In an effort to learn Mandarin, I started to use Anki. At the time the android app kept crashing. (Reviews seem to say it’s better now.) I also had a doubt that the current two variable algorithm is actually optimal.
So I set up a webpage with a mobile interface that lets you import decks and study using a variation of SuperMemo’s SM-2 algorithm. It has a small Gaussian randomness built into the easiness factor (decay constant). This might help determine if the algorithm should change.
I’m also worked on a way of sorting a language corpus. If you have sentences translated into another language, it’s not useful to have a long sentence presented, when you don’t know ANY of the vocabulary. Effectively, the algorithm runs through the items and sorts the items so that the challenge/novelty is fairly constant; the number of new words is distributed evenly over the entire set. As you might guess, this is iterative and still processor intensive… so I need to optimize it some more before letting web users run that function.
The other function still in the pipeline is a ‘remember on but not after’ deadline. I think this would only be useful for students, who need to remember something for testing, but can then forget it and use the internet if it ever comes up again.
http://www.superbrain.me looks useful. How does it differ from Anki—just that it’s web-based rather than needing to be installed? (I’m wondering how well it will work for those of us on slow connections.
I see from doing a site-search with Google that you can export to CSV—nice. I assume it’s possible to import the same way? And it looks like we can import from Anki—very nice.
In an effort to learn Mandarin, I started to use Anki. At the time the android app kept crashing. (Reviews seem to say it’s better now.) I also had a doubt that the current two variable algorithm is actually optimal.
So I set up a webpage with a mobile interface that lets you import decks and study using a variation of SuperMemo’s SM-2 algorithm. It has a small Gaussian randomness built into the easiness factor (decay constant). This might help determine if the algorithm should change.
I’m also worked on a way of sorting a language corpus. If you have sentences translated into another language, it’s not useful to have a long sentence presented, when you don’t know ANY of the vocabulary. Effectively, the algorithm runs through the items and sorts the items so that the challenge/novelty is fairly constant; the number of new words is distributed evenly over the entire set. As you might guess, this is iterative and still processor intensive… so I need to optimize it some more before letting web users run that function.
The other function still in the pipeline is a ‘remember on but not after’ deadline. I think this would only be useful for students, who need to remember something for testing, but can then forget it and use the internet if it ever comes up again.
Anyway, if interested: http://www.superbrain.me
http://www.superbrain.me looks useful. How does it differ from Anki—just that it’s web-based rather than needing to be installed? (I’m wondering how well it will work for those of us on slow connections.
I see from doing a site-search with Google that you can export to CSV—nice. I assume it’s possible to import the same way? And it looks like we can import from Anki—very nice.