Some things I would like to do given time:
- Adjust the parser to avoid the need for the fix-no-tag.py script (and even better, see if these cases are indicative of a programming bug)
- Add models that are trained on the CCK-style head rules
- Add example output / figures
- Add step by step instructions on installing the necessary libraries for the tagger
- Write a script to make the run process for the tagger a one-step procedure