The website and the API report difference sentence counts for CCMatrix

Open gregtatum opened this issue 1 year ago • 1 comments

https://opus.nlpl.eu/opusapi/?source=en&target=ru&preprocessing=moses&version=latest

[
  ...
  {
    "alignment_pairs": 139937785,
    "corpus": "CCMatrix",
    ...
    "version": "v1"
  },
  {
    "alignment_pairs": 139937785,
    "corpus": "NLLB",
    ...
    "version": "v1"
  }
]

Then from: https://opus.nlpl.eu/

corpus	doc's	sent's	en tokens	ru tokens	XCES/XML	raw	TMX	Moses	mono	raw	ud	alg	dic	freq			other files
NLLB v1	1	139.9M	2.7G	2.5G	xces en ru	en ru	tmx	moses	en ru	en ru				en ru		sample
CCMatrix v1	1	35.1M	581.4M	527.5M	xces en ru	en ru	tmx	moses	en ru	en ru				en ru		sample

Jan 16 '24 20:01 gregtatum

Yes, something seems wrong here. Better look at the re-designed website: https://opus.nlpl.eu/CCMatrix/en&ru/v1/CCMatrix That should be correct I hope.

Jul 17 '24 20:07 jorgtied