PB Forums

Full Version: AC clans database
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Hi,

This is a small project I did a couple days ago: the ac clans database.

I was wondering how hard it could be to code a script that would auto-detect clan tags, given a list of nicknames, so I fired up my favorite text editor and started experimenting. Turns out that my script works decently, although there are still a few false positives, so I decided to make it run on a large set of data.

I got a list of the players who connected to the TyD ladder (yes, the 2400 pages), and ran my script on it. 13 hours later, I had a database, all I needed was a webpage to share it with you guys !

Not sure what to do with this, I just did it for kicks, but maybe some of you will want to play around with the script, and maybe improve it.

Cheers.
Very cool! Nicely done.
Wow there are so many!
Give it to pwnage so he can update his ladder if it will work with his code. We still can't be part of the clan war in the TyD Ladder because our B} doesn't like the B} in........... ( BoB}<<<< ) GOOD WORK btwBig Grin

Ohh crap I just posted B}4 looking at it. I dunno if that would work. Still though Great work I love looking at this. Now time to alphabetize Tongue
Thanks.

(05-16-2013, 05:25 PM)charlymurphie Wrote: [ -> ]We still can't be part of the clan war in the TyD Ladder because our B} doesn't like the B} in........... ( BoB}<<<< )

It's a shame, but this also tells me that there's no automatic clan tag detection system in their ladder (the tags seems to be added manually, and then compared to the data with a simple strstr).

(05-16-2013, 05:25 PM)charlymurphie Wrote: [ -> ]Ohh crap I just posted B}4 looking at it. I dunno if that would work. Still though Great work I love looking at this. Now time to alphabetize Tongue

As far as I know the members are correctly reported to their respective clans, what's wrong ?

Alphabetize ? Not sure what you mean by that.

Anyway, like I said before, it is impossible for a script to make a difference between a prefix used by several players and an actual clan tag.
Example:
Code:
unarmed(fr), (fr)unarmed2
Both players use the same «tag», and any human would know that it's just a way to tell the others that they are french, but the code will pick this up as a clan tag.

As a result, the script will generate a list of tags that are probable clan tags (and as far as I've seen, the results are relevant in most case), but you will always need a human to read the generated data and eliminate false positives. However, this is still a fairly weak way of solving the problem of false positives: nobody can pretend to know every clan in the community, and maybe one day, one of the tags you had labeled as «false positive» will be used by a newly formed clan. One solution to the problem would be to let players who visit your ladder vote for tags that are not used by any clan, but again this adds complexity to the system (what if somebody tries to remove the clan from the ladder but mass voting, etc). This script is only one part of the huge system that would automatically add and maintain a list of clans.

Quick note: the script doesn't only detect fixed strings in a list of nicknames, it tries several combinations of the candidate prefix. Players tend to modify the clan tag in a way that it's still obvious for a human player to recognize it, but that makes it more difficult for a script to detect the tag.
Example:
Code:
.eG[unarmed1, unarmed2.eG[, unarmed3]eG.
The above example is correctly parsed by the script.

Finally, it's going to be hardly usable in any ladder at the moment, it took ~13 hours for my script to generate the tags list off of a single core (2.6GHz). And it will literally put your server to its knees if parallelism is involved.