Asklet - A learning algorithm for playing 20 questions
Overview
An algorithm for playing 20 questions, inspired by Robin Burgener's implementation, which is used to power the website 20q.net.
Note, Burgener's patent filing doesn't completely describe how the target rank is calculated, so I had to improvise my own method, but otherwise this should be a faithful implementation of his algorithm.
The project uses Django as a framework for organizing and managing a game domain, as well as recording games.
Installation
Some dependencies build Python extensions, so make sure you have the appropriate Python dev package installed for your platform.
e.g. On Ubuntu this would be something like:
sudo apt-get install python2.7-dev
or:
sudo apt-get install python3-dev
Then install the package via pip:
pip install asklet
If you're installing into the global Python install, as opposed
to a virtualenv, you'll likely need to prefix that with sudo.
Testing
For development work, checkout the project using git or download and
extract a release tarball.
Then, to run unittests, in the top-level project directory, execute:
python setup.py test
To run unittests for a specific Python version (e.g. 3), execute:
python setup.py test --pv=3
These commands will automatically create a local virtualenv and install all required dependencies.
Usage
For using the library as a Django app, start by creating basic Django project.
For reference, see the asklet/tests directory, which represents a minimal
Django project.
Then add asklet to your INSTALLED_APPS and run
manage.py syncdb; manage.py migrate asket; to install the models.
For using the library as a stand-alone classifier, TODO.