Add nounphrase_consolidate() ?

[kbenoit]

PR #119 adds the ability to extract noun phrases. Should we allow this list of noun phrases to be applied to a spacy_parsed object to be consolidated the same as entity_consolidate()? Seems like it would make sense, and provide an alternative to entity_consolidate().

We could call this nounphrase_consolidate(). It would provide this workflow:

1. sp <- spacy_parse() on input text to return a data.frame, classed as spacyr_parsed.
2. np <- spacy_extract_nounphrases() on the input text to get the noun phrases.
3. nounphrase_consolidate(x = sp, nounphrases = np) to turn the noun phrases into something that looks like a spacyr_parsed object but with the noun phrase sequences combined in a way similar to the operation of entity_consolidate().

Or: Is there a (more efficient) way to do this in one step, when calling spacy_parse()?

[amatsuo]

I have implemented nounphrase_extract() and nounphrase_consolidate(). See
https://github.com/quanteda/spacyr/blob/noun-phrase-v2/tests/misc/test_nounphrase_extraction.html

To have the same functionality as entity_*, nounphrase_consolidate() works independently from nounphrase_extract()`.

To try out, please install the noun-phrase-v2 branch.

There are a few points to do/consider:

• tests
• we need to make sure that once one of entlity_consolidate() or nounphrase_consolidte() is applied, the other method cannot be applied. My suggestions are:
  1. add another class to the output (entity_consolidated) or
  2. drop the field for the other (e.g. with nounphrase_consolidte() execution, entity field will be removed
• I am not sure I have NULL-ified all relevant names used in the function with data.table objects