04 AM by Ben Burkhill
|
Recent Blog Entries by Dromiceius
- I'm in ur toobz, hidin' ur spam. (10-24-2008)
- Fun with AutoHotkey (09-14-2008)
- Vocalise (08-23-2008)
- Hmm. (07-30-2008)
- Death Note. Ending. Sucked. (07-05-2008)
