Imagine surfing Wikipedia, reading an interesting fact you’d like to remember, and simply clicking a button next to it that added that fact to your daily spaced repetition learning database.
Because WP is unstructured natural language, this is impossible to automate completely. It might be possible to automatically generate flashcards from something more structured like the entries in Freebase, but would you want to learn them?
If you are willing to do a little work, Supermemo has an idea of structured note-taking/editing, called incremental reading. SM has the most support for it, but really it’s just copying the text into a text editor and expecting you to cut it down to notes and then to flashcards.
Also, while there are good spaced repetition programs/databases out there (like Anki), the databases aren’t collaborative like a Wiki, and in general could be much better.
Agreed. If I were to tackle this, I would start with the Anki folks (Mnemosyne is too slow and conservative), and I would do something like write an Anki plugin to scrape a page inside a Category on a wiki (probably the card contents would be formatted using tables or templates), and then everyone just edits the page and any future downloads get the modified version. This wouldn’t handle existing users getting updates, but that’s the general problem of distributed revision control, which is certifiably Hard.
Actually, collaborative abilities aside, I think even a very well kept but closed database could be quite valuable (more so than Rosetta Stone or other such software for learning a language for instance).
I think all the commercial providers of flashcards for SM have failed. Apparently it isn’t that valuable.
Because WP is unstructured natural language, this is impossible to automate completely.
I agree. What I intended was a markup language that editors could use. Then somebody could surf Wikipedia with some plugin installed so that marked facts would be highlighted in some way. Clicking on the fact would then bring up a box showing the relevant note card (or multiple variants), and you could select to add it to your database.
I think all the commercial providers of flashcards for SM have failed. Apparently it isn’t that valuable.
Links? I’m more keen on the collaborative idea personally, but I haven’t totally ruled out a commercial (curated) angle. I’m pretty stingy myself, but I would definitely pay for a high quality SR deck. Maybe I’m unusual though.
My theory is that hand-made cards are good enough and often better than purchased ones; it’s even harder for commercial ones to compete with the best freely-distributed ones. (Take a look at the samples for a number of the decks, like the assembler deck. My own assembler flash cards are better!)
The quality doesn’t seem particularly great for the SM bought decks. For instance, for a foreign language deck I’d like to see images and native pronunciations, preferably from multiple speakers. For a chemistry deck I’d like to see 3D representations of molecules, preferably in multiple manners (ball-and-stick, spheres, etc), even better if the flash card allowed for rotational viewing of the molecule.
I definitely wouldn’t buy any of the decks I saw on SM.
There are various plans to finally implement some sort of Wikidata—think “infoboxes on steroids” as an absolute minimum, and then unknown possibilities we haven’t even thought of yet. Erik Moller’s been asking around the Wikimedia Commons and technical lists for ideas of what we actually want in terms of a specification.
Because WP is unstructured natural language, this is impossible to automate completely. It might be possible to automatically generate flashcards from something more structured like the entries in Freebase, but would you want to learn them?
If you are willing to do a little work, Supermemo has an idea of structured note-taking/editing, called incremental reading. SM has the most support for it, but really it’s just copying the text into a text editor and expecting you to cut it down to notes and then to flashcards.
Agreed. If I were to tackle this, I would start with the Anki folks (Mnemosyne is too slow and conservative), and I would do something like write an Anki plugin to scrape a page inside a Category on a wiki (probably the card contents would be formatted using tables or templates), and then everyone just edits the page and any future downloads get the modified version. This wouldn’t handle existing users getting updates, but that’s the general problem of distributed revision control, which is certifiably Hard.
I think all the commercial providers of flashcards for SM have failed. Apparently it isn’t that valuable.
I agree. What I intended was a markup language that editors could use. Then somebody could surf Wikipedia with some plugin installed so that marked facts would be highlighted in some way. Clicking on the fact would then bring up a box showing the relevant note card (or multiple variants), and you could select to add it to your database.
Links? I’m more keen on the collaborative idea personally, but I haven’t totally ruled out a commercial (curated) angle. I’m pretty stingy myself, but I would definitely pay for a high quality SR deck. Maybe I’m unusual though.
Currently the only surviving seller of SM flashcards that I know of is SM itself: http://www.super-memo.com/collections.html
And if you look at their recent-addition list (http://www.super-memory.com/sml/accession.htm), #1 is http://www.super-memory.com/sml/2003/651.htm which was last updated… 2004.
My theory is that hand-made cards are good enough and often better than purchased ones; it’s even harder for commercial ones to compete with the best freely-distributed ones. (Take a look at the samples for a number of the decks, like the assembler deck. My own assembler flash cards are better!)
The quality doesn’t seem particularly great for the SM bought decks. For instance, for a foreign language deck I’d like to see images and native pronunciations, preferably from multiple speakers. For a chemistry deck I’d like to see 3D representations of molecules, preferably in multiple manners (ball-and-stick, spheres, etc), even better if the flash card allowed for rotational viewing of the molecule.
I definitely wouldn’t buy any of the decks I saw on SM.
There are various plans to finally implement some sort of Wikidata—think “infoboxes on steroids” as an absolute minimum, and then unknown possibilities we haven’t even thought of yet. Erik Moller’s been asking around the Wikimedia Commons and technical lists for ideas of what we actually want in terms of a specification.
That sounds like a great step in the right direction! I’d love to hear more about this as it evolves.