parse: Parse CoreNLP output formats.
In PolMine/bignlp: Fast and Memory-Efficient Annotation of Big Corpora

Parse output from CoreNLP.

corenlp_parse_json(
  x,
  cols_to_keep = c("sentence", "index", "word", "pos", "ner"),
  output = NULL,
  logfile = NULL,
  progress = TRUE
)

corenlp_parse_xml(x)

`x`	character vector, the JSON string(s) to be parsed
`cols_to_keep`	columns to keep
`output`	a destfile
`logfile`	a character string naming the file to an error log to; if provided, json strings will be written to this file if parsing the json string string fails
`progress`	logical

The JSON results of applying the Stanford CoreNLP annotators can be written to a streaming JSON file (ndjson format). corenlp_parse_json will parse a json string to a data.frame. If output is specified, the output will be appended to the file provided. If output is NULL, a data.frame is returned. Strings that cannot be parsed are written to the logfile, if it is defined. If filename is present, the function will process one or more files with the output of Stanford CoreNLP in a NDJSON format. If the argument output has been defined during initialization, results are written/appended to that file. Otherwise, a data.frame is returned.