ter

TER Translation Error Rate automated machine translation evaluation

measure_ter

 measure_ter (hypothesis_lines, reference_lines, normalized=False,
              no_punct=False, asian_support=False, case_sensitive=False)

Measuring TER on set of hypotheses and references

	Type	Default	Details
hypothesis_lines	List		Array of hypotheses.
reference_lines	List		Array of references.
normalized	bool	False	If `True`, applies basic tokenization to sentences.
no_punct	bool	False	If `True`, removes punctuations from sentences.
asian_support	bool	False	If `True`, adds support for Asian character processing.
case_sensitive	bool	False	If `True`, does not lowercase sentences.
Returns	Tuple		Array of arrays containing TER score, number of edits and reference length and sacreBLEU TER score in JSON format.

source

measure_record_ter

 measure_record_ter (hypothesis_lines, reference_lines, sourcelang,
                     targetlang, test_set_name, test_date, mtengine,
                     score_pathname, score_fname, domain='',
                     normalized=False, no_punct=False,
                     asian_support=False, case_sensitive=False)

Score hypothesis with TER score and record it to a specified metrics file

	Type	Default	Details
hypothesis_lines	List		Array of hypotheses.
reference_lines	List		Array of references.
sourcelang	str		Source language identifier.
targetlang	str		Target language identifier.
test_set_name	str		Name of test set.
test_date	str		Date of test as ISO-8601 formatted string.
mtengine	str		Name of the tested MT engine.
score_pathname	str		Path for results CSV file.
score_fname	str		File name for results CSV file.
domain	str		Domain string if applicable.
normalized	bool	False	If `True`, applies basic tokenization to sentences.
no_punct	bool	False	If `True`, removes punctuations from sentences.
asian_support	bool	False	If `True`, adds support for Asian character processing.
case_sensitive	bool	False	If `True`, does not lowercase sentences.