Applying Mobile Device Soft Keyboards to Collaborative Multitouch Tabletop Displays
1. APPLYING MOBILE DEVICE SOFT
KEYBOARD TO COLLABORATIVE
MULTITOUCH TABLETOP DISPLAYS :
DESIGN AND EVALUATION
SungAhn Ko, KyungTae Kim, Tejas Kulkarni and
Niklas Elmqvist
Purdue University, USA
5. Do you see what I see…
I’m a tabletop which is
hidden under
keyboards…
6. RELATED WORK
Layouts
– Quikwriting( Perline et al. ’98), Cirrin (Mankoff et al. ’98)
Performance Measure
– In search of effective text input interfaces for off the desktop computing ( Zhai et al. ’05)
– Evaluation of Soft Keyboard ( Soukoreff and MacKenzie , ‘95, ’99)
Tabletop Text Entry
– Examination of text-entry methods for tabletop displays ( Hinrichs et al. ‘07)
– SLAP widgets ( Weiss et al. ‘09)
– Typing on flat glass : examining ten-finger expert, typing patterns on touch surfaces( Findlater et al. ‘11)
7. DESIGN CRITERIA
Minimize Space
Minimize Occlusion
Maximize Robustness
13. INITIAL USER STUDY
• 15 Participants (11 males, 4 females)
• Age : 21 – 29 ( avg. 24.9, median 25)
• Normal or corrected-to-normal vision
• Experienced, not professional
• Duration : 1 hour
15. TASKS
Apparatus : 1.4m x 0.9m (81”) FTIR multitouch tabletop
display(1280x800), Microsoft Windows
Task 1 : Users are not allowed to correct errors
Task 2 : Users are force to correct errors.
Phrase : Standard Phrase*
Keyboards : Soft, Radial and Pinpoint
Size : Small, Medium and Large
15 participants x 3 keyboard types x 3 size allocations x
2 tasks x 1 rep = 270 total trials
* Proposed by MacKenzie and Soukoreff CHI ’03
18. FOLLOW-UP STUDY
Comparison between Pinpoint Keyboard and Soft Keyboard, SHIFT
technique
Keyboards : Soft, Pinpoint Soft with Shift
Task : Participants can make mistakes and correct theme as they see it
15 participants x 3 keyboard types x 12 repetitions = 540 total trials
20. DESIGN IMPLICATIONS
Standard QWERTY keyboards are the fastest in general.
Radial Keyboards may not transfer well to tabletops.
Pinpoint was promising, but still slower than QWERTY.
The Shift technique did not have a significant impact.
21. GENERALIZING RESULTS
Radial and pinpoint typing did not outperform soft keyboards. But!
Hybrid of pinpoint and soft keyboard is possible.
Occlusion reduction technique does not yield lower error rate,
especially when users have to type fast.
22. GENERALIZING RESULTS
Radial and pinpoint typing did not outperform soft keyboards. But!
Hybrid of pinpoint and soft keyboard is possible.
Occlusion reduction technique does not yield lower error rate,
especially when users have to type fast.
23. GENERALIZING RESULTS
Radial and pinpoint typing did not outperform soft keyboards. But!
Hybrid of pinpoint and soft keyboard is possible.
Occlusion reduction technique does not yield lower error rate,
especially when users have to type fast.
24. CONCLUSION
Evaluated popular soft keyboard-based text entry
method with two user studies
QWERTY soft keyboards are significantly faster.
A lot of future work needs to be done. Such as,
applying dictionary-based methods, click-to-tag
annotation, and possible speech input.
Additional applications of pinpointing.
You see a lot of keyboard layouts there...these are the different types of keyboards which has been proposed to cell phones, typically smart phones with touch screens. \nSo when it comes to touch screens, it seems like qwerty layout is the winner for alphabet.\n
While admitting qwerty’s dominating advantage on smart phone, familiarity, we started to question “isn’t it too big?” Because it actually covers the most of the screen space.\n\nWhen we type on smart phones, the keyboard layout hides the most of the screen contents, which does not actually hinder users, because the users brought up the keyboard intentionally which indicates that they are aware of context behind the keyboard.\n\nOur study was motivated by asking “Will there be any problem when applying this kind of text-entry method to tabletop display when multiple users are sharing the same screen?” \nWhen a user brings up the software keyboard, it will inevitably interferes others space. The finest advantage of tabletop display environment is that we can incorporate more users and more collaboration, but that will only make the problem worse. \n
This is a situation where four people are on the table collaborating, and they all brings up the keyboard to type something, and all of sudden, the keyboards hides the message from the tabletop display. If the keyboards were small, the message will be shown but will yield lots of error. \n\nSo, our study is to answer the question “does size matter” and “which layout or input technique is suitable for tabletop display.\n
Quikwriting and Cirrin both took advantage of radial layout. \n\nIn search of effective text input interface for off the desktop computing ... measuring the speed of hand writing voice recognition,gesture, stylus keyboard\nSoukoreff and MacKenzie 95 99 suggested standard way of measuring speed and error rate which we adapted.\nExamination of text entry : explored different types of text entry method for tabletop display in terms of physical and software\nSLAP widgets : used silicone rubber and markers to give tactile feedback which can be used in various way\nAnd very recently Findlater : study about whether giving visual feedback or displaying keyboard would help user or not\nOur contribution is not providing a novel layout or a method, but to incorporate existing technique which is applied to cell phones and evaluate on the tabletop display.\n
We decide three designed three criteria\nFirst factor is space. multiple users mean that the text input method should consume a minimum of space on the surface\nSecond factor is occlusion, fat finger problem\nThird, robust to calibration error.\n
highlight the touched key but useless when its small\n
draw a path instead of go tap\nbut in alphabetical order \n
During pilot study\n\ngenerated idea of why don’t we select the key outside of the keyboard\n
no Y axis occlusion\n
Did not consider vocabulary based correction typing system.\n\nFully understood users\nTwo were left-handed, all others right-handed.\n
\n
Size : important aspect\nsmall 58cm^2, 175^2 525^2 from pilot study.\niphone3 soft keyboard uses 240x320 pixel soft keyboard in vertical mode while small size in they was 130x50. Not completely fair to radial keyboard.\nRandomized\n
T1 SK is significantly faster than PK and RK. PK is faster than RK....no significance error rate.\n
MSD error rate\n\nerror rate of PK stays while other grows. reason why still SK is faster is DEL is cheap\n\nvery low error rate on small keyboard....entering password?\n
We removed radial keyboard. Smaller sized was tested but not useful.\nParticipants\nFifteen paid adult volunteers (10 males, 5 females) participated in the experiment. \n\nNo participant in this experiment was also a participant in the previous experiment. \n\nAges\nranged from 22 to 27 (average 24.5, median 25). All participants\nwere experienced, although not professional, typists.\nThree participants were left-handed.\n\n\nInstead, participants\ncan make mistakes and correct them as they see fit—the\nerror metric can cope with both fixed and unfixed errors. We\nused the same phrases as before\n\n\nSoft Keyboard (SK): The standard virtual keyboard used\nin the previous experiment (Figure 1(a)).\n Pinpoint Keyboard (PK): The pinpoint keyboard used in\nthe previous experiment (Figure 1(c)).\n Soft Keyboard w/ Shift (SH\n\n\n15 Participants\n 3 Keyboard Types K (Standard, Pinpoint, Shift)\n 12 Repetitions\n540 Total trials.\n
Surprisingly, SK and SH lower error rate because of \n
Main difference between tabletop mobile display and tabletop is DPI. We are using 1280x800, some of cell phone has 720p resolution. \nStandard QWERTY soft keyboards allow for the fastest\ntyping speed on tabletops in general;\n\nRadial keyboards optimized for mobile devices may not\ntransfer well to tabletops without further design iterations;\n Pinpoint typing was more promising than radial keyboards\nbut did not outperform soft keyboards; and\n The Shift [36] occlusion-reduction technique did not have\na significant impact on error rate.\n\n\nIn general, standard QWERTY soft keyboards\nare faster because of their familiarity, the possibility\nto use both hands, and the fact that mistakes are cheap.\n\nWe adapted radial keyboards from mobile computing, but our\nresults indicate that the method is unsuitable for tabletops.\nRadial keyboards were significantly slower than other techniques,\n\nresults suggest that pinpoint typing,\nwhile somewhat promising due to its low error rate in the\ninitial user study, is still not a viable alternative to standard\nsoft QWERTY keyboards on tabletop displays—not even on\nsmall size allocations. This seems to indicate that our initial\ndesign constraints did not have such a significant impact on\ntabletop text entry as we had anticipated.\n\nApple iPhone, did not yield significantly lower error rate or\nhigher speed in the follow-up user study\n
Cellphone ... why not on tabletop\n\nHybrid for password\n
Cellphone ... why not on tabletop\n\nHybrid for password\n