掲示板 Forums - A couple of strange things happened

Yesterday, it was the 2nd or 3rd day that I had switched my kanji schedule to slow and thorough, and suddenly I got many questions about the trickiest aspects of kanjis. Mostly, it's about the kun-yomi of kanjis that do not have any common words with that reading. Or even no word at all that I can find, such as かけ。る for 県. That's far beyond what most people should learn. I'm guessing it's a word that disappeared from common use centuries ago? That's the most extreme case I got, but at least a dozen other questions were way beyond what I could reasonably expect to learn, with no common word using the rare reading that was asked. The kun yomi of 権 is maybe worse than the kun yomi if 県, because it's 3 things to remember (omori, kari, haka.ri), all that for an alternative spelling of はかりどの, which is a "temporary shrine (houses the object in which the deity resides when main shrine is under repairs)". Most Japanese people probably don't know this word, or the kun yomi of 県 or 権 (after high school), I guess.

In case it was related to the setting, I switched back to quick and simple, but it didn't seem to help, today, also it's strange that it didn't start the first day of me using this setting. Maybe it's due to a change on the site, around the word merging process?
The second strange thing was this morning, when I opened the main page it said that all my schedules were fully studied. Of course when I opened each schedule, it appeared I had about 130 vocab and 40 kanjis to review. That was odd.
Then a 3rd strange thing happened during review... which I forgot while describing the first 2. Let's see if it comes again...

1. Go to your site settings > Kanji > what does your "hide rare readings" setting set to?
2. I've heard this from others from time to time. So, if you go to your dashboard, everything is marked as finished? Only when you click in does it change? What if you refresh the dashboard after hitting it the first time?

Unfortunately, I'm not sure how to debug this unless I can see it myself (in which case, it should be fairly easy). If you posted this around the same time, then your morning is around my 4pm. If you can replicate this, then *not* study or go to the schedule page and send me an email (michael@renshuu.org), I can try and jump in and check it out. Should I not respond within a couple of hours, then you can go ahead and continue to study.

3. Do you remember what the sentence said/was?

Also, when one votes down a sentence, it seems it disappears forever. Could it instead be thrown to the bottom of the list and greyed out? Currently, if someone votes down a sentence because it's way too complicated, it will be still be gone 5 years later when the user speaks fluent Japanese?
(there are many, many sentences that I'd gladly push down to the bottom because they're way too complicated for me, but I don't precisely because of this, and because some of them are the only examples of some words)

Thanks for clarifying what's to know for the kanken.
All normal with the schedules today. The problem yesterday could have been related to the OS update that had just happened? I don't see how, but maybe you do.
Schrödinger's sentence above still seems half correctly parsed, half broken...
Another strange thing right now: I got quizzed for the on-yomi of 玉. I didn't know it, so I looked below the answer for the words in gyoku. None were displayed, although 玉（ぎょく), 珠玉 and 玉砕 are all common words, which should be displayed, according to my settings. I see a total of 14 words, the maximum (in the settings) is 15. Although 玉 (ぎょく) is missing, it's probably not an issue related to words merging because one of the words listed under 玉 is 玉子, which is an alternate of 卵.

Even better (worse? I can't tell... I think it's good because if you solve this you could have thousands of sentences newly linked to thousands of words... ): although the word dictionary shows 0 sentences with 忠実(ちゅうじつ), a sentence search shows 38 occurrences, with 忠実 being correctly parsed in 18 of them. Maybe the clue is that in the sentences where 忠実 is parsed, only 1 has it as まめ, it's ちゅうじつ in the 17 others. The sentence with 忠実（まめ）is correctly attributed as the only exemple sentence for this word. It's "彼女って忠実だね。"
Maybe, somehow, the association between that sentence and the word 忠実 (まめ）blocks the search of sentences containing 忠実（ちゅうじつ）？
The problem is not in the sentences:
あの 翻訳は 原 文 に 忠 実 だ そうだ 
appears as an example sentence of 原文, just not as an example sentence of 忠実.

.

と言いました。

There's that too!... Also, many sentences that don't appear for a word, because in the sentence it has a suffix like 権　度　的　語...
I don't have a good example in mind, but my private sentence with 加速度 doesn't appear as an example sentence for 加速. For some words it's a problem because they're taught as JLPT words of their own, but in fact are always used with a suffix. Japanese being an agglutinative language, any sentence that is an example sentence for 加速度、加速度計、加速度器、加速運動、加速度原理、加速度時間... is also an example sentence for 加速 (and 時間　原理　運動...). I guess that would be pretty difficult to take it into account? You'd have to teach the site the concept of agglutination, that a Word can be simultaneously 1 and 3... (So fascinating, maybe one could use this concept as the basis for a religion?). Maybe the mystery of the 17 half-parsed/ half-unparsed 忠実 sentences is a good start to improve this part of the site :o)

1 and 2: Nicely done!
It's maybe even more work than that, because it's not only one-kanji suffixes. Often it's 4-kanji words made of two 2-kanji words.
Such as 民主主義 which has 32 examples while 民主 has 0, although it's supposed to be a top-500 words, a top-12000 word in newspapers and a JLPT word. And sure enough, if you search for examples anywhere, the word 民主 appears as part of a bigger word: democratization, democratic party, and 民主主義 which makes using 民主 alone almost impossible (completely impossible?). jisho manages to show sentences containing 民主主義 and the other 民主* words as examples for 民主.
You also have other forms, like the ~180 words in *がない. Most of them are compound more set expressions, or compound words, like 腕が無い, 能がない etc etc, which are examples of 腕, 能 etc.
Or words like the dozen of words starting in 加速* (there's one too many in the list, 加速度センサー is given as かそくどセンサー and then as かそくどセンサ).
I can't imagine how this could be completely automatised. But you could have a very good start if you can make a list of all the 4-kanji words ABCD words in which the first half AB and the second half CD are both words by themselves, because then it should be rather quick to confirm (crowd-sourced?) whether or not the word is a compound of the first half and the second half, i.e. if AB and CD can both be considered to be present by themselves whenever ABCD is used in a sentence (as the words 民主 and 主義 are both present in any sentence containing 民主主義).

I just wrote the recap of the 4 issues here that are not yet resolved, but it crashed when I added a second screenshot, so I'll start again, without the screenshots, and from my comp on which the Japanese keyboard doesn't work :/

1-I don't get asked new rare kanji readings, but this morning I was asked for the ka.keru reading of ken (prefecture), among others. Is it possible to stop all questions about rare readings, once the "hide rare readings" options is turned back on?

2-From my list "tautologicals" I can still see only 3 examples for jiko (oneself), out of the 22 that I can see from the dictionary (and much more from the sentence search, but most are in the compound word "jikoshoukai"). It's the only word on the page with this problem, and jiko is the 50th and last of the page, so that must be where the problem is from.

3-Previously there was the problem that rare or obsolete readings (mame instead of chuujitsu and tokihakari instead of tokei) were taking example sentences away from the usual reading. Now there's the opposite reading: If someone searches tokihakari in the dictionary, he will have no indication that it's an obsolete reading, no connection to the usual word tokei, and no example sentence. For words that are written the same as more common words (marked "common" and/or with example sentences), could there be a link "see also tokei"?

4-chuujitsu now has 18 example sentences, much better than the previous 0, but there are in fact 38 sentences with that word (and that's not because they're in compound words, they're just not parsed)

1-Oh, so that's what this button is for! That will fix the problem within a few days, thanks! Solved
2-Hm... everything is fine. Solved
Maybe I started to have hallucinations. Yesterday or the day before I thought I saw filters in the kanji dictionary: by class, key, learning status... all gone when I reloaded the page.
3-There's no rush... it will be an improvement whenever you get to it. Coming soon(-ish)!
4-Sure, I'd gladly fix some sentences here and there if you add this button. Not thousands, but as we say in French, little streams make big rivers. Solv...ing

I saw the button, but here's an issue I hadn't mentioned: When I enter a private sentence, the parsing doesn't load. It's the same here, see this screenshot. Waiting several minutes doesn't help.


So I can't use it, yet.