New Flashcard Features

mikelove

皇帝
Staff member
Thanks for those two suggestions. We're definitely planning to improve the button input in the new PlecoDict flashcard system, probably with some sort of user-configurable mapping like you describe; there are such a wide array of button configurations on Palms now that it's ridiculous to force everyone to use the same one. But because of that wide variety of configurations it's unlikely we'll adopt an onscreen button layout that matches the hardware buttons; palmOne alone has three different configurations, 2 buttons side-by-side, 4 buttons side-by-side and 4 buttons arranged in a circle, and then there's the Sony UX50 and the Tapwave Zodiac and various other devices with unusual button layouts.

And simplified/traditional switching is also something we're planning for flashcards; in fact, now that our font system supports it we may even start allowing people to display simplified and traditional versions at the same time.

Thanks again,

Michael Love
Pleco Software
 

Ants

Member
mikelove said:
Ants said:
- It would be helpful if the Flashcard software could hold its state when the PDA is switched off, so that one could return to the same position without having to restart a test each time.

I assume you mean when the dictionary is exited, correct? Simply turning off your Palm won't affect the flashcards at all, when you turn it on again it'll go right back to where you left it. We're probably going to add the ability to save your session progress in the next version of the flashcard system, along with a host of other ways to store user data.

This must be a record for late replies, sorry abt that :). On my Treo, I have PalmOS 4.1 set up with password protection, so your observation is really interesting - with password protection switched on, Palm OS 4.1 automatically exits the dictionary on power-off. If there was a way you could by pass this, then that would be perfect, but having saveable sessions is a reasonable workaround.
 

Ants

Member
Could we have SuperMemo functionality?

I am studying for a master's degree in Chinese studies and over the last academic year have been using the Chinese-Oxford dictionary's built in flashcard system. However, with just six weeks to my finals, I had so many characters that it was very time consuming going through all the characters I had learned over the year. Removing characters that I feel are in my long-term memory could be one solution, however, there are times when even these characters need refreshing.

My conclusion is that flashcard management (i.e. the frequency by which a given card should be shown) for building long-term memory cannot be achieved by manual means - it requires software intervention. Just six weeks before my exams, I found software that could help me with this problem (http://www.supermemo.com) and it has made a huge difference. If Pleco want a killer app, then I would recommend they seriously look into building similar functionality into its software.

In brief, the functionality required is the creation of metadata on every single flashcard that describes the number of times the card has been tested, the success rate, and an assessment of difficulty (calculated from the history of that card for that user). The flashcard system then parses this metadata using a simple algorithm that shows the "difficult" cards more frequently and "easy" cards less frequently. The system is responsible for the complete management of the cards over days, weeks, and months, using well established principles from research into long-term memory.

This kind of implementation would, as I say make Pleco dictionaries killer apps, with huge competative advantage over any rivals.
 

mikelove

皇帝
Staff member
Something like this is actually already in the works for PlecoDict. It's unlikely we'll roll out anything quite as sophisticated as the SuperMemo algorithm, since even if we went to the (considerable) effort and/or expense required to put something like that together, it's doubtful that it would be the best fit for most of our customers. But we're certainly planning a feature that allows you to automatically adjust the frequency with which flashcard words come up based on how well you perform on them. The current design is kind of a hybrid of the task-based and frequency-adjusted models discussed elsewhere in the forums; it's pretty flexible, and with built-in presets it should be possible to realize almost any reasonably straightforward system for automatically organizing flashcards.
 

MichaelK

秀才
flashcard Lessons???

Right now I use another flashcard program and am very happy with it; but, like everything else it could be better and can think of no better place than in my dictionary. I know it's kinda late to be incorporated into the summer release of the PlecoDict ... but better late than never.

I enter my study material into my flashcard program which ups my retention of what I studied. The process of entry takes time and is tedious (yes I'm very lazy).

The point...

Do you think it's profitable for you (Pleco) to contact the better Chinese language learning publishers to get rights to adapt their curriculum to a palm aide/extension as an add on for your flashcard feature? What I'm thinking of is "A Key to Chinese Speech and Writing" by Joel Bellasssen and Zhang Penpeng, Sinolingua, Beijing. Anything of the same caliber would also be very welcome. (I'm thinking that if you could amplify that it would give them more international "electronic" exposure (a new market for them) for no work on their side (because the books are already in ready format for computer editing). Hopefully, it can be a free license for you thus benefitting both parties and me) (grin).

Well just a thought for sellable educational flashcard filesets.
 

mikelove

皇帝
Staff member
Actually several of them have already contacted us, offering lists of vocabulary for us to put into our flashcard system, so it looks like we might indeed have free commercial-quality flashcard lists available at some point in the not too distant future. Now that we're adding a text import feature to the flashcard program it should be very easy to bring in these sorts of lists, so it's really just a question of getting publishers interested, and I suspect most of them would gladly give away word lists for free in exchange for increased exposure for their books.
 

Mads

举人
Does this mean, that if later on, when we get the oppertunity to exchange flashcards here on the site, we can't exchange wordlists from books where you don't have a deal with the puplisher?! :shock:
 

mikelove

皇帝
Staff member
Well we're certainly not planning to police it that closely - if a publisher wrote and asked us to remove a list from our website, we'd probably have to comply, but barring that unlikely event you should be OK. Frankly I doubt that very many publishers would mind, most of them would probably be happy to have the additional exposure for their books and would (correctly I think) surmise that most of the people downloading the lists owned the actual books and were simply using the lists to supplement them. They really aren't all that useful by themselves.

As far as I can tell, this is kind of a gray area of copyright law anyway - one can easily argue that creating such a list constitutes fair use, or that what's being excerpted contains only a minimal amount of original content anyway, so it certainly doesn't seem like a blatant copyright violation to let people post these word lists. Of course if people start copying actual content from the books (usage explanations etc) and putting them into custom dictionary entries, that's a different story and we probably would have to put a stop to that, but hopefully that won't happen.
 

MichaelK

秀才
One thing I don't like about the "other" flashcard program.

Characters.

I can now recognize a bunch of charaters and meanings ...
but write it ... no go.

I liked the Wenlin because it showed me this shortcoming.


Soooo.....

Is it possible to include a character recognition aspect into your flashcard program ... so you know you have the right shi.

or de.

(along with a tone marker for Chinese characters, (actually I think this is revolutionary because a non native chinese speaker doesn't have the 20 years or so of hearing the proper tones, hence a tone marker for the Sinograms, pls forgive my heresy).

but now your moving into wenlin territory.
I hope your relations are with the ABC and not the App.
 

mikelove

皇帝
Staff member
Well it wouldn't be all that difficult to implement, but I'm not quite sure what advantage that would offer over simply letting people scribble down the characters on the screen in that writing area we have now - our recognizer isn't 100% accurate, and in fact for some people it's considerably less than that, but if you draw the character yourself and compare it to the correct answer you can be sure you're checking yourself accurately. Plus you can vary your tolerance for mistakes, which the recognizer can't do.

Once we release our new, revamped handwriting recognizer (however many months from now that is) we may be able to start looking at features like this, since that will give us the ability to do things like limit the recognizer's character set to a few dozen "shi's" and test only to see if you've drawn the correct one, but for now I don't think that a character recognition feature in the flashcards would work well enough to be worth including.

Oh, and we've already gotten a few suggestions about tone marking for Chinese characters; we've got an interesting (and as-yet unannounced) new feature in the works that should not only cover this but provide some other benefits for study as well.
 
flashcards

Three suggestions for the flashcard system, forgive me if I'm repeating something that has already been mentioned.
1. A learning mode that retests mistakes during the SAME test. So for example I can take a learning test of 50 flshcards, and the test won't end until I get every card right a couple of times. I think this was already mentioned, but I just wanted to emphasise the repeat asepect of it. Then of course a test system that tests me on all the cards once and gives me a percentage score immediately following the test, I think this is a given.
2. Is there anyway to create categories for flashcards? So for example using the Flash! flashcard program, I have different flashcard decks for each of my books, and then within each deck I have the cards sorted by chapter. So this way it is easy for me to pick which book I want to review and then go through the chapters accordingly, or test myself on the whole book if I want to. After studying Chinese a couple of years, things in the flashcard deck can get a little messy.
3. Be able to add new entries to the dictionary or customize existing ones, i.e. if I want to add a grammar point to a definition, or happen to find some chengyu that isn't in the massive ABC dictionary, I can add it without too much hassle. This may be impossible to implement in the dictionary section, but is there anyway to make it happen in the flash card part?
Also I just want to add that I am in full support of it being simple to select which fields to be tested on and then which fields are given as the answer. So for example if I want to be tested on the English definition, but I only want the answer to be shown in Hanzi (no pinyin) then have the pinyin show with the touch of the button. Then if I want to do it the opposite, have it be a SIMPLE process to change what fields I want to be tested on.
I haven't tried the 2.0 flashcard system very much (mostly due to the lack of repeating cards until they are correct) so hopefully I'm not talking to much old news.
Been using the Oxfard palm dictionary for over two years now, can't wait till the full version of the ABC dictionary comes out and I never have to search through a radical index ever again! Paper is so antiquitated.
 

mikelove

皇帝
Staff member
Thanks for your feedback.

We actually have something along the lines of #1 in the works in the form of a "continuous review" system that keeps bringing up new cards until you tell it to stop - it's not quite the same as what you describe, though, since it won't keep recycling the cards until you get every card right. But actually that's an interesting idea... we're retooling the flashcard review algorithm right now so while I can't promise it'll make it into the finished version we'll definitely consider it (and if it's not in 1.0 it might still appear in some future release).

#2 is already in place for the PlecoDict 1.0 flashcard system - we've been surprised at the number of requests we've gotten for this, actually, when we designed the original flashcard system we didn't think anyone would use it for more than a few hundred words at a time.

#3 is also in the works for PlecoDict 1.0 - you'll be able to add custom dictionary entries at will. Actually you can already do this in the Oxford E&C flashcard system - if you just tap on the "New" button while editing a flashcard list you can create a new customized entry with whatever word you want in it.

Varying which fields you're tested on is another feature that's already in Oxford E&C, but we're making it much better in PlecoDict: you'll be able to choose the fields shown both before and after you hit the 'reveal' button and you'll be able to vary them by specific flashcard lists or 'ranks' (our system for adjusting flashcard frequency based on performance).
 

KingCat

Member
Super-Memo Style

Mike said:
It's unlikely we'll roll out anything quite as sophisticated as the SuperMemo algorithm...

Actually, Mike, you can create an algorithm similar to the supermemo one without a whole lot of trouble, as long as you make it for an "average" person, and not personalize it. If you download a free version, and look through all of their super-duper-smart matrices, you can simplify the number of days to skip to something like A * n ^ B, where n is the number of times answered correctly in a row, A depends (slightly) on how many times answered incorrectly, and B is a constant. I did it myself for a flashcard program I wrote a while ago.

Supermemo goes through a whole lot of trouble optimizing, when a much simpler single equation gives an answer (days of delay) within 2% error.
 

mikelove

皇帝
Staff member
Well that still doesn't change my basic gripe with the SuperMemo concept, which is that in language learning the frequency with which you see a word in a flashcard program has little bearing on the frequency with which you encounter it in classes or outside in the real world.

However, PlecoDict's flashcard system is sufficiently powerful that one could probably configure it to approximate this behavior - I say "probably" because a few features are still up in the air for 1.0 at least, but given the number of SuperMemo fans among our customer base it would probably be worth adding some sort of a mechanism for repetition spacing.
 

Jim

榜眼
While most of your customers are students a few of us are working full time and studying in whatever time we can spare. I find the larger my flashcard list the less often and less thoroughly it gets done. It would be nice to be able to set a maximum number of cards in each category and in total. Every time I add a new card it would drop one of the cards with the best test record or if none had tested well then the oldest one. If I run across a word in daily use that has been deleted I would just add it again which would keep the list focused on words I need most.
 

mikelove

皇帝
Staff member
Interesting idea, though I'm afraid it may not be practical to add any more features like this in 1.0... we might also offer the option to move words to another, unlimited-length list instead of dropping them altogether.
 

Jim

榜眼
That would be easy to do myself as I could just create another category. However, if I did that then each time I come across a word I have to remember it is in the list or if I need to add it anew. If it is in the list I would have to track it down. Unless each category has either pin yin or some other sorting it could take awhile to find it which means I would be quicker to just delete a word and add it in again when I come across it. Still, I can live without this and it is only a "would be nice to have". You can add it to your long list of things to be done someday.
 
[WARNING: SuperMemo Fascist]

mikelove said:
Well that still doesn't change my basic gripe with the SuperMemo concept, which is that in language learning the frequency with which you see a word in a flashcard program has little bearing on the frequency with which you encounter it in classes or outside in the real world.
I don't quite get this logic. SuperMemo is suppose to help you memorize items. Once something is memorized the frequency you see it in a flashcard program should be irrelevant.

I still have cards like 我, 是, and 美國 in my testing list, but they aren't a bother because they come up like once every 2 years now.

mikelove said:
However, PlecoDict's flashcard system is sufficiently powerful that one could probably configure it to approximate this behavior - I say "probably" because a few features are still up in the air for 1.0 at least, but given the number of SuperMemo fans among our customer base it would probably be worth adding some sort of a mechanism for repetition spacing.

I'll have to take a look at the Pleco flash card program, but I can't imagine it besting supermemo (even with the low rezness of it). It it truly is "sufficiently powerful" for me to configure it to mimic supermemo and have features that (I'd use that) the current palm version lacks, maybe I could make a switch.

I just haven't found anything that does a better job at manages the thousands of flashcards one needs (or wants) when studying Chinese.

---

I haven't even installed the dictionary yet :oops: . All was going to say was it'd be cool if this flashcard program could somehow integrate the SuperMemo algorithm.
 

mikelove

皇帝
Staff member
I wasn't agruing with the idea of repetition spacing, just the idea that an algorithm like SuperMemo's can effectively optimize the spacing intervals for learning vocatulary.

The whole idea behind SuperMemo is that a word will come up at the absolute minimum frequency necessary to maintain it in your memory, thus allowing you to learn the maximum number of words possible in the time you have available for memorization. However, with all of the extra exposure you're getting to Chinese outside of SuperMemo, it's inevitable that you'll end up reviewing a lot of words more often than you need to, so however good the algorithm is, it simply won't have enough information to determine the optimum spacing interval.

I'm sure SuperMemo is very useful for memorizing vocabulary, but given this handicap, I can't imagine it offers anything more than an incremental improvement over a much simpler algorithm like ours. With other types of information (licensing exam questions, state/national capitals, baseball statistics, etc) it's probably a lot more efficient, since you're much less likely to encounter those in class or a textbook or McDonald's or wherever.

But we certainly don't think our system is perfect - we hope it's good enough for the vast majority of our users, but if you find that SuperMemo helps you learn better then by all means keep using it.
 
mikelove said:
I'm sure SuperMemo is very useful for memorizing vocabulary, but given this handicap, I can't imagine it offers anything more than an incremental improvement over a much simpler algorithm like ours. With other types of information (licensing exam questions, state/national capitals, baseball statistics, etc) it's probably a lot more efficient, since you're much less likely to encounter those in class or a textbook or McDonald's or wherever.
I see your point now and I'd agree with that. I can't check out the flashcard program until I buy it (it's installed now :)). The fact that I'll just be able to click on a button and add a new card as opposed to copying, pasting and reformating a card to use it in SuperMemo (for palm) might be enough to sell me on it.

After I get my registration ID and play with the Pleco flashcard program expect to see me back :D.

It looks like I'd be able to use the Exported text file to make databases for SuperMemo (which would make things a lot easier).
 
Top