This directory contains a simple end-to-end example to use vowpal wabbit for
multi-class classification. The task relates to DBpedia ontology classification
where the labels are one of 14 ontology classes (e.g. Company, Film, Athlete)
and the input data consists of the title and content associated with the entity
in DBpedia.

This is adapted from the classification-example.sh from fastText
(https://github.com/facebookresearch/fastText).

In order to run:

1) Run `make` in the vowpal_wabbit root directory
2) Make classification example executable: `chmod +x ./classification-example.sh`
3) Run `./classification-example.sh` which will:
  (a) Download and parse the data to convert it into the vw format
  (b) Train a model
  (c) Evaluate the model
  (d) Output the predictions

You can also evaluate the predictions (you'll need python and scikit-learn and
its dependencies installed). After running the script above you can run the
evaluation script (thanks to @Scott-Graham-Bose):

  python eval.py --truth data/dbpedia.test --pred result/dbpedia.test.predict
