Write entire data set to a Feather file and read back in
Display a parse tree
Retokenize with a BERT subword tokenizer
Show reconstructing a sentence's span using group by and aggregation
Run document text through the Stanza EWT dependency parser (https://stanfordnlp.github.io/stanza/available_models.html) and compare the outputs against the gold standard. Or alternately use SpaCy's parser, with the caveat that it's trained on OntoNotes which has a slightly different schema.
Create an example notebook that shows analyzing the EWT data set
Major steps: