this exactly. Parallel texts may seem like the strange inclusions, but they are still published with the learner in mind
I think where it can get confusing is the line between textbooks & graded readers (some textbooks try to teach through stories… which can make things confusing). Generally, I’ve just thought if the vast majority of the content is story rather than notes, then it can fall in graded readers. Also, if the thrust of the book is to understand stories rather than teach grammar or something like that, then it probably falls in graded readers.
Nothing in particular I can’t sort out with your answers! The book advertised parallel text for learners, but I just wanted to double-check before I submitted anything. I feel like it’d be really disappointing to find a new textbook/graded reader, buy it, open it up, and find out the person who submitted it mis-categorized it. So just being overly cautious, I think!
I’m sorry, I’ve got another question: story type “Short Story”. I’m guessing this is for short story collections and not literally a short story, since those aren’t typically published stand-alone (that I’m aware of)?
Most of the Aozora short stories are actually stand alone short stories available by themselves, but collections of short stories also go in that category. You can usually tell them apart by page length
The blurry bit comes when some things are lightly connected short stories vs wholly unconnected. I personally put connected short stories into the novel category and unconnected into short story.
Just noticed that my selection for the activity feed - e.g. “Following” - always gets reset back to “Global” when I open a book page and go back to the dashboard. Doesn’t seem to happen with other parts of natively. Is that just me? Is that intended?
Guessing this has probably been discussed before, but I couldn’t find it in a quick search.
Why doesn’t natively automatically grade your books once it has enough information?
If you have three books - A, B and C - and you grade C > B and B > A, why does it still ask you for the relation between C and A? Shouldn’t it just grade it as C > A automatically?
Seems like it would make grading a lot easier for people that have a lot of books of similar difficulties
That’s a good question and potentially something to think about in the future. Granted, you could make the case that grades are inherently a little noisy, so leveraging this way could exacerbate that noise. However, I think you’re probably right… i think it makes sense to do it at some point
You should find out how often gradings are inconsistent, where the user has contradicted themselves. I bet it happens fairly often. And if it does happen, getting that information could be useful since the contradictory gradings will cancel out (at least somewhat I would expect), negating the user’s (apparent) lack of ability to grade those particular books accurately.
Yes if i went ahead and did this I’d have to build a ‘implied grading’ calculator which you could probably use to view inconsistent grading and see if it truly makes sense to use this leveraging.
TBH, since people usually always finish their available grading, you could make the case too that it’s not worth it to implement yet. It’s a good thought though and makes intuivitive sense.
Yeah, it makes sense. I just know I’ve contradicted my own gradings before, which made me wonder how often that happens and what impacts it might have. (Also, as you know, I love grading the books, so I’m happy to do extra comparisons, even if they are redundant.)
Doesn’t that depend on the accuracy and precision of the data, how far apart the automatically graded books are in perceived grade and how big the contradiction is?
E.g. a user has five books he perceived as A<B<C<D<E, the user has filled in everything but A<E. Since the distance in grade should be fairly large relative to noise, automatically grading shouldn’t propagate error, but mistakingly grading A ≈ E would introduce new error.
I wonder if the same could be said about the C>B>A example, since having an entire book between C and A might already be enough that noise doesn’t impact C>A. Would be interesting to see the math on this
It does seem intuitive though that for books that are fairly close in perceived grade having redundant information would reduce the impact of noise.
Potentially! Maybe longer links could resolve things, but the longer the links you require, the more complicated it would be to calculate (i think) and also the less beneficial it’d be.
TBH, I’d probably be more inclined to surface inconsistencies to users first or try to make a personal difficulty ordering, purely for personal evaluation before having it impact actual gradings generated. Prioritizing the display of inconsistencies first would make the situation a lot more clear
But, yeah like I said, users seem happy to finish their gradings so probably won’t do for a while… as it’s simpler for evaluation and for an algorithm
Not sure if this is just my browser, but if you update the reading percentage for a book from your dashboard, it looks like you need to refresh the page again after the popup closes out for it to actually show the percentage on the dashboard.
I’m on Chrome 104.0.5112.102 (64-bit) on Windows 10.
Hey, thanks for the site and all its functionalities. It’s really useful, I’m finding tons of stuff at my level and reading them confirms the sites rating’s are pretty accurate!
One question, is there a way to filter series by length? I like to read manga so I would like to search like “N3 level between 5 and 10 volumes” or something.
I don’t think there’s a way right now, but that sounds pretty useful! I made a similar feature request if you want to upvote that. If you want specifically a filter and not sorting as I suggested, you might need to make a new request.
Also just tried again on both Firefox and Chrome and it seems to work just fine without refreshing, so maybe was a one-time weirdness with Chrome or something…
Thanks for the help!
