Stanford

The Stanford output is included mostly as a convenience and a debugging tool, but if you need a more reliable corenlp.run implementation here you go.

This is the result of the Stanford CoreNLP analysis of the text using the following annotators:

• tokenize: recognizes individual words.
• ssplit: recognizes sentence boundaries.
• pos: identifies the token's part of speech.
• lemma: identifies the root form of the token.
• parse: creates a syntax tree of the sentence. Main clause, independent clause, noun phrase, verb phrase, etc.
• depparse: creates a dependency graph of the sentence. Subject, verb, object, modifier, auxilliary, etc.
• ner: recognizes named entities. Is this a person? Is this a place?
• coref: links pronouns to antecedents.
• quote: recognizes quotations and links quotes to speakers.
• sentiment: classifies the emotional content of a sentence on a scale from most negative to most positive.

The structure of the STANFORD json is as follows:

• document
  • various document level annotations
  • corefs
  • quotes
  • sentences
    • various sentence level annotations
    • tokens
      • various token level annotations
      • word
      • lemma
      • pos
      • ner