Some ideas for the future
- Thread starter MarcoD
- Start date
ckatt said:
I suspect that they'd offer whatever this is as a compatibility library for older phones - Google knows as well as anyone how long it takes to get a plurality of Android users on a particular new OS. It may actually just take the form of an improved Google Drive - they're already adding more and more of these sorts of features, a few more tweaks and they'll pretty much have it.
That would mean you'd still have to search the term, or part of it, and it would prioritize terms in the flashcard database?
I was thinking more of a crib sheet, quick access (I suppose like we do with the desktop shortcuts), brings up the "favorites" list. I actually keep a notepad doc at the moment with terms I usually forget but its a hassle to access.
My personal use of Pleco is for communication, so I'm usually typing messages / emails and looking up words, phrases etc as I need them. Actually, having seen the new Facebook chat heads and how they sit ontop of everything else, is that something you guys have thought of for a quick-access dictionary lookup? Hit a pleco "head" and lookup, hit the "head" and it disappears and you're back into the other app you're using. This is just my personal taste though.
I was thinking more of a crib sheet, quick access (I suppose like we do with the desktop shortcuts), brings up the "favorites" list. I actually keep a notepad doc at the moment with terms I usually forget but its a hassle to access.
My personal use of Pleco is for communication, so I'm usually typing messages / emails and looking up words, phrases etc as I need them. Actually, having seen the new Facebook chat heads and how they sit ontop of everything else, is that something you guys have thought of for a quick-access dictionary lookup? Hit a pleco "head" and lookup, hit the "head" and it disappears and you're back into the other app you're using. This is just my personal taste though.
Yes - sorry, that's what I thought you were asking for.
Ah. We do list recently created flashcards in the history screen now - so you'd like a way to quickly access a list of flashcards in a particular category from the main screen in the same way? How about if we just (say) added a new app shortcut option that would instantly bring up Organize with your favorite category? (in a new task)
Probably doable, but a lot of hacking - it's unlikely we'll attempt many more low-level Android hacks like this until we drop support for 2.x/3.x, which is still a ways away given that about 1/3 of our users are using them. (that's also the point where we're likely to do our next major UI revamp, though that won't delay cross-platform features like Cantonese audio or the implementable-on-Android aspects of our new type design)
burdenofhope
进士
Just a quick question: what kinds of dictionaries are still on the horizon? I read something about a possible Cantonese dictionary?
Thanks, Michael.
Thanks, Michael.
Cantonese yes, we've acknowledged that one in a few places. The others are tough to say - we've got a number of other things licensed but we don't know when or even if they'll all come out, but on the other side of that, with publishers getting better and better about supplying data in nice clean formats there can sometimes be as little as a few weeks' delay between licensing something and having it out in Pleco, so the fact that we haven't licensed something yet doesn't mean it won't appear soon.
I'm not sure about this one - I don't use the flashcard system at all in the app, and am probably alone in that, I find the method of learning doesn't work with me.
If its tailored so that its a list of "favorites", displayed like a dictionary search, each clickable etc. Thats the sort of thing I'd find useful. Hopefully I'm not alone in this! I tend to find these phrases etc get into my brain after regular use but that initial stage of learning them requires me to look them up every time I want to use them.
Flashcards can also be used as favorites lists - there's nothing forcing you to actually test yourself on them, but it seemed like it would be easier / more consistent to only have one system for saving lists of words rather than two. So if you simply tap on the + button to "favorite" a word you can browse recently added words easily via the history screen.
Hello,
it's a long time I want to post here but I had never taken the time to create an account.
First of all, I would like to tell you that I love Pleco. I started using it intensively since September 2012 as my main learning tool. I use it every day (especially the paid flashcards add-on), and all my Chinese friends are surprised about my fast level increase.
As a computer science student and an intensive Pleco user, I would like to tell you which features I would like to see in a future release.
1) I love your handwriting input add-on. For users who bought it (or got it for free thanks to the offer on Android), it would be very nice to get this as a standalone input method on Android.
I've tried many Chinese handwriting input methods on Android, but none of the algorithms is better than yours. I've always been disappointed by the mistakes made by the recognition algorithms of other tools.
Your algorithm is so smart that it would be great to use its power to make it a standalone handwriting input method (it would be the best Chinese handwriting IME of the Android market).
Google Pinyin IME (handwriting part) is not bad, but the drawing frame is very small, and the user cannot take his time to draw a character because after a given timeout, the drawing screen will be cleared and the user has to start over again (it's OK for native Chinese who draw fast, but not OK for beginners who take their time and may stop in the middle of a character to remember the remaining parts
).
2) It would be nice to get very sophisticated statistics about the learning in the flashcards system. It would give motivation to learn.
Not only figures, but also charts would be nice. One data would be very interesting : literacy rate (number of unique characters known) over the time.
I have to do the stats manually using a bash script (I first have to export the XML, then keep only the Chinese expressions, then put one character per line, then make the list be unique and finally count the number of lines, which gives me the total number of unique characters I know).
I repeat this process (I export the XML every week) in order to draw a chart of the evolution of my literacy rate.
This kind of data gives me a lot of motivation, when I see that I know 500 unique characters and 1200 expressions in only 10 months.
3) I'm currently using the flashcard system with 3 profiles :
- reading ;
- understanding ;
- writing.
For the reading part, I have to enter the pronunciation of the words with the tones. I show the character and the meaning, and the pronunciation is asked, so that for "更", I can know which one is meant (geng1 or geng4) (the displayed definition gives me the advice).
For the understanding part, I see the characters and the pronunciation of the words, but I have to think of the meaning. I always know which meaning is asked because I can see the characters and the pronunciation (so that I can distinguish 子 (son) and 子 (noun suffix)) when I'm guessing the meaning.
For the writing part, I display the pronunciation and the meaning. These 2 data enable me to know which character I have to draw.
Some characters appear several times in different flashcards. For example, I have a flashcard "是", another "是不是", another "是否".
It's good to see them all when I'm doing my reading or understanding test.
However, for the writing, I lose a lot of time drawing several times the same character (though it's useless since it's not like the pronunciation : the writing of a character doesn't change depending on which character is beside it).
So it would be very nice (like I read in a previous message of this topic) to automatically split the "multi-characters-flashcards" into single ones.
I would be able to only review and score each individual character for the writing part, and continue like I do right now (mixing of single and multi-characters flashcards) for the reading and understanding part.
Let's say I added the flashcard "是" in September 2012 when I started learning Chinese using Pleco.
Now, in June 2013, the score of this flashcard (the writing part) is very very high (let's say 250 days).
I recently added a new flashcard "是否". The score is something like 15 days high.
The fact that the "是" is part of these two flashcards lead to the following : when the "是否" will be due for today, I will also write "是", and it will fool the algorithm, because I review "是" much earlier than if I had waited for the 250 days of the single-character flashcard "是".
4) When using SRS option in a flashcard profile, I would like an option like Anki like "limit the total number of cards per day".
I recently lost my phone (I had fortunately saved the whole database before). Once I bought my new smartphone one month later, I restored the Pleco data into the new phone.
The restoration worked fine, but to Pleco, it was as if I just stopped learning for one month.
The consequence was : 600 cards due for reading, 600 due for understanding and 600 due for writing. It took me several days to go over the 1800 questions.
I would like to be able to limit the number of questions per day (not per session, but per day). I would like to see "there is no more card due for today" even if there are still 500 to review, but if the limit has been reached for today.
5) The pronunciation of the 5th tone is too much high. In Pleco, it is pronounced like a 1st tone. But it should be much lower (pitch and volume at the same time). If I play the word "頭髮" in front of my Chinese friends, they will laugh because they hear 頭發.
6) I'm bilingual French / German, and I also understand English, so when I learn Chinese, I use the three languages to cross the definitions to understand more. It's good, because some words are only translated in the German dictionary, some only in the French, etc.
But there is an inconsistency between some words of different dictionaries regarding tones.
I'm a learner, so sometimes I don't know which tone is a word because several dictionaries say different things (most of time, the problem is only about the 5th tone).
For example : PLC, ADS, CF and CC dictionaries say 早上 is pronounced zao3shang5. But HDD and CMT say it's zao3shang4.
7) Your handwriting input is wonderful. I can draw fast, slowly, like a computer, or like "human script" using only one stroke to draw 20 strokes, it always works.
But there are some exceptions. It seams like some words are recusant.
There are some (very few) words that are almost never recognized, no matter how well I draw it.
I just have two examples :
- 日 ;
- 塊.
For the first one (日), the Pleco algorithm succeeds recognizing it let's say 70% of the time.
But 30% of the time, there is only one candidate which is shown : it's always "八".
So, when I see this slight bug, I just write it again, and usually, it works the second time.
For the second one (塊), Pleco almost never succeed to recognize it.
Pleco displays lots of the candidates, but none is the right one. An example of candidate which is shown: 瑰 (which looks very similar, just the 土 replaced by 王). But if I scroll through the candidates, I 95% of the time never see the right one.
It's possible that it doesn't come from your algorithm and comes from my bad writing skill, but I think I'm drawing quite well, and I respect the exact stroke order.
8) This is not a feature request, but just a question :
How do you decide the splitting of a characters parts in the "char" tab of the handwriting add-on ?
For some characters, it's obvious. Like 否 = 不 + 口, or like 是 = 日 + the part below it.
But for some others, it feels like you have to "make decisions" about the splitting, because it has no Chinese sense, no historic sense. For example, the splitting of 亭.
For me, I just see the 亠 + 口 + 冖 + 丁, but I don't understand your splitting, it seams you created custom Unicode chars, like one which contains in one word "亠 + 口 + 冖". Maybe you wanted to create this character because you noticed that lots of words share this part (so it reduce the number of links in your graph). But does it really make sense for learners?
I'm looking forward to the new version of the Android Pleco app!
it's a long time I want to post here but I had never taken the time to create an account.
First of all, I would like to tell you that I love Pleco. I started using it intensively since September 2012 as my main learning tool. I use it every day (especially the paid flashcards add-on), and all my Chinese friends are surprised about my fast level increase.
As a computer science student and an intensive Pleco user, I would like to tell you which features I would like to see in a future release.
1) I love your handwriting input add-on. For users who bought it (or got it for free thanks to the offer on Android), it would be very nice to get this as a standalone input method on Android.
I've tried many Chinese handwriting input methods on Android, but none of the algorithms is better than yours. I've always been disappointed by the mistakes made by the recognition algorithms of other tools.
Your algorithm is so smart that it would be great to use its power to make it a standalone handwriting input method (it would be the best Chinese handwriting IME of the Android market).
Google Pinyin IME (handwriting part) is not bad, but the drawing frame is very small, and the user cannot take his time to draw a character because after a given timeout, the drawing screen will be cleared and the user has to start over again (it's OK for native Chinese who draw fast, but not OK for beginners who take their time and may stop in the middle of a character to remember the remaining parts
).2) It would be nice to get very sophisticated statistics about the learning in the flashcards system. It would give motivation to learn.
Not only figures, but also charts would be nice. One data would be very interesting : literacy rate (number of unique characters known) over the time.
I have to do the stats manually using a bash script (I first have to export the XML, then keep only the Chinese expressions, then put one character per line, then make the list be unique and finally count the number of lines, which gives me the total number of unique characters I know).
I repeat this process (I export the XML every week) in order to draw a chart of the evolution of my literacy rate.
This kind of data gives me a lot of motivation, when I see that I know 500 unique characters and 1200 expressions in only 10 months.
3) I'm currently using the flashcard system with 3 profiles :
- reading ;
- understanding ;
- writing.
For the reading part, I have to enter the pronunciation of the words with the tones. I show the character and the meaning, and the pronunciation is asked, so that for "更", I can know which one is meant (geng1 or geng4) (the displayed definition gives me the advice).
For the understanding part, I see the characters and the pronunciation of the words, but I have to think of the meaning. I always know which meaning is asked because I can see the characters and the pronunciation (so that I can distinguish 子 (son) and 子 (noun suffix)) when I'm guessing the meaning.
For the writing part, I display the pronunciation and the meaning. These 2 data enable me to know which character I have to draw.
Some characters appear several times in different flashcards. For example, I have a flashcard "是", another "是不是", another "是否".
It's good to see them all when I'm doing my reading or understanding test.
However, for the writing, I lose a lot of time drawing several times the same character (though it's useless since it's not like the pronunciation : the writing of a character doesn't change depending on which character is beside it).
So it would be very nice (like I read in a previous message of this topic) to automatically split the "multi-characters-flashcards" into single ones.
I would be able to only review and score each individual character for the writing part, and continue like I do right now (mixing of single and multi-characters flashcards) for the reading and understanding part.
Let's say I added the flashcard "是" in September 2012 when I started learning Chinese using Pleco.
Now, in June 2013, the score of this flashcard (the writing part) is very very high (let's say 250 days).
I recently added a new flashcard "是否". The score is something like 15 days high.
The fact that the "是" is part of these two flashcards lead to the following : when the "是否" will be due for today, I will also write "是", and it will fool the algorithm, because I review "是" much earlier than if I had waited for the 250 days of the single-character flashcard "是".
4) When using SRS option in a flashcard profile, I would like an option like Anki like "limit the total number of cards per day".
I recently lost my phone (I had fortunately saved the whole database before). Once I bought my new smartphone one month later, I restored the Pleco data into the new phone.
The restoration worked fine, but to Pleco, it was as if I just stopped learning for one month.
The consequence was : 600 cards due for reading, 600 due for understanding and 600 due for writing. It took me several days to go over the 1800 questions.
I would like to be able to limit the number of questions per day (not per session, but per day). I would like to see "there is no more card due for today" even if there are still 500 to review, but if the limit has been reached for today.
5) The pronunciation of the 5th tone is too much high. In Pleco, it is pronounced like a 1st tone. But it should be much lower (pitch and volume at the same time). If I play the word "頭髮" in front of my Chinese friends, they will laugh because they hear 頭發.
6) I'm bilingual French / German, and I also understand English, so when I learn Chinese, I use the three languages to cross the definitions to understand more. It's good, because some words are only translated in the German dictionary, some only in the French, etc.
But there is an inconsistency between some words of different dictionaries regarding tones.
I'm a learner, so sometimes I don't know which tone is a word because several dictionaries say different things (most of time, the problem is only about the 5th tone).
For example : PLC, ADS, CF and CC dictionaries say 早上 is pronounced zao3shang5. But HDD and CMT say it's zao3shang4.
7) Your handwriting input is wonderful. I can draw fast, slowly, like a computer, or like "human script" using only one stroke to draw 20 strokes, it always works.
But there are some exceptions. It seams like some words are recusant.
There are some (very few) words that are almost never recognized, no matter how well I draw it.
I just have two examples :
- 日 ;
- 塊.
For the first one (日), the Pleco algorithm succeeds recognizing it let's say 70% of the time.
But 30% of the time, there is only one candidate which is shown : it's always "八".
So, when I see this slight bug, I just write it again, and usually, it works the second time.
For the second one (塊), Pleco almost never succeed to recognize it.
Pleco displays lots of the candidates, but none is the right one. An example of candidate which is shown: 瑰 (which looks very similar, just the 土 replaced by 王). But if I scroll through the candidates, I 95% of the time never see the right one.
It's possible that it doesn't come from your algorithm and comes from my bad writing skill, but I think I'm drawing quite well, and I respect the exact stroke order.
8) This is not a feature request, but just a question :
How do you decide the splitting of a characters parts in the "char" tab of the handwriting add-on ?
For some characters, it's obvious. Like 否 = 不 + 口, or like 是 = 日 + the part below it.
But for some others, it feels like you have to "make decisions" about the splitting, because it has no Chinese sense, no historic sense. For example, the splitting of 亭.
For me, I just see the 亠 + 口 + 冖 + 丁, but I don't understand your splitting, it seams you created custom Unicode chars, like one which contains in one word "亠 + 口 + 冖". Maybe you wanted to create this character because you noticed that lots of words share this part (so it reduce the number of links in your graph). But does it really make sense for learners?
I'm looking forward to the new version of the Android Pleco app!
Thanks for this detailed feedback.
1. Making our handwriting recognizer a standalone IME would be difficult for licensing reasons, unfortunately; we pay less to license that handwriting recognizer on Android than we would otherwise (and actually much less than we pay on iOS, where IME'ing isn't an option) specifically because we *don't* offer it as an IME. In theory we might be able to launch that as a separate product, but we're too busy with our other projects at the moment to consider it.
2. This is coming, probably early in 2014 along with greatly revamped flashcard internals in general. (2013 is more about the dictionary end of things)
3. This should also be coming, it's been frequently requested and ties in nicely with the stroke order test (among other things); just a little difficult because we have to sort out the correct character flashcard in cases of one-to-many jianti/fanti mappings and characters with multiple Pinyin. (even trickier when you add in Cantonese)
4. That should be coming along with #2, but it's tricky because in SRS, if you don't study enough cards per day you're (at least in theory) going to have a very hard time memorizing any of them; you need to "catch up" or else you'll end up with every card reviewed behind schedule and a whole lot of forgotten cards. So we've been working on an algorithm to deal with that problem, one that will catch up / reschedule cards and will also in some circumstances remove cards from active study (and reintroduce them as new cards later) if there's simply no way to stay on top of your reviews otherwise.
5. For the male or female speaker?
6. Yes, this sort of inconsistency is sadly quite common among dictionaries - we're working on a system to display all dictionaries' entries on the same screen (will debut first on iOS for performance reasons but should also come to Android soon), and that will include coverage of variant pronunciations to tell you all of the listed pronunciations for a word (and which dictionary is responsible for which). But in most cases we're contractually forbidden from modifying this sort of thing, so all we can really do is try to resolve the differences when merging entries.
7. These are known issues - we've actually licensed a newer / even nicer recognizer which doesn't have these problems, but it's currently only available in our iOS app because it requires an ARMv7 processor and too many of our Android users have devices with older processors. (Apple started requiring developers to drop support for iPhones with pre-ARMv7 processors last fall, so while that wasn't very nice for the users of those devices, it did make the handwriting upgrade much simpler) We think we have a way around that which we should be able to release later in the year, though.
8. That database actually comes from our stroke order diagrams feature, so the breakdowns are a little quirky because of that. We've got a separate database that we're working on ourselves that should be more intelligently constructed for component breakdowns, though.
1. Making our handwriting recognizer a standalone IME would be difficult for licensing reasons, unfortunately; we pay less to license that handwriting recognizer on Android than we would otherwise (and actually much less than we pay on iOS, where IME'ing isn't an option) specifically because we *don't* offer it as an IME. In theory we might be able to launch that as a separate product, but we're too busy with our other projects at the moment to consider it.
2. This is coming, probably early in 2014 along with greatly revamped flashcard internals in general. (2013 is more about the dictionary end of things)
3. This should also be coming, it's been frequently requested and ties in nicely with the stroke order test (among other things); just a little difficult because we have to sort out the correct character flashcard in cases of one-to-many jianti/fanti mappings and characters with multiple Pinyin. (even trickier when you add in Cantonese)
4. That should be coming along with #2, but it's tricky because in SRS, if you don't study enough cards per day you're (at least in theory) going to have a very hard time memorizing any of them; you need to "catch up" or else you'll end up with every card reviewed behind schedule and a whole lot of forgotten cards. So we've been working on an algorithm to deal with that problem, one that will catch up / reschedule cards and will also in some circumstances remove cards from active study (and reintroduce them as new cards later) if there's simply no way to stay on top of your reviews otherwise.
5. For the male or female speaker?
6. Yes, this sort of inconsistency is sadly quite common among dictionaries - we're working on a system to display all dictionaries' entries on the same screen (will debut first on iOS for performance reasons but should also come to Android soon), and that will include coverage of variant pronunciations to tell you all of the listed pronunciations for a word (and which dictionary is responsible for which). But in most cases we're contractually forbidden from modifying this sort of thing, so all we can really do is try to resolve the differences when merging entries.
7. These are known issues - we've actually licensed a newer / even nicer recognizer which doesn't have these problems, but it's currently only available in our iOS app because it requires an ARMv7 processor and too many of our Android users have devices with older processors. (Apple started requiring developers to drop support for iPhones with pre-ARMv7 processors last fall, so while that wasn't very nice for the users of those devices, it did make the handwriting upgrade much simpler) We think we have a way around that which we should be able to release later in the year, though.
8. That database actually comes from our stroke order diagrams feature, so the breakdowns are a little quirky because of that. We've got a separate database that we're working on ourselves that should be more intelligently constructed for component breakdowns, though.