Elias Ponvert

Switch the code from Apache 2.0 to MIT (or Simplified BSD) to make this work GPL compatible

Requires #6, removing chunk.py and ensuring this project is organized correctly as a library

Convert README to Markdown, update links to Github repo

When passing a test file to chunk.py, if it does not have annotations for evaluation, we currently have this error: ``` Exception in thread "main" java.lang.AssertionError at upparse.corpus.UnlabeledBracketSet.fromTokens(UnlabeledBracketSet.java:113) at upparse.corpus.UnlabeledBracketSet.fromString(UnlabeledBracketSet.java:97)...

And remove chunk.py This will facilitate operating upparse as a library

Currently chunk.py expects a train / test setup. Allow user to only supply a single file, to chunk, output results and optionally evaluated output.

For a slightly more realistic train / evaluation setup, the cascade parser currently chunks the evaluation dataset at each level right after it trains on that level in the training...

This affects cascade parsing. Currently, the chunk.py script must run (with full training) chunking twice -- once to re-chunk the training dataset to train the next level in the cascade,...