1. Justify demotion in principle: Look at the mnemosyne-proj data so
far (or even just your own data), and show (I suspect) that the pass
rate after a failure is unnecessarily high. This would justify
demotion rather than failure.
2. Just as the "initial ef" is autoadjusted based on the individual's
accumulated results, the global "demotion factor" could be set by
looking at the pass rate after demotion. For example, start with some
safe guess, such as /4, then adjust over time so that the cumulative
pass rate after demotion is, let's say, about the same as the
individual's overall pass rate, indicating that these cards are being
put back about where they belong.
3. Furthermore, optionally, once the above is done, some data will be
available that I don't think can be deduced from the accumulated
mnemosyne-proj data: For a given demotion factor look at pass-rate-
after-demotion versus length-of-interval-before-demotion. I suspect
that the longer the interval before the miss, the bigger the demotion
required. (Here is my reasoning for this guess if you want to see it:
Upon missing an old card, you receive a little reminder of something
you "know very well". You say to yourself: Oh yeah, I knew that! But
the *next time* you see this card, it might be the case (I experience
this) that the answer comes not by recovering the old (supposedly well
known) memory, but by remembering that little reminder, which is
*newer*.)
The simplest way to handle this would be with a discrete function: one
divisor for short interval, one (or more) for longer ones.

Conclusion: I've never trusted that "6" in the SM2 algorithm. *In the
context of the above*, what it is in effect doing is imposing this
policy, of more severe demotion for longer intervals, in the bluntest
possible way.
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"mnemosyne-proj-users" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/mnemosyne-proj-users?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to