Learning Algorithm customization.

[GoogleCodeExporter]

What is the feature you want?
We want to customize the learning algorithm.

How important is it to you?
Sometimes people think the learning period is too long or too short. After 9.0, 
AnyMemo support the decimal scheduling period and separate the algorithm logic 
in code. So it is easier to do algorithm customization.

Do you think other people also want this feature and why?
There are some request about this.
Star this issue if you want to get it prioritized.

Original issue reported on code.google.com by mrlhwlib...@gmail.com on 31 Mar 2012 at 12:45

[GoogleCodeExporter]

Yes, it will be great.

Original comment by mchin...@gmail.com on 31 Mar 2012 at 2:17

[GoogleCodeExporter]

YES, I need it~

Original comment by Joshuap...@gmail.com on 1 Apr 2012 at 2:20

[GoogleCodeExporter]

Yes, it's exactly what I need - thanks!!

Original comment by bernie.c...@gmail.com on 9 Apr 2012 at 3:59

[GoogleCodeExporter]

Question for you guys: Do you need the algorithm be global (like global 
options) or local to each database (like db settings)? 

This is an important question needs to be answered and will determined how I 
implement it. You can't choose both for now.

Original comment by mrlhwlib...@gmail.com on 9 Apr 2012 at 5:56

[GoogleCodeExporter]

都可以……

Original comment by Joshuap...@gmail.com on 9 Apr 2012 at 6:28

[GoogleCodeExporter]

Global is enough for me.
But you can make it like DB settings.
(Every DB has its own settings and global default settings).
You can do default algoritm, that I can change.
Every new DB inherites this algoritm, but I can change it as I want for every 
DB.

Original comment by mchin...@gmail.com on 11 Apr 2012 at 11:04

[GoogleCodeExporter]

As I have said, you can't have both just like "Settings". The main deal is that 
the option will lose if you uninstall AnyMemo. But it is much much easier to 
code for me to use global option and the transition can be done without the 
database format upgrading.

Original comment by mrlhwlib...@gmail.com on 11 Apr 2012 at 4:52

[GoogleCodeExporter]

No needs for me to edit algorithm for every DB.
So I vote 1 global algorithm.
(Maybe you'd better make voting on main anymemo page.)

Original comment by mchin...@gmail.com on 12 Apr 2012 at 2:47

[GoogleCodeExporter]

I would really appreciate this feature.

I think the SRS numbers are a bit off though. For example on one word, the 
ranges are 

0 forgot(+0.0)
2 very hard (+4.7)
4 easy (+5.0)
5 very easy (+5.2)

Shouldn't it have more spread something like this for example?
0 forgot (/2)
2 very hard (+1)
4 easy (+3)
5 very easy (+7)

Original comment by maxwell....@gmail.com on 22 Apr 2012 at 6:44

[GoogleCodeExporter]

Did this feature get implemented? If so, how does one use it? Thanks.

Original comment by RTHid...@gmail.com on 13 May 2012 at 8:36

[GoogleCodeExporter]

It is planned in next version. It will be in Options.

Original comment by mrlhwlib...@gmail.com on 13 May 2012 at 8:46

[GoogleCodeExporter]

This is great news!

Original comment by Schinkel...@gmail.com on 13 Jun 2012 at 8:59

[GoogleCodeExporter]

Original comment by mrlhwlib...@gmail.com on 22 Jun 2012 at 11:11

• Changed state: WIP

[GoogleCodeExporter]

Maybe you need to add "restore settings by default" button only for Learning 
Algorithm customization ???

Original comment by mchin...@gmail.com on 23 Jun 2012 at 5:56

[GoogleCodeExporter]

Nightly build with customisation algorithm.
When failed to remember a card
default grades:
0-0 1-0 2-1 3-1 4-1 5-1
I change:
0-0 1-0 2-1 3-1 4-1.5 5-2
I open DB french-body-parts
I push forgot+0
After I restart BD, I still get
0 0 +1 +1 +1 +1
I expected to see 0 0 +1 +1 +1.5 +2 (0 +1 +1.5 +2 for Anki style)

Or I understand this in wrong way....

Original comment by mchin...@gmail.com on 23 Jun 2012 at 8:25

[GoogleCodeExporter]

I just added the UI, it is not functional yet. 
Reset could be a good idea. I dunno if it is doable.

Original comment by mrlhwlib...@gmail.com on 23 Jun 2012 at 8:27

[GoogleCodeExporter]

The initial implementation of algorithm customization is done. It should be in 
today's nightly build. If anyone installed the previous nightly build, you need 
to reset all your preference or AnyMemo will force close when opening any new 
db.
The code is still very rough for today's nightly build though.

Original comment by mrlhwlib...@gmail.com on 27 Jun 2012 at 4:57

[GoogleCodeExporter]

No reset 'Learning queue size' after 'Reset cistomized scheduling algorithm'.
Not important, just notice.

Original comment by mchin...@gmail.com on 28 Jun 2012 at 9:24

[GoogleCodeExporter]

Easyness factor change - impossible to enter numbers with 'minus' from number 
keyboard.

Original comment by mchin...@gmail.com on 28 Jun 2012 at 9:30

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

Open any DB as Preview/edit.
Open any card Detail.

last_learn_date - numeric keyboard (ok)
next_learn_date - common keyboard (numeric keyb would be better)

Original comment by mchin...@gmail.com on 3 Jul 2012 at 3:25

[GoogleCodeExporter]

Hay, I suggest you file another issue for the numerical keypad issue. This 
needs to be fixed but not now. And it is not related to learning algorithm 
customization.

BTW, regarding one of your suggestion for constant timing, I will do something 
to meet that need. I forgot where you mention it though.

Original comment by mrlhwlib...@gmail.com on 3 Jul 2012 at 6:03

[GoogleCodeExporter]

I posted suggestion for constant timing here but I deleted it after.
It wont do in the way I described.
I suggested to redefine one of DELETE button, but it would ruine conception 
0-1-2-3-4-5.
I still dont realise beatiful way for this feature.
The main idea: to be able to set number of days for scedulind and have a button 
or menu item for this.

Original comment by mchin...@gmail.com on 3 Jul 2012 at 6:15

[GoogleCodeExporter]

This feature is released in 9.1. For any additional request that isn't done in 
this issue, please copy/ paste your request to a new issue.

Original comment by mrlhwlib...@gmail.com on 15 Aug 2012 at 4:31

• Changed state: Fixed