RECENT TOPICS » View all
Shinkanji-search-free for iphone
Index » Learning resources
- « Previous
- 1
- Next »
Hello,
this is to announce a few things.
1) since the last time I have written in this forum I have made multiple improvements to Shinkanji. In particular I have added a new handwriting recognition method for searching a kanji. This is (in my opinion) more accurate than the one provided by apple and it has several other avantages other it: Larger area to draw (important for finger drawing), stroke undo redo (important since stroke order is taken into account like in most online recognition engines including the one used by apple), more suggested results
2) a new free demo apps has been released called Shinkanji-search-free. This contains all the search methods of Shinkanji but you can only see the english meanings of the found kanji as you can't get to the main kanji description screen with readings, stroke order diagrams and structural navigation. The intent is that between this and Shinkanji-lite, all functionalities of Shinkanji can be tried out for free while keeping an incentive to try the pay version. In particular the handwriting recognition I am talking about in 1) can be tried out for free.
If any of you wants to give me feedbacks on the handwriting recognition (which can be tried for free, see point 2)) please do so.
Best regards
Benoit
http://shinkanji.bcerrina.com
Thank you Benoit. I bought ShinKanji several months back, and hadn't opened it in a while. I was blown away by the updates. I really appreciate the continued effort to improve the product (at no extra cost to me).
While in Japan recently, I used the handwriting recognition to decipher a few vaguely familiar kanji, and was able to come up with the correct reading. My kanji fingerpainting is awful, so I am impressed. On some occasions, it took more than one or two tries, but overall it works. Very much worth the price for the iphone user.
The handwriting recognition looked awesome, but it had a lot of trouble recognizing a great majority of the characters I put in... I started with 鑿, which, granted, is both rare and rather complex, and went from there. Even simple characters like 壊 and 爪 either required fiddling with or were too far down the list to make any sense. My current 電子辞書 recognizes my stylus written 鑿 first try, without a problem-- that's on a tiny tiny, 2cm x 2cm writing surface.
I know you probably don't want to hear this, but the UI is rather off-putting. The radical search is horrible. Go take a look at the iphone version of Jisho.org for an idea of how to redo the radical selection interface. I've never been a fan of the huge scrolling wheels in the iPhone UI and the radical search interface is definitely a place NOT to use them.
The wheel works OK in the meaning search. In fact, I like it better than the way results are displayed in Kotoba or 日本語 (because it seems faster), but the semi-transparent grey back is odd and un-iphone like.
In Shinkanji Lite, it blows my mind that you opted for scrolling only a section of the screen, rather than just doing a full width top bar (non-floating) and rendering the information display as a full scrolling page (think safari).
What would make Shinkanji even better (and I know you've probably heard this before) is say, a list of common compounds along with the readings derived from edict and/or JLPT lists, along with a definition. It's somewhat of a waste of time to look up a kanji individually and then have to close Shinkanji and open another dictionary to actually find the meaning of the word you're trying to decipher.
Last edited by FutureBlues (2009 January 26, 11:57 pm)
Hello,
thank you for your feedbacks. I want to start my reply by saying that I DO want to hear UI feedbacks as it is the only way I can improve it. Now here for the detailed answer.
I have to say that I have not updated the free version at the same speed as the for pay version.
In particular regarding the click wheel, this is not used anymore, I now present the result in a table and tapping any result automatically bring you back to the main screen.
Regarding the handwriting recognition, I have had mixed report about it most people thinking it work really well, what you have to understand is that it is using the stroke order for the recognition, if your stroke order is a little off it will probably still find the kanji (it is more forgiving than jstroke based algorithm) but if it is completely off or if your stroke number is completely off it will not be able to recognize. In the case when you cannot guess the stroke order the radical search is better. Regarding a comparison with dedicated denshi jisho, there is not much I can say except that stylus input is much much superior than touch input and that I am certain that large japanese company have better japanese handwriting recognition algorithm at their disposal than me.
On the other hand, many people (including me) think the current system works better than the one from apple if only because: 1) the area to draw is bigger ==> better for touch 2) the undo redo support (you complain about fiddling, you can't fiddle with the apple implementation and on a complex kanji it is a problem) 3) more results shown 4) it just works better. You can still use the apple input method if you want in the direct search input method though as I understand that your milage may vary.
Finally about the support for compounds, this is in work, in fact there will not be any functional updates until this is in (bug fixes will be delivered).
Benoit Cerrina
http://shinkanji.bcerrina.com
Last edited by benwulf (2009 January 27, 8:48 am)
FutureBlues wrote:
I know you probably don't want to hear this, but the UI is rather off-putting. The radical search is horrible. Go take a look at the iphone version of Jisho.org for an idea of how to redo the radical selection interface. I've never been a fan of the huge scrolling wheels in the iPhone UI and the radical search interface is definitely a place NOT to use them.
The wheel works OK in the meaning search. In fact, I like it better than the way results are displayed in Kotoba or 日本語 (because it seems faster), but the semi-transparent grey back is odd and un-iphone like.
I am interested in understanding why you think the scroll wheel works in the meaning search but not the radical search. I have tried to have a consistent interface and while I can understand that it doesn't work in all cases for displaying the kanji results, I don't understand why it would work in one case and not in others.
Regarding the way it is done in jisho.org, I would like to know which part you like best compared to shinkanji, is it that you can directly select the radical without first going to a zoomed view of the selected row? This was done after feedbacks showed that it was hard to select the correct radical directly as people have big fingers.
In there UI what I don't like is that we have to scroll way down to see the results. What I like is that the radicals which are incompatible with the current selection are deselected as we go.
I think a set of improvements which you may find satisfying (and hopefully others) are the following:
1) have an option for direct selections of the radical as opposed to the current 2 step with zoom selection
2) reduce the size for the result table list (it is not a click wheel anymore so that can be resized) and increase the size for the radical selection list table.
I still want to keep a split screen with the top used for the criteria definition and the bottom for the result list.
Benoit Cerrina
http://shinkanji.bcerrina.com
FutureBlues wrote:
In Shinkanji Lite, it blows my mind that you opted for scrolling only a section of the screen, rather than just doing a full width top bar (non-floating) and rendering the information display as a full scrolling page (think safari).
What would make Shinkanji even better (and I know you've probably heard this before) is say, a list of common compounds along with the readings derived from edict and/or JLPT lists, along with a definition. It's somewhat of a waste of time to look up a kanji individually and then have to close Shinkanji and open another dictionary to actually find the meaning of the word you're trying to decipher.
There are several reasons for this.
First the main purpose of shinkanji-lite is to show off those functionalities which are original in shinkanji. The main one is not the kanjidic information it is the stroke order diagram and the navigation within the kanji structure accessible when you tap on that diagram.
Because of that I want the diagram to be visible at all time.
Second in most case the kanji information part of the screen needs to be scrolled only to see meanings in non english languages. In the full version there is a preference setting to select which is your preffered language and if you do select it there will always be only one meaning. In this context there is almost never a need for scrolling the kanji information part
Again thanks for the feedbacks
Best regards
Benoit Cerrina
http://shinkanji.bcerrina.com
I've been playing around with the full version of Shinkanji and I have some more feedback on the interface and assorted issues.
Without a unified interface (with a top bar like Safari, Notes, most of Apple's applications, and a lot of other gems like Tweetie, etc.) canceling out of a search is confusing and unintuitive. If you try to do a search by reading, intitally you get a keyboard and a prompt. To cancel out of this, you have to tap a random spot on the top interface to slide the keyboard away and then find the "Done" button which was previously hidden by the keyboard. Even worse, if you then re-enter the reading search screen, the keyboard doesn't automatically appear, leading to further confusion and at least 1 required tap to even begin searching beyond the initial tap to get to that screen. If there was a unified top bar, you could put a back button up there at the top left and completely do away with the problem on this screen and all related search screens.
As far as entries are concerned, tapping the kanji grid seems to serve no clear purpose. If you removed the animation controls from the bottom of every screen and put them in the box that pops up when you tap to zoom on the diagram, you could use that bottom space for all the different search methods, again, in a unified bar that persists throughout screens (See Tweetie, Google, Notes, etc.).
I'm also confused by the tutorial. Sometimes it tries to take over the UI while I'm trying to look up a kanji and I get really frustrated that I can't somehow disable it.
I think we may disagree here, but I don't think the diagram is any more important than anything else as far as scrolling is concerned. If compounds are, in fact, coming to a future release, I imagine most users will be more interested in those anyway. Again, I think the whole screen (barring two unified UI bars) should scroll fluidly-- what you have now (with horizontal and vertical scrolling in a weird blended box) looks out of place and just bad, really-- especially with the drop shadow that appears when you scroll past the end.
When you search and then go to bookmark something, the bookmark screen is unintuative-- it should either pop up an option like "View Bookmarks, Bookmark, etc." as in mail (replying, etc.) or at least go to a screen that has an "Add bookmark" button enabled by default. As it stands now, you have to go to the bookmark screen, search for the new bookmark button, stop, search for the edit button, tap the edit button, tap the add button and then add a bookmark. Way too many taps to get to the default action that most people want when they hit bookmarks.
On the options screen, I don't understand what the SoD options are referring to. Is there some clearer term you can use here? I have no idea what these do. (If it's related to the grey backgrounds, they don't seem to work in realtime in the options screen). Preferences could also probably go under Settings, in order to free up screen space for that bottom bar I'm championing. There aren't a lot of settings you need to access all the time in the current prefs menu.
I haven't had a lot of time to draw in oodles of kanji to see what works and what doesn't, but I tried a few and one that I couldn't get it to recognize no matter how hard I tried to draw it was: 修. 鑿 also remains un-findable, as far as I can tell. It seemed to work pretty well for a lot of other characters though.
Last but not least, a failed direct search produces no error message. I accidentally was typing meanings into the kanji direct search and I'd press enter and be taken back to the initial screen without a single change-- no error message, no empty search, nothing. There needs to be some feedback here to prevent user error.
Well, lots of criticism, I'll have to see what can be done to address it.
1) First the question of a "unified interface" I have to say I have tried to have as consistent as possible an interface for the different input methods, we do not have something as simple as an address bar but that is because the different methods for searching kanjis are too varied. I don't think that the next critic has any relationship with this one anyway.
2) Too many clicks to cancel a search which shows a keyboard. This is correct and I think I can do something about it, I'll see to adding a cancel button in the part of the screen which is visible when the keyboard is up
3) The keyboard is not shown when the direct input method is selected if it was previously hidden. I can do something about that too
4) lack of a topbar, well I tried that in a non published (beta) version of 2.2, the feedback from all was that it was a big regression as it took too much space and required too many clicks. The interface had a menu like in shinkanji-search free as a root, you would select a search which showed the related search panel (but with a back button) selecting a kanji would push the main window with the back button to return to the search panel. People complained about having to do 3 taps instead of one to select a different search method.
5) "As far as entries are concerned, tapping the kanji grid seems to serve no clear purpose. " I don't understand what you are talking about.
Bottom bar for the search method is a problem as I already have several other buttons there. I can see if I can move the play closer to the animation control (maybe remove when the animation is not enlarged but just having a large play overlay to show that it can be tapped like movies in safari). Settings is a problem as I think putting an app settings in the main settings app is an error, experience shows that this is about the same as not having any settings as 90% of the users won't even realize that your app has some special settings. Bookmark would still need to be somewhere, next previous would still need to be somewhere, I also tried using a regular toolbar with a more button with a similar remark from my beta testers that they found it cumbersome to switch between search methods. Conclusion, you may be correct that the interface could be improved but it is much more easily said than done, believe me I have tried.
6) whole screen scrolling I'll have to experiment but you are right I am doubtfull about this, I definitely put a high importance to the diagrams. I don't know what you are talking about regardign horizontal scrolling there is none as far as I am concerned. The shadow that appear past the end is an apple thing.
7) Bookmark screen, it is probably unintuitive but not in the way you are describing, what you are talking is adding a new category and you are confused because at the beginning there is none. Adding a bookmark is one tap to a previously created category. I should create a new default category when none exists to make things clearer.
8) SoD in the option screen means Strok order diagrams I'll have to change it everybody is like you and doesn't understand
9) I'll try the 2 kanjis you mention to see what is wrong
10) I agree a message is needed for a direct search
Thank you for the usefull feedback I don't think all of your ideas will work but there is definitely room for improving several of your pain zones.
best regards
Benoit
Last edited by benwulf (2009 March 23, 2:01 pm)
The most recent upgrade is a major improvement - that being the addition of lists of kanji compounds. The lists contain quite a few compounds, with definitions and readings. When you look up a kanji, the lists are automatically displayed under the definition and readings of the kanji. And they scroll nicely.
This is a feature that many of the Japanese learning apps are sorely lacking, and it really is vital to actually learning to read. Shinkaji gets a big bump upwards in my list of favorite japanese learning apps. The radical search plus the compounds makes encountering unfamiliar kanji a pleasure rather than a struggle.
The next step from here would be sample sentences - but that is obviously a big step, and probably too much to ask for.
I am still a bit frustrated with the navigation within the app, and the fluidity of its function. I haven't mastered the bookmarks or recent lists. Keep working on the interface. Regardless, this app keeps moving in the right direction.
Hello,
Thanks for the positive comment.
Regarding the adjunction of sample sentences, that could be part of the display of the compound information (as opposed to the kanji info, sentences are to words what compounds are to individual kanji).
I think there is an available sentence corpus so I'll look into it.
I'm sorry to hear you still think the interface use is not intuitive enough and I'll look into that too. For now though the next worked planned for me is:
* Site update with new screenshots and new videos.
* DB update with more animations(you may have notice weekly updates, recently I add about 100 stroke order animations a week all of "Heisig Kanji").
* customizable background and fonts for better readability (I like it as is but I have had people saying it is not as easy to read as it could be).
* Meaning groupped by reading (all meanings do not apply to all readings)
* German kanji dictionary (will cost more unfortunately as I have to license a commercial dictionary, but it will go well with the iphone OS 3.0 in App purchase)
* User defined annotations on kanji
So you still have a lot to expect...
Best regards
Benoit Cerrina
http://bcerrina.shinkanji.com
Last edited by benwulf (2009 March 23, 2:01 pm)
Thanks, Benwulf for the reply. I'm thankful that you keep improving this app, and glad I bought it. At this point, it is the best tool I have for learning to read a new kanji when I'm out and about - where I typically encounter them. i'll have another look at the website and demos, which may help with some of my difficulties with the interface.
A new version has just been uploaded to apple, hopefully it will be in the appstore in about a week, the main improvements are increase in the dictionary size (full jmdict), speed increase, customizable background
- « Previous
- 1
- Next »