High-dan game network

Comments, questions, rants, etc, that are specifically about KGS go here.
walleye
Dies with sente
Posts: 124
Joined: Wed Apr 28, 2010 7:56 pm
Rank: IGS 1k
GD Posts: 0
Has thanked: 5 times
Been thanked: 13 times

Re: High-dan game network

Post by walleye »

Mef wrote:Perhaps a better graph would show connections for people who post in the same threads (instead of playing each other in games) (=

It would be even better to make a bipartite graph with two kinds of nodes: 1) usernames and 2) thread topics. You need some sort of a binary relation between the nodes to make a social network. People who post in the same thread are not necessarily connected to one another, rather they are connected to the thread.

I would particularly want to see this network done for the old godiscussions forum. Someone just needs to write a little program to harvest the data: a list of thread topics and a list of people who posted in each thread. I will then make the network image easily.

Another idea is a network of quotations. If you make a post that quotes what someone else has written (like I'm quoting Mef here), this constitutes a binary relation which one needs to make a network, a directed network in this case. Again, the problem is to harvest the data.

Yet another idea is to make a network of references for the sensei library. Most pages in the library reference some other pages. This can be used to make a network. I'd really want to see this network too.
User avatar
Dusk Eagle
Gosei
Posts: 1758
Joined: Tue Apr 20, 2010 4:02 pm
Rank: 4d
GD Posts: 0
Has thanked: 378 times
Been thanked: 375 times

Re: High-dan game network

Post by Dusk Eagle »

Not to tout my own horn, but http://lifein19x19.com/forum/viewtopic.php?f=8&t=3052 contains some graphs I made showing how often two people post in the same thread. Granted, it doesn't implement your idea of using thread names as nodes, but it's still something.

If you want to make more graphs with the data, you can ask me and I will provide you the raw data I used to generate my graphs. Granted, it doesn't contain thread names for the list of posters in a thread, but it would not be hard for me to modify my parsing script to include the thread name as the first element of each list of posters. Also, due to technical limitations, I could only gather the first twenty posts of each thread, so the data is a bit incomplete. I couldn't find a good solution to download all the pages of a thread or to get all the posts of a thread onto one page for wget.

As for parsing godiscussions.com, that's unfortunately not currently possible, as any thread containing FlashGoban or Go diagrams refuses to load. If we had access to the database that the L19 admins have, then it might be possible.

The senseis library idea does sound interesting. If you or someone else doesn't look into it first, I might look into it on Thursday (my next day where I have free time).
We don't know who we are; we don't know where we are.
Each of us woke up one moment and here we were in the darkness.
We're nameless things with no memory; no knowledge of what went before,
No understanding of what is now, no knowledge of what will be.
Post Reply