2.1 Feature Requests

[Captain Planet]

Flashcards: in the free-answer mode asking for the character, after answering I can tap on the correct character and look up words with it, etc. It would be nice if I could also tap on the wrong character I wrote and look up what it is. [screenshot 1]

Also, there's a small bug in Preferences -> Fonts -> Zoom font size: I can never choose 72 pt: it seems saved, but doesn't change, and after I open the preferences again, it's back to 60 pt. (It'd be nice if you could add larger font sizes, too, like 100 pt or sth.) [screenshot 2]

And one more vote for character components / characters based on this character.

Of course Pleco is already great! Back to my flashcards now.

 

[mikelove]

Thanks for the note on that font bug - should be easy to fix in 2.0.5.

Tapping on incorrect characters in free-answer to look them up has actually been suggested on iPhone too; haven't added it yet but it's definitely seeming like it would be a good option to include (and trivially easy to implement).

 

[Captain Planet]

Thanks for the answer, Mike!

Something else, not terribly important, and not sure if it's a bug or feature. It seems some characters cannot be recognized from handwriting, I found 怱 and 鷄 (it's however possible to write 鸂). Also, there's the onomatopoeic character 呣 I needed to remove from my flashcards, because its pronunciation (m5) is never accepted as correct. (You can see I've been using Pleco intensively these days!)

 

[HW60]

Component breakdowns are a "maybe" on WM at this point - we don't even really consider that feature to be finished on iPhone yet (and it probably won't be until the next semi-major update, 2.2) since there are a number of components like 心 that aren't correctly linked with different-looking versions of themselves, and it's unlikely we'd roll it out on WM until we had it fully implemented on iPhone, so the question is whether there'll still enough interest in our WM software at that point to justify rolling out a new feature that people would have to pay for.

Is a character component database - like table of characters and components available on the world wide web or elsewhere? It seems that component breakdown will not be finished before I finish my chinese studies, but it would help learning a lot, especially for those characters where the component is not a radical at the same time (i.e. to find characters with a 殳 in the right position not being radical which among many others I like to mix up).

 

[mikelove]

Captain Planet - the recognizer doesn't support some particularly rare characters; it's a necessary tradeoff, the more characters it recognizes, the more likely it is to come up with a false positive when one of those rare characters is similar to the more common one you're actually entering.

HW60 - I haven't seen one, at least not one that's in a situation licensing-wise where we'd be able to use it.

 

[HW60]

In Flashcards / Export the records cannot be sorted, in Manage Flashcards / Batch I cannot select the categories I would like to, but can sort the output. I would like to sort the exported records like Manage Flashcards / Settings or select all kind of categories in Manage Flashcards to apply a batch export.

 

[mikelove]

Good idea - not sure which of those would be a better approach, probably the multi-category support in Manage since it would facilitate some other cool things too; thanks!

 

[flameproof]

In Flashcards / Export the records cannot be sorted, in Manage Flashcards / Batch I cannot select the categories I would like to, but can sort the output. I would like to sort the exported records like Manage Flashcards / Settings or select all kind of categories in Manage Flashcards to apply a batch export.

...and while the flashcard function is designed... having the option putting all 'don't knows' into a (temporary) category could be a useful function.

 

[mikelove]

Couldn't we accomplish that simply by adding a more specific "last answer" search to Manage Cards?

 

[Captain Planet]

Just a small suggestion: as a left-handed person I really appreciate it that you thought about the option for a left-handed flashcard interface. However, what this option should actually change to be useful is the character-writing box, which would be much more convenient to have on the left side of the screen. This way I wouldn't be pressing the screen in other places while writing a character with my left hand. The alignment of the buttons underneath is OK either way, and that's why I'm not using this option now, even though I'm left handed.

 

[mikelove]

That actually is an option, but it's done through a different setting - "Left-handed input interface" in the "Input" panel in Settings. (we figured that if someone wants the handwriting box on the left side of the screen normally, they'll probably want it there in flashcards too)

 

[Captain Planet]

Thank you again, Mike!

(I always use the full-screen input, never noticed this option before. )

 

[jiacheng]

I vaguely recall the idea of saved searches on flashcards being suggested before, but I don't remember if anyone mentioned the idea of "meta categories" i.e. a category in hierarchy that would be bound to a particular search criteria. It would also be cool if we could anchor a search criteria to a particular scorefile instead of just scorefile of the active profile.

The reason I'd like to do this is, I keep 3 main scorefiles, recognition, comprehension/writing, and expression. I'd like to be able to automatically exclude a card based on whether it has been tested in one of those other scorefiles on that particular day.

Currently to do this, I update categories after my sessions, but this is fairly cumbersome and it would be very nice if it were all automated.

 

[mikelove]

I vaguely recall the idea of saved searches on flashcards being suggested before, but I don't remember if anyone mentioned the idea of "meta categories" i.e. a category in hierarchy that would be bound to a particular search criteria. It would also be cool if we could anchor a search criteria to a particular scorefile instead of just scorefile of the active profile.

Doing this with categories would be a little tricky, but we might be able to add a "Card Filter" which would exclude any card that was included in the results of a particular saved search. You can already choose a different scorefile in Search Cards (bottom option on the screen), so we could certainly make that apply to saved searches as well as active ones.

 

[HW60]

Repetition-spaced card selection is a powerful tool to remember cards, but it has one drawback: The dayly amount of cards to be reviewed can be more than you are willing to review. This is especially the case when

• you just start using Pleco and have a big number of cards not yet reviewed.
• you add a considerable amount of new words (e.g. 100 per week) to your flashcards.
• for some reason you had no time to do your dayly review and want to get back after 2 weeks "doing nothing"

While the first and second problem can be helped a little bit by Advanced settings/Cards/Limit # of unlearned cards, there seems to be no help from Pleco for the third problem: you almost have no chance to catch up, when there are ten times more cards to be reviewed than you are willing to review.

I suggest to add a "Number of cards" option for repetition-spaced (as realised in frequency adjust) and then let Pleco choose the cards that are appropriate to catcch up. The problem is: what is the right order to select cards? I think it should be

• new cards, limited by "Limit # of unlearned cards" set (you cannot stop learning new cards being in class). Of course "Number of cards" should be bigger than "Limit # of unlearned cards.
• cards with low score come before cards with higher score

That means, that cards with a higher score, e.g. with several correct answers in a row, are reviewd after repeating the cards that urgently need a review. Pleco cannot find a solution, when "Number of cards" is to small to catch up. But as long as "All flashcards" * "Points per day" / "Maximum Score" is bigger than the chosen "Number of cards", there should always be a chance to catch up. (For example: With 500 cards all together, maximum score 10000 and points per day 100, Number of cards should be bigger than 500 * 100 / 10000 = 5).

 

[mfcb]

i had that problem a year ago, solved it like that:

- i limited my flashcard session to cards above a certain score (lets say 2500), if the number of cards is not enough for that day lowered the limit even more... that session adds 15% on each correct, so the cards disappear into the future, when i have them right...
- for new cards anyway i have a separate profile (same score file) where i specify the upper limit of the cards... (also uses different scoring method, adds fixed amount after 2 correct)

i was done with catching up, when both limits were again the same value (usually 200 in my case)...

 

[HW60]

I solved the problem (more than 1000 cards were overdue) too: first looked at scores of 100, than scores up to 600 and so on; in the meantime to reduce the amount of cards a little bit more I looked at cards with a score between 5000 and 10000, then slowly the lower limit (600) went up and the upper limit (5000) went down. But I think this could be a simple problem for Pleco.

 

[mikelove]

Thanks for the feedback on this.

I continue to feel that adding an actual "number of cards" option to spaced repetition is a bad idea, because it can potentially result in you not reviewing any of your cards often enough to remember them. Consider this scenario: you've capped Pleco's system at reviewing 100 cards per session, but you have so many cards that you're averaging about 120 due for study every day. Of course the remaining 20 would get studied the next day, but after a week you'd have studied 700 cards but been expected to study 840, so you'd now have 140 cards that were running behind schedule, most by 1 day but some by a whole 2 days (which could already be disastrous for newly-added items). And as you review cards later and later, you remember more and more of them incorrectly, meaning even more extra cards to study and even more older cards falling back in the priority list until you forget them too - even just assuming that 120/100 disconnect, after a month you'd be behind by 600 cards or 6 days of study. Preventing the introduction of new cards until you'd worked through your queue of old ones might help matters a little, but the basic problem remans that you'd be falling farther and farther behind schedule without realizing it.

So we do need to do something about the problem of having too many rep-spaced cards to study, but I don't think letting people simply cap the sizes of sessions is it; some other approaches might work better, like adding a command to randomly bury some cards (maybe playing off of the priority system / giving it a bigger role): accept the fact that you don't have time to practice all of the vocabulary that you want to retain, and give up on some words. Adding a little wiggle room (per johnh113's suggestion in the iPhone forums) to intervals could also work - in addition to letting you review cards a bit early (5-10% or so of their interval), you could also review them a bit late, so if you were behind you could temporarily delay some cards but they'd still come up for review in a few days and you'd still have to clear the backlog eventually.

 

[HW60]

I continue to feel that adding an actual "number of cards" option to spaced repetition is a bad idea, because it can potentially result in you not reviewing any of your cards often enough to remember them. Consider this scenario: you've capped Pleco's system at reviewing 100 cards per session, but you have so many cards that you're averaging about 120 due for study every day. Of course the remaining 20 would get studied the next day, but after a week you'd have studied 700 cards but been expected to study 840, so you'd now have 140 cards that were running behind schedule, most by 1 day but some by a whole 2 days (which could already be disastrous for newly-added items).

Well, actually the problem of too many cards exists, and [mfcb] and I described a possible solution to solve it, though this is a difficult procedure. If you only cap Pleco's system reviewing 100 cards per session when 120 are due without any systematic approach (e.g. selecting from the 120 due cards 100 by the methods used with the frequency adjusted way), I agree with your answer. But when you prefer cards with a low score, the not reviewed cards are usually cards with a higher score, and it probably does not matter too much if you review cards after 60 days that are due after 50 days.

I remember a forum entry where somebody asked how to learn 500 cards with 20 cards a day. I think this is not a problem of "playing off the priority system" or "not have time to practice all of the vocabulary", but a problem how to get these 500 cards into the dayly repetition-spaced card selection within some weeks.

• If you only limit the number of unlearned cards to 20 in this case and have 80% correct answers on the first day, you will get 20 + 20% of 20 = 24 cards on day 2, and with the same assumption of 80% correct the number of cards rises steadily until day 7 (score 600 for the correct answers), when you get the 80% of 20 reviewed for the second time together with all the other cards. Of course cou can stop after 20 cards and wait for the next day, but then you have no idea what cards Pleco selected (new or old, high score or low score).
• But you could start with 20 on day 1 and review 4 old (wrong) cards and 16 new cards (instead of the limit of 20) on day 2 to 6, then 20 old cards on day 7 to 12 (16 correct from day 1 to 6 and 4 incorrect fromthe day before), then again 4 old and 16 new and so on.

This is surely just a first idea how to solve the problem, maybe not even a good one, but it is a way it could solved. If you manually without support by Pleco try to catch up, you do not want to look for the percentage of right answers to limit the number of new cards, therefore this is a typical problem to be solved by a program.
Actually there is no written help at all, and no built-in solution in Pleco, and the problem what to do after 2 weeks without repetition-spaced reviews is more difficult than starting only with new cards.

One final remark: actually with every correct answer the number of days until the next review increases and therefore the number of due cards decreases - that is the reason, why you can get 500 new cards into Pleco's repetition-spaced system with 20 cards a day, even if 20% of your answers is wrong all the time. Maybe this should be taken into account in your above example with permanently 120 due cards.

 

[mikelove]

The frequency adjustment system just randomly selects cards but with cards with higher scores coming up more often - it doesn't do any special magic to ensure that you see all of the cards you haven't seen in a while, or prioritize cards by anything other than their scores. (this is why I keep murmuring about removing it - I don't really feel like it offers any benefit over spaced repetition at this point because it's too random) So it's not a good idea to hang a spaced-repetition-optimizing system on that.

My problem with the card limit wasn't that you'll occasionally have a few days where you miss / have to study more (or review cards a bit late) to catch up, my problem with it is that people might keep falling farther and farther behind; decide they only want to study 100 cards a day and eventually be reviewing cards weeks after they're due. Yes, the manual might specifically tell you that you shouldn't do that / should only do that for a few days, but people wouldn't read it or would ignore its advice and weeks later they'd be wondering why they were having such a touch time remembering vocabulary. We still have to prioritize cards based on how far past due they are - otherwise, even if you were only a little behind you could still end up having a card's review get delayed a week because it would keep not showing up.

My 120 card example was operating on the assumption that you had so many old cards to review that the combination of those plus cards answered incorrectly from previous sessions would keep you at 120 per day (even assuming no new cards) - if the maximum score is at 10000 and you're reviewing even the highest-score cards every 100 days, having 12000 old cards (not at all out of line for some people) would leave you with 120 per day even if you kept answering everything correctly and didn't add any new ones. Now you can argue that we ought to have a higher maximum score by default, and we probably should, but then again a lot of people aren't even taking the time to review 50 cards a day, let alone 100.

I think the best solution to this problem might be to simply make "limit unlearned" also check for the number of cards currently due for review and refuse to introduce new cards if it was above a certain limit; combine that with a more user-friendly system for redistributing / extending the timelines of some cards (already supported through the Batch screen, but most people don't know about that), a slightly more flexible interval check that looks ahead to the next few days and brings some cards forward / pushes some cards back to even things out, and a higher maximum score (which we'd automatically re-set for everyone but then allow people to back and change again) and we could make vocabulary pile-ups much less of a problem than they are now. The redistribution system could be really beautifully done - a bar below the session start button telling you how many cards you had due, a button below that allowing you to redistribute some cards, a count of the number of times you'd done that recently / alert if it was too many, etc - and would be completely transparent, but having it be a button instead of a set-it-and-forget-it number-of-cards count would make it clear that this would be destructive to your word retention if done too often.