some news / ideas

Hi all,

Some informations / news / ideas in a short version:

- Carsten Niehaus, from KDE-edu team, joined us yesterday.
He helps with KMemAid, but as far as he told me - might be that he won't have 
spare time for KMemAid after this weekend. :-(
Anyways, I hope he will help us, even with hints what should be changed,  etc. 
with getting KMemAid to KDE. :)

- My MemAid UI versions - nothing serious happens here, sorry.
No revolutionary changes, I don't have so much spare time now, but I have 
experimented a lot anyways - with Fox Toolkits versions, also made a GraphApp 
version of GUI, and I plan to make (an additional) MemAid GUI based on GTK.
More or less - I am still learning GUI programming. It takes me time - but it 
means that in the future I will be able to make better GUIs that I am now.

In the meantime: I ask all people having access to CVS (but not anonymous one! 
cvs for 'anonymous' is very lagged on sourceforge since some time - SF has 
some problems that should be fixed this/next week) that use my MemAid GUI to 
try version from CVS and report any problems - if there are any.
I attached a simple script that should download itself all needed files.
There are no binaries in CVS - you have to compile MemAid by yourself. (BTW: 
Please note that current version needs Fox Toolkit in stable version: 1.0.)

Ideas / to-do:

- Categories.
I need to know what would be the best implementation of categories from GUI 
programmer point of view (Klemens? Carsten? Or anybody who knows. Any 
hints?).
I think my simple idea:
	Every item has a number - category.
	Every number correspond to a category described by simple string
	e.g. char *(int category); would return pointer to category description.
won't be bad, but I am still not so sure about it and about some details of 
storing it on files - I simply haven't though about it seriously for more 
time. And I should look on some widgets (FXListTree in case of Fox Toolkit) 
and see how implementation would look like.

More ideas - once we will have 'categories':

- Difficulty.
Every category could have a difficulty. Let's say: "easy", "normal", 
"difficult". All items would have it's own difficulty variable, but would 
have by default difficulty of the category they are in.
This idea has obvious advantages (one can set "easy" difficulty for topic/
category that is easy - and these elements would have higher intervals, as 
they are not needed to be reviewed so frequently.)
But it has also a disadvantage:
yet another neuron in ANN, not trained ANN to this, change of format, etc. - 
mess in code.
But I think this idea will be worth implementing - but first we should have 
categories. (and trained ANN - current shipped with core ANN is not well 
trained to use all currently implemented features - we shouldn't move so 
fast, I will need to collect user_data.bin files and train ANN using them 
before going to this stage - implementing "difficulty".)

- Speed of learning.
Ah... here I just wanted to say something about my fresh idea - variable 
"speed of learning" (AKA "Forgetting Index", that can be set up from 80% to 
99% IIRC, in SuperMemo), but after some thinking I see that... I should think 
about it more before speaking out about this one. ;-)


-- 
David Calinski
http://neodave.civ.pl
/* FreeBSD: When we say free, we mean it. */

#!/bin/sh

cvs -d:ext:YOUR_SOURCEFORGE_LOGIN_HERE-SzB4B6P5alysTnJN9+BGXg@public.gmane.org:/cvsroot/memaid checkout memaid
cd memaid/src/
cvs -d:ext:YOUR_SOURCEFORGE_LOGIN_HERE-SzB4B6P5alysTnJN9+BGXg@public.gmane.org:/cvsroot/memaid checkout core