Our data is released under various Creative Commons licenses.더 많은 정보
해당 콘텐츠를 좋아한다면 기부도 고려 부탁해요.
메뉴
I made a visualization of Tatoeba translations.
http://tguinard.github.io/
It shows how often two languages share a direct translation link, and you can view up to 50 of Tatoeba's most commonly translated languages.
Enjoy!
pretty interesting, thanks :)
Nice visualisation. ☺
Thank you for creating this, it’s amazing! I always wanted to do something like this, but had no idea about how to display it in a comprehensive way. I have two comments about it.
Sentences that are not translated into any language are not included, right? It could be cool if they were included, as a part of each arc, that would be not connected to anything.
I’d like to know technical details about the implementation, and the licence of your code. I’d love to have something like this included in tatoeba.org. We could probably even make it near-realtime since I think the data can be directly queried from the search engine.
Thanks, gillux and everyone else!
The graph displays translations, not sentences. If a sentence in language X is translated to every other language, this will count toward every edge in the graph originating from language X. Sentences that do not have any translations are completely ignored. I don't think there is a good way to incorporate these sentences as it is, as I am visualizing the links between sentences, not the sentences themselves.
As for the implementation, I have all of the code I used available on github.
https://github.com/tguinard/tatoeba_visualization (all code including data processing)
https://github.com/tguinard/tguinard.github.io (this is what is running in the link)
It isn't the most robust code; I kind of wrote it over the weekend, not expecting it to be used longterm. You are more than welcome to include this in tatoeba.org, and I can help with whatever integration you need. I don't have any license on it as of now, but if you need me to add any, I'll release this under any open source license.
> The graph displays translations, not sentences.
The links between the arcs are translations, but the arcs themselves are sentences, right? The bigger the arc, the more the sentences. Actually, strictly speaking: the bigger the arc, the more the sentences translated into other languages. My suggestion is to add a section in each arc that would not be connected to anything, and would represent sentences that are not translated into any language.
> The bigger the arc, the more the sentences
No, not exactly. Say sentence A is translated from English into 5 other languages, and sentence B is translated from English into 1 other language. Sentence A will account for 5 times the arc length (of English) as sentence B.