ZDT: Suggested features

[dorritg]

Hi,

Thanks for making this fantastic program available. It has been a great help to me as I start studying Mandarin. There are several additional features I would suggest:

• Add a flashcard mode that shows the English and requires the user to enter the pinyin
  
• After reviewing a deck of flashcards, give the user statistics for both that specific session as well as overall statistics.
  
• Allow the user to specify the number of times to repeat a card after getting it wrong, or after trying a card, allow the user to mark it for repeat, even if the user got the answer right.
  
• In "self-review" mode, allow the user to mark cards right or wrong and to indicate whether or not to repeat them in the session. Keep statistics on right/wrong answers for self-review mode separately from the other modes.
  
• Keep overall statistics and also separate statistics for each mode.
  
• Add a "chinese-to-english" mode that does not use multiple-choise. Instead identify keywords in the definition and mark the card correct if the user enters at least one of the keywords (or phrases).
  

Thanks again for making this tool available. I love it.

[bogleg]

Hi dorritg,

Thanks for your suggestions. I am recording them down for future consideration. Is there a reason why you want to separate out the statistics for the different test modes? I'm kind of reluctant to do that unless there's a good reason to...

Chris

[dorritg]

I want separate statistics for the different test modes because it is useful to me to know which mode needs the most practice. For example, I may be pretty good at recognizing the character for a particular word, but still not reliably know what the word means; or I may be pretty reliable at coming up with the English for a word given the Chinese, but have a difficult time coming up with the Chinese given the English. Separating the statistics for the different modes will help me know what needs the most practice.

Hope this helps clarify. Thanks for taking my suggestions into consideration! If you decide to implement any of them and would like help testing I'd be happy to do so.

Best,

Dorrit

[onebir]

# Allow the user to specify the number of times to repeat a card after getting it wrong, or after trying a card, allow the user to mark it for repeat, even if the user got the answer right.

# In "self-review" mode, allow the user to mark cards right or wrong and to indicate whether or not to repeat them in the session. Keep statistics on right/wrong answers for self-review mode separately from the other modes.

I'd like to suggest slightly different functionality:

- ideally, a supermemo-type repetition spacing algorithm (it's described here, and built into open-source mnemosyne/, so it shouldn't be too difficult to copy it)

- if that's not possible, ability to use OR as well as AND in the smart filter (which I imagine would be a lot simpler, although I guess it might raise questions about the order of evaluation of the operators? :s)

This would save users a lot of time being presented with characters they already recognise. The supermemo algorithm uses past performance to guess when you'll be on the brink of forgetting an item, and only shows you it at that point.

AND/OR evaluation would help by allowing selection rules like this:

- show me characters I've got right less than 3 consecutive times OR that I haven't seen for N days

I think being able to select OR for 'last tested' is most important - so you can ensure you see each character regularly, but not too often. (Supermemo-type algorithms improve on this by increasing N - the gap between repetitions - each consecutive time you recognise the character correctly)

(It'd also be useful to apply the filter when exporting. I've written a game for practicing hanzi & it'd be nice to be able to just export the characters that need work, & use them as input for the game. But perhaps no one else needs this feature; my programming skills are limited, but I imagine I could write my own filter inside the game without too much trouble...)

[bogleg]

Hi Onebir,

What do you think of the interval testing concept found in the Mac flashcard program, IFlash? (http://loopware.com/iflash/screens.php?p=7) I've already added this functionality to the latest ZDT build.

Thanks for the smart filter suggestions and the supermemo link. I'm going to look into them.

Chris

[onebir]

Hi Chris - I think the Iflash approach goes 90% of the way to solving the problem, & makes my comments about the filter less important. It's great news for me because I just got back to China, been reminded of my illiteracy (amongst other thing) & struggling to find software to help with it... So thanks for your continued heroic efforts on ZDT!

But some characters are easier/harder than others to remember - due to complexity, similarity with other characters etc. With the Iflash system, tricky characters get the same intervals as easier ones each time you get it right. This means you're likely to see tricky characters several days in a row, finally grasp them, but perhaps forget them before you get to them again, because their difficult doesn't affect the intervals. Then you're back to seeing them very frequently for a while again - perhaps unneccessarily.

I don't know how serious this problem is. But the early Supermemo algorithms offer a solution by associating an 'easiness factor' (EF) with each item, which is updated based on the (0-5) rating you give it when you see it - a high score increase the EF, a low one decreases it. This reduce the "one size fits all" problem, because the repetition intervals are scaled by the EF - 'high easiness' items get a longer interval, because the standard interval is scaled by a higher EF (and vice versa).

If you felt the urge, you could bolt this onto the Iflash approach with just one extra parameter for each item (an EF), & possibly a more finely graded - rather than binary - rating system. But I think you've solved the main problem already, so

There's a further remaining problem posed by the Iflash system (& the early supermemo algorithms) - what if the user's choices for the score/inteval relationship don't reflect their ability to remember things? Under Iflash, this would become apparent when they started getting a disproportionate number of items wrong at a particular score, relative to previous performance at lower scores. Since most people probably haven't got a clear idea of what these intervals should be, some system for rejigging them might be helpful. [Later supermemo algorithms have something like this, but actually I think using score rather than previous repetition interval (as in supermemo) to capture the item's state makes it clearer how it should work: if the % of correct/good answers over the last few days at score X is lower than at score X-1, reduce the interval, & vice versa. Repeat for all X (with enough answers for the %age not to be skewed; perhaps 20?, until the % correct levels out - perhaps around some predefined level)

[Odysseus]

Hi,

I would like to make a plea for the following feature.

Make it possible to edit the description.

This would permit a non english speaking student to use the flashcards to work with his native language. For me this makes it possible to lookup english-> chinese. Add a certain definition to a category and then edit the english definition to change it to Dutch (or any other language). Now I could use the flashcard to train Chinese--> dutch---> chinese.

I would like to think that this would make the software more widely useable. Anybody speaking moderately well English could define his own translation and make this set available to non-english speaking persons.

Thank you for considering this request.

[bogleg]

Hi Odysseus,

Do you mean editing the definitions? You can already do that. Just select the entry in the category you want to edit, and press Ctrl-E. A dialog box should appear which you can use to edit any aspect of the entry you want, including the definition.

Hope that helps

Chris

[mnglfiddle]

When going through the flash cards, I occasionally come across a word such as nü or lü (female, or green, for example). I cannot figure out how to enter the u with two dots above it (ü), and it will not allow me to paste it from the character map, like I did to put it in this note. No matter what I try, it always tells me I am wrong, so I can't even properly finish the quiz!

Any help on this would be appreciated.

Thanks,

Matthew

[Luobot]

nu:3 should work for 女

lu:4 should work for 绿

etc.

(Side note for thread management: More of a "How do I" than a "Suggested feature")

[mnglfiddle]

Thanks for the help. I had actually just figured it out by looking at a file in zdt itself.

Sorry for the OT question. I haven't figured out on this list how to start a brand new thread, so I looked for something with a hint of "zdt flashcard" in it. I'll try not to do it again.

I am really enjoying ZDT more and more. Am looking forward to the next version. The recommendations sound great. Can't think of anything else myself. Great program!

Matthew