|
gptkbp:instance_of
|
gptkb:psychology
|
|
gptkbp:can_be_combined_with
|
human evaluation
|
|
gptkbp:developed_by
|
Papineni et al.
|
|
gptkbp:first_introduced
|
gptkb:2002
|
|
gptkbp:higher_scores_indicate
|
better translation quality
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
BLEU
|
|
gptkbp:is_compared_to
|
machine-generated text to reference text
|
|
gptkbp:is_criticized_for
|
not accounting for semantic meaning
|
|
gptkbp:is_evaluated_by
|
precision and recall
|
|
gptkbp:is_implemented_in
|
various software libraries
|
|
gptkbp:is_influenced_by
|
length of the reference text
|
|
gptkbp:is_language_independent
|
gptkb:true
|
|
gptkbp:is_often_used_in
|
NLP community
other metrics like NIST
|
|
gptkbp:is_part_of
|
evaluation framework for machine translation
|
|
gptkbp:is_popular_in
|
research papers
|
|
gptkbp:is_related_to
|
ROUGE
|
|
gptkbp:is_used_in
|
automatic evaluation of translations
|
|
gptkbp:measures
|
n-gram precision
|
|
gptkbp:modifications
|
brevity penalty
|
|
gptkbp:range
|
0 to 1
|
|
gptkbp:suitable_for
|
short texts
|
|
gptkbp:used_for
|
machine translation
|
|
gptkbp:bfsParent
|
gptkb:COCO
|
|
gptkbp:bfsLayer
|
5
|