...Pipelines produce CoreDocuments, data objects that contain all of the annotation information, accessible with a simple API, and serializable to a Google Protocol Buffer. CoreNLP generates a variety of linguistic annotations, including parts of speech, named entities, dependency parses, and coreference.