Datayê ma binê lîsansanê biçeşît yê Muşterekê Afirnayoxî de yeno neşrbîyene.Hîna zaf melumatî
Eke şima na muhtewa ra hes kenê, xeyrê xo seba yew bexişkerdişî bifikirîyê.
Menu
* Tatoeba Top 30 Languages For 14 Days *
Tatoeba Top 30 Languages For 14 Days Interactive Graph has been updated:
https://tatoeba.j-langtools.com.../index-14.html
Thank you, good job !
Thanks! :-)
@Sharptoothed
Many thanks.
I am really happy and proud that Kabyle is the first during these 14 days. I also participated to the score of French and English (thanks to @AlanF_US who helped me so many times) since I always translate my Kabyle sentences to make them visible, and with the goal of better comprehension with the other languages. I don't manage Russian as well. That's a pity! But I did my best.
Greetings
Some job has to be done to recruit more contribs for Kab. We are still late. We need more to launch other projects, MT among them. Isn't it?
@belkacem77
You are right.
Unfortunately, I am too old to learn a little more computing science than my out-of-date Apple Basic (of the 1980s).
Thanks a lot! :-)
@Sharptoothed
Tanemmirt sɣuṛ-i.
It's up to me to thank you.
C'est moi qui vous remercie.
Thank you.
Could you provide the statistics for proofreading ratios (both cumulative and individual)? I mean the ratio of sentences that are rated or tagged OK to the total number of sentences.
It's possible adding such a ratio to "Tatoeba Sentences & Translations Stats" chart but I'm not sure if it's valuable. For example, in Russian part of the corpus we usually don't mark sentences with OK tag unless they were added by a non-native. This means that only some of the sentences are tagged OK. I suspect the same is in other parts of the corpus too.
I previously calculated comment/sentence ratios of some languages and Russian was one of the highest ones, so I assume it's well-maintained and review marks may be trivial for you, but I think it's essential for the Turkish corpus. We have a lot of unnatural sentences added by native speakers. Monitoring the proofreading progress is therefore important and valuable for us, and the same may go for others. Having this information may also encourage users to rate/tag sentences which I think is beneficial for the overall quality of the Tatoeba corpus.
Anyway, I tried to work out the ratio for the Turkish corpus and left it as a comment on #8739923. I may update it from time to time to show the progress.
And here are the ratios for the top ten languages. (as of May 2nd)
English: 58.14%
Russian: 2.39%
Italian: 5.3%
Turkish: 3.06%
Esperanto: 23.6%
German: 2.81%
French: 1.62%
Portuguese: 3.96%
Spanish: 1.46%
Hungarian: 9.21%
* Sentences that are both tagged and rated OK, and sentences that are rated by multiple users are counted once.
* Sentences that have mixed rate marks (green-yellow or green-red) are not excluded.
* Self-approvals are not excluded.
OK, I'll try adding OK ratio to my chart.
Thanks a lot. :-)
Please, take a look :-)
https://tatoeba.j-langtools.com/transtop
Thank you very much. My OK ratios are a bit higher than yours. I guess yours include the OK-rated (green marked) sentences, but not the OK-tagged ones. I had tried to add them together and remove the duplicates to avoid counting both tagged and rated sentences twice.
I count both OK-tagged and OK-rated sentences but if a sentence has multiple OK-ratings, it counted only once. If a sentence has both OK-tag and OK-rating, it counted once as well.
According to your chart, 5948 Hungarian and only 3 Hebrew sentences are OK, but the advanced search shows many more sentences that are tagged OK.
https://tatoeba.org/eng/sentenc...&sort_reverse= (20,613 results)
https://tatoeba.org/eng/sentenc...&sort_reverse= (124 results)
It was a bug in my script. :-( Now the results should be better.
Everything seems fine now. Thank you for implementing that category.