Convenience method for doing a full evaluation of a parsing model on a set of test parses.
Convenience method for doing a full evaluation of a parsing model on a set of test parses.
the parser to use
(comma-separated) filenames of the gold parse files to test the parser on
file format of the gold parse files
where to find the files (i.e. "datastore" or "local")
size of the n-best lists to generate for oracle scoring
Command-line for parsing a test set in CoNLL format.
Command-line for parsing a test set in CoNLL format.
Usage: ParseFile [options]
-c <file> | --config <file> the file containing the JSON configuration for the parser -t <file> | --test <file> the file containing the test parses to parse and compare against (CoNLL-X format) -d <file> | --datasource <file> the location of the data ('datastore','local')
see above
Determines the oracle score for n-best parsing a sequence of parses.
Determines the oracle score for n-best parsing a sequence of parses.
the parser to use
data source for the gold parses to check against
size of the n-best lists generated by the parser
Re-parses a sequence of parses, and compares the results to the gold standard.
Re-parses a sequence of parses, and compares the results to the gold standard.
the parser we want to use to parse the sentences
the gold parses