Feature Suggestions

[harrycallahan]

OK, I see I may have sparked some interest in this which is good. And I've learnt something new, had never heard of Amazon Mech Turk before. I actually had an idea to set up an open source style project to have that reference database built, let the online community parse each character of it's sub-components. Then I see sub-components mentioned here...

That deluxe score based solution shouldn't be too hard once you know sub-components. It doesn't have to be right all the time, if I'm getting options which tend to be close to the test character then it's a large step up for me. The learning for me comes via many tests over time. And you said you know position as well (may), that's even better. AMT would really be great though as all the fuzziness of human interpretation is incorporated. Also the automated solution is much preferred over manual, there's just too many characters to manually cater for.

I think the learning capability of flashcards would really take a giant leap with this system, it makes the mind focus on the detail, something you rarely have to do with purely random selections.

 

[mikelove]

Mark Kang - Sorry, which screen are you hitting the OK button too many times on? It'd be really tricky to redesign every screen to have a differently-located OK button... you actually should be able to close dialogs with a hardware button on most Windows Mobile devices, though; is there any button on your handheld with an "OK" on it, or something else that suggests it can be used to close windows?

harrycallahan - yeah, Mechanical Turk is a great tool, though it's a bit difficult to choose the sorts of workers you get on it. We've thought about setting up a similar system ourselves specifically for simple Chinese-language-related tasks like character matching, correcting / expanding / adding dictionary entries, etc; that way we could be a bit more rigorous in verifying that someone actually had some Chinese expertise, and could commensurately pay more (and reward longtime customers / fans / etc if there was a lot of interest in a particular job) - it'd be a cheap way for us to get large amounts of editorial work done and a great way for students / expats / etc to make some extra cash.

For character matching specifically, though, I think we'd end up using a combination of statistics from that distributed project and component / layout information from our various databases - that'd let us apply what we learned about commonly-confused radicals / strokes / etc to every character we have information on. (since it's not really worth the trouble to specifically test if, say, 靐 is frequently confused with 龘, given how unlikely it is anyone's actually going to be learning either one of those)

 

[Mark Kang]

Mark Kang - Sorry, which screen are you hitting the OK button too many times on? It'd be really tricky to redesign every screen to have a differently-located OK button... you actually should be able to close dialogs with a hardware button on most Windows Mobile devices, though; is there any button on your handheld with an "OK" on it, or something else that suggests it can be used to close windows?

Most screens on Pleco are dismissed via Winmobile's OK button. It seems all programs rely heavily on this button. I have a Dell X51V, which doesn't have a special hardware OK button, but I've mapped one to it now. I just noticed many of the Pleco screens have extra real estate, and since when working those screens I often already have my stylus out...

Ok, how 'bout this then- You've already got the tone color thing going, what I'd really like is to be able to highlight dictionary entries. Either to emphasize an example sentence, are to mark that a certain definition is actually used in the real world on occasion. For marking the definition, just a marking for the number would be sufficient.

 

[mikelove]

Good idea, but another thing that won't be done for your birthday, I'm afraid, since it fits in beautifully with the new annotation feature we're working on - that'll allow you to add usage notes / extra examples / etc to any entry in any dictionary, and should also work with abstract words (if you want the note to show up in any dictionary's entry for a particular word) - once we're doing that we could provide a way to highlight / add a note to a particular portion of an entry too.

 

[ben_gb]

Hi,

One thing I thought would be a handy addition is in the Flashcards "Prioritize by", add an option to introduce most recently added cards first (ie reverse date order).

This would help in my current situation, where I've got lots of 'older' cards I've collected together over the years, many of which are not a particular priority, but I do want to cover them eventually. However, I also just started on a new chinese course, and so I've got 'high priority' new words which I want to come up ASAP.

In the past on here, various ways of ensuring that cards can be jumped to the head of the queue have been discussed, e.g. by messing around with categories and profiles etc. Whereas I realised just adding the reverse date order introduction might be an easy way to cover many of the situations.

Or....

yet another way to do this would be to have a 'category' priority order, where you can choose the order of categories from which cards are introduced (eg in the category selection tab of "New Flashcard session"). So in my case I might put my categories like this:

-New cards category
-Old cards category

and Pleco will first introduce unreviewed cards from the "New cards category" before going onto the "Old cards" (but, within each category, still keeping the order you've chosen for "Prioritize by"). Other people might prefer to have the categories the other way around.

Anyway, just two ideas. I thought the first one might be quite easy to implement....?

Regards,
Ben

 

[mikelove]

This fits in nicely with the priority-based alternative to repetition-spacing we're contemplating for 2.1, actually - reverse date order would be extremely easy (wouldn't even need to re-sort the card list after we fetched it from the database, as with some other methods - we'd just query the database with that sort order) but I do like the idea of category-based prioritization for the "limit unlearned" feature too.

Interface-wise, actually, what we'd probably want is two panels specifically for this stuff, one for determining the order in which new cards are introduced via "limit unlearned" (number, prioritizing method, learned threshold, perhaps an option to remove some cards again if you fall way back in your pool, etc) and another for determining the order in which cards appear during sessions. It'd be a lot cleaner than what we have now, anyway.

 

[daniel123]

It would be great to have a kind of writer that allows pinyin or hanzi based input and let me use the dictionaries to find the correct hanzi characters (when used pinyin). This can be done as an improvement to the reader. It can help you writing longer hanzi texts like for example Wenlin does. Could be used for writing emails and sms in hanzi without need to know the characters exactly.

 

[mikelove]

We've been thinking about something like that for a while, actually - in fact the iPhone version includes a rudimentary text editor function, though really it's just designed to let you use our much-better-than-Apple's handwriting recognizer to edit text and doesn't include its own Pinyin system or anything like that. We've got some design sketches for a full-on Chinese text editor, with features like a built-in E-C query (enter English word, pop up entries for it, select a word in one of those entries and insert it into your document), pop-up C-E entries for the words you're about to enter, example sentences, phrase templates, etc, but we're not sure how many releases it'll be before we actually get that implemented in our software.

 

[goulniky]

Re: Feature Suggestions (Japanese)

I'm back from 2 weeks vacations in Japan : I had learnt the hiragana and katakana before going, but that was the extent of my knowledge of the language (loved the sound of it, and proceed to start learning it, but that's beside the point).
Knowing a large number of Chinese characters was of tremendous help, but I still found myself back to pre-Pleco times in terms of dictionary lookup, to get at both pronunciation and Japanese specific meanings.
I read the questions about Japanese support for Pleco, and Mike's replies in terms of priorities, need for a robust product supporting kanji / kana combinations etc. However, I was wondering if a custom user dictionary wouldn't work as an interim step : since most kanji will be recognized by the handwriting recognizer anyway, should a Japanese character dictionary based on EDICT or such like not be straightforward to load up? There does not seem to be much more than a hundred kanji unique to Japanese (kokuji), not all being frequently used, so those can be memorized :wink:

 

[mikelove]

I suppose you could create a user dictionary like that, but the lack of good support for mixing kana and kanji would be very limiting I think - there's also the issue of our handwriting recognizer being designed for Chinese stroke order (inasmuch as it cares about stroke order at all), which is in a few cases different from Japanese order (though they're generally pretty close). Plus we haven't really done any testing with kana and there might be a few cases where they're erroneously treated as one type of character by the search indexer when they're actually supposed to be treated as another type. If anyone wants to try it, though, the tools are all there, it's just a matter of formatting EDICT correctly for our user dictionary importer (maybe with a smaller subset of EDICT's entries) and running it through that to see what you get.

 

[xavyeah]

reading kangxi codes

Hi all,

Anyone could help me out understanding how kangxi codes work?
e.g. 建 = 0353.040 ; Just discovered Pleco had this function and
never used those before.

Pretty sure the reply in somewhere in this post, but a bit reluctant
in going through the 42 pages!

Thanks a lot in advance,

Xavier

 

[mikelove]

Anyone could help me out understanding how kangxi codes work?
e.g. 建 = 0353.040 ; Just discovered Pleco had this function and
never used those before.

Those are actually Kangxi dictionary page references; see the manual at:

http://www.pleco.com/manual/unihan.html#AlphabeticalListing

for a description of each of the fields accessed through that screen.