5 ms·
I wanted to see how it would handle non-ascii text so I typed 안녕하세요 and it romanized it to annyeonghaseyo. (This is "hello" in Korean). I did not expect that.
by jbaudanza 6y ago
I wanted to see how it would handle non-ascii text so I typed 안녕하세요 and it romanized it to annyeonghaseyo. (This is "hello" in Korean). I did not expect that.
I really like the idea of a text-only social network, so I hope this takes off. I think the author should consider supporting other languages though. Seeing romanized text isn't very helpful to people that actually want to read or write in other character sets. It's a cute feature though.
- grishka 6y agoDoes the same thing with Russian as well. (I wonder how it would tell the difference between Russian, Ukrainian and Bulgarian though?)
- inopinatus 6y agoThere are other romanisations of Korean besides, although this is the government official one. For maximum confusion there are also cyrillizations, how would it go with that:)
- cinquemb 6y ago>>> import langdetect >>> probabilities = {x.lang: x.prob for x in langdetect.detect_langs(text) Works for me, for differentiating Russian and Ukrainian text.
- 0-_-0 6y agoPython strings have a method called casefold() that normalises text in a similar way (although it doesn't seem to work on Korean).
- blotter_paper 6y agoFrom the about section (emphasis added): >Subreply was created by Lucian Marin from the desire of a having a simple to use, English only, public forum that has nothing in common with ancient and untrustworthy social networks. [...] >Limitations >480 characters per reply ASCII only because it works everywhere I could see this being a xenophobic thing, but I could also see it being more about the limited character set (for minimalism). I'm not making a claim about the motivations, but either way your desire seems counter to his vision.
- unityByFreedom 6y agoWhere does Unicode not work? It’s been everywhere for years
- jchook 6y agoHe even has emojis for the "avatar"...
- aabbcc1241 6y agoBecause the content density varies in different languages. For example, each english word has ~7 ascii chars, but each chinese / japanese / korean words only has ~2 unicode chars
- unityByFreedom 6y agoThat doesn’t have anything to do with the tech being standard and easy to implement.
- jdck1326 6y agoIt could also be a moderation thing. If the people working on the website mostly just speak English, then it will be much easier for them to moderate an English-only website.
- alpb 6y agothat's xenophobic :)
- bugmen0t 6y agoNames of people, events, trends…
- waste_monk 6y ago> I think the author should consider supporting other languages though Maybe it's a moderation problem? I personally wouldn't feel comfortable running a non-english service, as I feel it would be too hard to moderate it effectively (or at least until you rearch the point where you have paid or volunteer moderators fluent in both the "main" language (to communicate rules and policy clearly) and any other supported langage). You can try auto-translating foreign text but you won't necessarily get a good translation, let alone handle cultural references (not Korean, but e.g. [1])and so on. [1] https://www.italki.com/article/92/the-dark-meaning-of-5-chinese-terms https://www.italki.com/article/92/the-dark-meaning-of-5-chin...
- techstrategist 6y agoI don’t think Lucian has needed to moderate yet, it’s been a small and positive community afaik.
- numpad0 6y agoI tried カレーライス, 高田馬場, ﾀﾃｲｽｶﾝﾅﾆﾗｾ, Llanfairpwllgwyngyll. Results were "kareraisu", "Gao Tian Ma Chang", and error message "Use spaces or shorter words" for the latter ones. Something funky is going on.
- gwbas1c 6y ago> I think the author should consider supporting other languages though Looks simple enough that someone could clone it to run "Subreply for Korean." IMO: I find mixed language social networks and threads difficult because I usually don't understand the other language.