This program contains (will contain) a classification engine for news articles that distinguishes articles that are centrally about topics of charity and altruism from such that have nothing to do with those topics or only touch on them peripherally.
This engine is (will be) used to run a Tumblr or Twitter account that publishes links to these articles based on a broader news feed from Google News or Kuerzr.
Our envisioned reader is interested in topics that maximize the following in roughly that order:
1. Altruism 2. Effectiveness 3. Timeless relevance
You may have to install some packages globally:
sudo aptitude install python-numpy python-scipy libzmq-dev g++
Then just execute the following commands and hope for the best.
python2.7 bootstrap.py bin/buildout
You may have to run
bin/buildout twice if it can’t download the NLTK data on the first run when
bin/python does not yet exist, but theoretically that shouldn’t happen.
Running the Manual Classification
See the usage message with the
--help option for explanations. Here an example.
bin/manual_classification -m 0 -f data/altruism.2014-05-13.json link_url content.title content.body
Generating the Corpora
Running the Evaluation
Running the Daemon
Running the Tests