Nos données sont distribuées sous différentes licences Creative Commons.Plus d'informations
Si vous aimez ce contenu, pensez à faire une donation.
Mur (7 152 sujets)
Astuces
Avant de poser une question, soyez sûr d'avoir lu la FAQ.
Nous cherchons à maintenir une ambiance amicale pour des discussions civilisées. Veuillez lire nos règles contre les mauvais comportements.
sacredceltic
il y a 7 heures
sacredceltic
il y a 14 heures
AlanF_US
il y a 15 heures
sacredceltic
il y a 16 heures
lbdx
il y a 20 heures
Seael
hier
sacredceltic
hier
frpzzd
hier
sacredceltic
hier
frpzzd
hier
Ça m'exaspère de passer mon temps à éviter les phrases contenant "Tom". Je n'en peux plus. Pourrait-on envisager un flag dans le profil, d'évitement des phrases contenant "Tom" ? Voire d'autres mots-clés ?
Ça m'éviterait de sur-consommer de l'énergie et de causer un excès de réchauffement de la planète, car je dois, à chaque fois, passer à la phrase suivante, car je n'ai AUCUNE intention de traduire de telles phrases.
This page on the Tatoeba wiki gives examples of some more advanced search queries that are possible, besides what is listed in the Advanced Search interface:
https://en.wiki.tatoeba.org/art...ow/text-search
For instance, it shows how you can perform searches for sentences omitting certain words, such as "Tom":
https://tatoeba.org/en/sentence...rom=eng&to=und
Does this help you?
No it doesn't. it's too complicated because you must set it up each time. it should be a profile's FEATURE .
My idea is, each contributor should be able to dress a list of UNDESIRED keywords. Based on it, the display of sentences should be PERMANENTLY filtered accordingly.
Well, I agree that it's very inconvenient to fill out the advanced search menu each time you want to perform a common search. However, you can also save advanced searches by copying the URL (and deleting the "rand_seed" argument, if you want it to re-randomize each time) and saving it for later. For example, I have several query URLs saved in my user bio for easy access, so that I can just copy-paste them instead of having to fill out the Advanced Search form.
Some sort of "global filter" would be nice too, though. A related feature has been requested in this Github issue, which asks about globally hiding sentences by specific users:
https://github.com/Tatoeba/tatoeba2/issues/3099
I'm still unconvinced...
I also long for a bit more Tomless Tatoeba so I tried the how-to-exclude examples contained on https://en.wiki.tatoeba.org/art...w/text-search# but they don't seem to work:
https://en.wiki.tatoeba.org/%20...-small+the&to=
https://en.wiki.tatoeba.org/%20...C*%7Cz%5C*&to=
For some reason, a space was inserted into those links. I've fixed them on the wiki page. Here are the correct links:
https://tatoeba.org/en/sentence...-small+the&to=
https://tatoeba.org/en/sentence...C*%7Cz%5C*&to=
It's very unnatural.
it should be much easier for users to exclude words and names.
Why not make a user's list of undesired words and expressions available?
There are lists such as "Rebalanced English ✂️" that prohibit a word from occurring more than 10 times as often as in a reference corpus.
You can browse these lists at https://tatoeba.org/en/sentence...direction=desc
That’s not the solution since Tom appears across languages.
Well I guess we’ll have to spoil more data-center fuel to dodge these sentences…
Je viens d'accéder à ta liste "rebalanced French" et c'est très curieux:
1) la majorité des phrases sont écrites par des non-natifs.
2) il y a plein de fautes, sans doute conséquence de 1)
Je propose d'élire l'ennemi numéro 1 de l'espéranto sur ce site.
Qui, selon vous, est le "bigot" qui dissuade le plus les utilisateurs de contribuer dans cette langue ?
Si vous avez peur de le dénoncer publiquement, envoyez-moi votre réponse par message privé.
Korpuspflegerkandidat für Toki Pona
Corpus Maintainer Candidate for Toki Pona
Kandidato por la bontenado de la tokipona korpuso
Candidat pour la maintenance du corpus Toki Pona
Кандидат в ответственные за корпус для токипоны
Eugene (EugeneGS): https://tatoeba.org/user/profile/EugeneGS
🇩🇪 Schickt uns wie immer gerne eine Privatnachricht, um uns eure Meinung mitzuteilen (auf den Link unten klicken).
🇬🇧 As usual, please feel free to send us a private message to share your opinion (click on the link below).
🟩 Kiel kutime, ne hezitu sendi al ni privatan mesaĝon pri via ĉi-rilata opinio (alklaku la ĉi-suban ligilon).
🇫🇷 Comme d’habitude, n’hésitez pas à nous envoyer un message privé pour nous faire part de votre opinion si vous le souhaitez (cliquez sur le lien ci-dessous).
🇷🇺 Вы можете, как обычно, отправить нам личное сообщение, чтобы поделиться своим мнением (кликните по ссылке ниже).
https://tatoeba.org/private_mes...rsichbaeumchen
Le contenu de ce message va à l'encontre de nos règles et a donc été caché. Il est seulement montré aux administrateurs et à l'auteur du message.
✹✹ Stats & Graphs ✹✹
Tatoeba Stats, Graphs & Charts have been updated:
https://tatoeba.j-langtools.com/allstats/
Thank you! ⵜⴰⵏⵎⵎⵉⵔⵜ!
You're welcome! ⴰⵏⵙⵓⴼ ⵢⴻⵙⵙⵯⴽ! :-)
Heureux anniversaire, Lisa ! 🎂
Besser spät als nie. Herzlichen Glückwunsch zum Geburtstag, Lisa!
Vielen Dank! 😊
Auch Dir alles Gute zum Geburtstag, Christian! 🎉🎊🎈🎂
Je me joins aux félicitations !
Le contenu de ce message va à l'encontre de nos règles et a donc été caché. Il est seulement montré aux administrateurs et à l'auteur du message.
🤣
Hello, Tatoeba was updated today. What’s new?
- Some listings have been limited to display the last 1000 sentences: the list of sentences having audio, the list of sentences in user-created lists, and the list of sentences having a specific tag. Background on this new limitation: https://github.com/Tatoeba/tatoeba2/pull/3199
- It is now possible to use @ immediately followed by a username to notify a member in Wall posts. It was already possible to do so in sentence comments, but not in Wall posts. Now you can do it in both.
- The API served at https://api.tatoeba.org was improved. While it is still a work in progress, it is now close to become stable. Technical details: https://github.com/Tatoeba/tatoeba2/pull/3201
Interface improvements:
- Error pages (such as 404 error) can now be localized, thanks to the contribution of SrulyP: https://github.com/Tatoeba/tatoeba2/pull/3198 The error pages can be localized just like any other part of the website, for how to contribute, please look at https://en.wiki.tatoeba.org/art...ce-translation
- For sentences in Chinese, the converted simplified/traditional script is now displayed correctly. Details: https://github.com/Tatoeba/tatoeba2/issues/3188
- 52 language icons should look nicer. They have been vectorized, which means they won’t look pixelated no matter how much you zoom in. A total of 153 out of 429 language icons (35%) are now vectorized. Today, it is my honor to publish the vectorized brazilian/portuguese flag of our deeply regretted fellow member Ricardo.
Le contenu de ce message va à l'encontre de nos règles et a donc été caché. Il est seulement montré aux administrateurs et à l'auteur du message.
Le contenu de ce message va à l'encontre de nos règles et a donc été caché. Il est seulement montré aux administrateurs et à l'auteur du message.
Hey all! Wanted to share that recently I wrote some scripts that help find sentences to translate on based on lists of vocab. Taking as input the Tatoeba corpus and a tiered list of vocab (e.g. lists of words ranked by the chapters in which they appear in a textbook) as well as some limited grammar concepts (cases, parts of speech, verb forms, etc), it generates tiered lists of sentences such that the sentences at level N only use the vocab/grammar at level N or lower.
I've been working through a Russian textbook and using it to find sentences here that appropriate for attempting to translate into English/Spanish based on the chapter I'm on. My scripts use the spaCy Python library to rank sentences by chapter, and this library is sometimes flawed in terms of lemmatization and grammar annotation, but so far it has been a huge help for me. (Of course, my translation attempts can also be very flawed, as the Russian natives here know... thank you marafon, Ooneykcall and EugeneGS for your patience! :-) )
As an example, here are some sentences (as of the current data dumps) at chapter 10 level based on my book:
1006495 Да, у меня есть дочь.
1006500 Вы изучаете английский.
10074251 У нас всё свежее.
10101554 Вода слишком холодная.
10110989 Кофе слишком горячий.
And here are some at chapter 20 level:
10031177 Интересно, пойдёт ли снег.
10031179 Интересно, будет ли снег.
10038848 Ничего странного я в этом не вижу.
10043008 Я ушёл домой пешком.
10055872 Он бежал в школу.
Just wanted to share this info here in case any other learners would find this helpful for their own studies. Whether you intend to translate sentences found this way or not, it can be helpful for finding lots of example sentences at your level when working through a textbook. If this sounds useful to you, feel free to reply / send me a message and I'll be happy to try generating ranked translated/untranslated sentence lists for your choice of language too.
That’s amazing, Franklin, thank you for telling us. As I understand, you are making a curated list of Tatoeba sentences organized by textbook level. Do you plan to make your list available on Tatoeba or elsewhere?
That is correct!
The sentences are selected programmatically, and I could make them available as lists on Tatoeba (as I have a list of sentence IDs comprising the lists), but I'm not aware of a feature on the site allowing users to form lists other than by manually clicking to add each sentence. Right now, the Russian sentences ranked by my textbook number over 100,000 so of course I cannot add them manually this way. As far as I understand, the Tatoeba API is also read-only for the moment. Is there a way to do form lists by uploading (long) lists of sentence IDs that I'm not aware of?
In any case, I may post some of the lists on Github corresponding to my particular textbook. But I wanted to mention it here in case anyone is similarly working through another textbook for any other language and could also make use of ranked sentences. I am currently only using this for Russian but it would be easy to do for many other languages, provided a vocab list.
Hello, Franklin! I'm not sure if there's currently a way to upload lists directly on Tatoeba, but I once made a small program for myself that can automatically add sentences to a list by their IDs. If you'd like, you could share your sentence ID lists somewhere (for example on GitHub), and I could help import them into Tatoeba lists. (By the way, the program is rather slow — in my testing it processes about a thousand sentences per hour, so a large list would take some time.)
Hi Eugene, that would be fantastic! Here's a link to a TSV file in which the first column contains IDs of Tatoeba sentences in Russian, and the second column contains the "rankings", which are numbers 3-29 corresponding to chapters in the book. Ideally each chapter would get its own list.
https://gist.githubusercontent...._sentences.tsv
I haven't even added all of the book's vocab into my input data yet, since that is a slow-going process. For that reason, from chapter 24 onwards, ranking is based purely on grammar (as estimated by spaCy) and not by vocab for now.
Okay! I've started it. It will take a few days (around 4–6, maybe a bit more). Here is the link to all the lists:
https://gist.github.com/kilsens...4a27b9658e5e55
I'll update it with new lists as they appear. If you need, I can also rename the lists.
Thank you very much, this is fantastic! This will be a huge help. Previously I was copying over the IDs into the URL, which was a very clunky workflow.