Suggestomatic is a project mainly written in PYTHON and C, based on the MIT license.
Item based collaborative filter recommendation engine
Your starting point is a CSV dump of (user_id,group_id) tuples.
If you're using MySQL, you dump a table in the proper format with:
mysql -e 'select concat(user_id, \",\", group_id) as r from memberships)'
Data preparation:
python ./prepare.py_data --membership-csv=(file you dumped in #1)
When this step is done, it will print the command line arguments needed to load that data file into Suggestomatic
Run Suggestomatic:
make suggestomatic ; ./suggestomatic 1311574636.32-set-ids.bin 1311574636.32-set-members-index.bin 1311574636.32-set-members.bin suggestions.csv 100
derwiki@fordfiesta:~/src/suggestomatic/test$ nice ./test.py --only-csv --recommendation-file=test_data --num-groups=1000
Invalid option --recommendation-file=test_data. Try ./test --help for help.
derwiki@fordfiesta:~/src/suggestomatic/test$ ls -lh
total 1.7M
-rw-r--r-- 1 derwiki derwiki 1.7M 2011-07-25 22:24 test_data
-rwxr-xr-x 1 derwiki derwiki 5.0K 2011-07-25 22:17 test.py