Properties (54)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
evaluation metric
|
| gptkbp:criticism |
not accounting for semantic meaning
|
| gptkbp:evaluates |
precision and recall
summarization systems translation systems |
| gptkbp:grades |
better translation quality
|
| https://www.w3.org/2000/01/rdf-schema#label |
BLEU
|
| gptkbp:impact |
length of translations
|
| gptkbp:includes |
brevity penalty
|
| gptkbp:introduced |
2002
|
| gptkbp:is_a_key_player_in |
WMT_(Workshop_on_Machine_Translation)
|
| gptkbp:is_a_popular_spot_for |
evaluating translation quality
|
| gptkbp:is_a_time_for |
automatic evaluation of translations
evaluating machine translation |
| gptkbp:is_a_tool_for |
quality assessment
|
| gptkbp:is_designed_to |
machine translation systems
|
| gptkbp:is_essential_for |
all translation tasks
|
| gptkbp:is_evaluated_by |
comparing_n-grams
|
| gptkbp:is_part_of |
evaluation frameworks
automatic evaluation |
| gptkbp:is_popular_among |
human evaluation
|
| gptkbp:is_recognized_for |
gptkb:PyTorch
TensorFlow various software libraries NLTK SacreBLEU |
| gptkbp:is_used_in |
ROUGE
research papers academic research natural language processing translation models different translation systems machine translation competitions other evaluation metrics |
| gptkbp:isUsedFor |
other metrics
|
| gptkbp:language |
true
|
| gptkbp:measures |
machine translation quality
|
| gptkbp:performance |
can be automated
is often cited in literature is used in industry is widely recognized can be computed quickly can be used for large datasets can be used for small datasets is easy to implement provides a score |
| gptkbp:produces |
Papineni_et_al.
|
| gptkbp:range |
0 to 1
|
| gptkbp:related_to |
F1 score
|
| gptkbp:relatedTo |
reference translations
machine-generated text |
| gptkbp:sensors |
word order
|
| gptkbp:traditions |
machine learning evaluation
|
| gptkbp:uses |
n-gram precision
|