Last updated by kyle_mathews on Sun, 2008-10-05 04:00
This was what I had to do on Ubuntu Hardy with memetracker alpha 6:
# Get python dependencies
apt-get install python-numeric
apt-get install python-numpy
# Make sure we have gcc
apt-get install libtool
# Other python stuff with magic snake powers
apt-get install python2.5-dev
# Get the remaining library
cd ~
wget http://bonsai.ims.u-tokyo.ac.jp/~mdehoon/software/cluster/Pycluster-1.43...
tar zxvf Pycluster-1.43.tar.gz
cd Pycluster-1.43
# Install it
python setup.py install - Install memetracker and machinelearningapi modules.
- Go to q=memetracker/1 and read directions.
- Go to q=memetracker/1/edit-feeds and add feeds.
- run cron
The menu structure of the memetracker seems to be broken. You can't access q=memetracker. Rather, you have to access q=memetracker/1 even if you've never created a meme before.
Personally, I think creating a memetracker is overkill. I think it should find memes in all of the site content automatically, with little or no configuration, and should offer one default view (akin to q=node), plus views integration.
The memetracker found no memes on my site. Not sure why.
[From Kyle -- internal meme detection is broken at the moment :( My focus has been on getting external memetracking (on stuff brought in by Feedapi) working as that's the use case I'm most interested and and I suspect most others. I need to get that working for the next release. Sadly my time is very limited now that school has started up again.]