I'm using the Google Cloud Language API (GCL API) to do syntactic analysis of English sentences. In particular, I'd like to identify all noun phrases. I don't care how they are used (subject, object, etc.) but I just want the sequence of words corresponding to each noun phrase.
Using some simple rules based on parts of speech, I can get a lot of them (e.g., DET NOUN, DET ADJ NOUN) but I'd like to do better. Noun phrases that include verbs and hyphenated words get complicated.
The response from GCL API includes a lot more than part of speech, such as
- A head token index of a dependency tree (see here)
- A parse label with about 40-50 options (same link) such as "Adjectival complement" or "Adverbial clause modifier"
Is there a relatively straightforward way to use this extra info to parse out the noun phrases?