Top Banner
Niels Henze University of Oldenburg Enrico Rukzio University of DuisburgEssen Susanne Boll University of Oldenburg 100,000,000 Taps Analysis and Improvement of Touch Performance in the Large
34

100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

Jul 09, 2015

Download

Technology

Niels Henze

Touchscreens became the dominant input device for smartphones. Users' touch behaviour has been widely studied in lab studies with a relative low number of participants. In contrast, we published a game in the Android Market that records the touch behaviour when executing a controlled task to collect large amounts of
touch events. Players' task is to simply touch circles appearing on the screen. Data from 91,731 installations has been collected and players produced 120,626,225 touch events. We determined the error rates for different target sizes and screen locations. The amount
of data enabled us to show that touch positions are systematically skewed. A compensation function that shifts the users' touches to reduce the amount of errors is derived from the data and evaluated by publishing an update of the game. The independent-measures experiment with data from 12,201 installations and 15,326,444 touch events shows that the function reduces the error rate by 7.79%. We argue that such a compensation function could improve the touch performance of virtually every smartphone user.
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

Niels  Henze  University  of  Oldenburg  

 Enrico  Rukzio  

 University  of  Duisburg-­‐Essen    

Susanne  Boll    University  of  Oldenburg  

100,000,000  Taps    

Analysis  and  Improvement  of  Touch  Performance  in  the  Large  

Page 2: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

Touch  me  How  can  it  be  Believe  me  The  sun  always  shines                      on  my  screen  

Page 3: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     but  sJll...  

 we  cannot  see  what  we  touch    fingers  are  bigger  than  the  elements  we  touch  

Page 4: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Work  about  mobile  

touchscreens?    special  interacJon  techniques

       

Page 5: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Work  about  mobile  

touchscreens?    special  interacJon  techniques    performance  for  different  target  sizes  and  posiJons          

Y.  S.  Park,  S.  H.  Han,  J.  Park,  Y.  Cho:  Touch  key  design  for  target  selecJon  on  a  mobile  phone.  Proc.  MobileHCI,  2008.  

Page 6: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Work  about  mobile  

touchscreens?    special  interacJon  techniques    performance  for  different  target  sizes  and  posiJons    taps  might  be  systemaJcally  skewed              

Y.  S.  Park,  S.  H.  Han,  J.  Park,  Y.  Cho:  Touch  key  design  for  target  selecJon  on  a  mobile  phone.  Proc.  MobileHCI,  2008.  

Page 7: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Degrees  of  freedom  

 different  sizes          

Page 8: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Degrees  of  freedom  

 different  sizes    different  posiJons      

Page 9: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Degrees  of  freedom  

 different  sizes    different  posiJons    different  devices      

Page 10: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Degrees  of  freedom  

 different  sizes    different  posiJons    different  devices    …      

Page 11: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

   Large  amount  of  

touch  data    game  published  on  the  Android  Market  

   

Page 12: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

   Large  amount  of  

touch  data    game  published  on  the  Android  Market    we  inform  the  player  about  the  study  

 

Page 13: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

   Large  amount  of  

touch  data    game  published  on  the  Android  Market    we  inform  the  player  about  the  study    just  looks  like  an  ordinary  game  

   

Page 14: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

   Large  amount  of  

touch  data    game  published  on  the  Android  Market    we  inform  the  player  about  the  study    just  looks  like  an  ordinary  game    parJcipants  get  some  introducJon  

   

Page 15: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

   Large  amount  of  

touch  data    game  published  on  the  Android  Market    we  inform  the  player  about  the  study    just  looks  like  an  ordinary  game    parJcipants  get  some  introducJon  

   

Page 16: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

   Large  amount  of  

touch  data    game  published  on  the  Android  Market    we  inform  the  player  about  the  study    just  looks  like  an  ordinary  game    parJcipants  get  some  introducJon    they  tap  the  targets  

   

Page 17: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

   Large  amount  of  

touch  data    game  published  on  the  Android  Market    we  inform  the  player  about  the  study    just  looks  like  an  ordinary  game    parJcipants  get  some  introducJon    they  tap  the  targets    we  vary  targets’  size  and  posiJon  

   

Page 18: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     We  collected  

 data  from  91,731  installaJons    

   

Page 19: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     We  collected  

 data  from  91,731  installaJons    >  50  different  devices          

The  ten  most  common  devices  

Page 20: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     We  collected  

 data  from  91,731  installaJons    >  50  different  devices    120,626,225  touch  events          

Page 21: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Using  the  data  

 we  determined  the  error  rate  for  different  sizes        

Page 22: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Using  the  data  

 we  determined  the  error  rate  for  different  sizes    and  the  error  rate  for  different  posiJons  

error  rate  

HTC  Wildfire  

Page 23: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     SystemaJc  skew  

 found  that  taps  are  systemaJcally  skewed  

     

HTC  Wildfire  

Page 24: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     SystemaJc  skew  

 found  that  taps  are  systemaJcally  skewed    determined  offsets  for  each  screen  posiJon  

     

Samsung  Galaxy  S  

Page 25: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     SystemaJc  skew  

 found  that  taps  are  systemaJcally  skewed    determined  offsets  for  each  screen  posiJon    derived  how  to  shid  taps  to  decrease  error  rate  

     

devices  with  480x800  pixels  resoluJon  

Page 26: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     SystemaJc  skew  

 found  that  taps  are  systemaJcally  skewed    determined  offsets  for  each  screen  posiJon    derived  how  to  shid  taps  to  decrease  error  rate    approximate  a  smooth  shid  funcJon  

     

devices  with  480x800  pixels  resoluJon  

Page 27: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Experimental  

validaJon    implemented  shid  funcJon  for  4  resoluJons  

control  condiJon   experimental  condiJon  

Page 28: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Experimental  

validaJon    implemented  shid  funcJon  for  4  resoluJons    published  an  update  of  the  game    

control  condiJon   experimental  condiJon  

Page 29: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Experimental  

validaJon    implemented  shid  funcJon  for  4  resoluJons    published  an  update  of  the  game    randomly  assigned  players  to  a  condiJon      

control  condiJon  nc=6062  

experimental  condiJon  ne=6139  

Page 30: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Shid  funcJon  

 decreases  the  error  rate  

Page 31: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Shid  funcJon  

 decreases  the  error  rate    increases  the  precision          

Page 32: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

     Shid  funcJon  

 decreases  the  error  rate    increases  the  precision    for  all  considered  resoluJons          

Page 33: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

   

Summary  interacJve  task  

implemented  in  a  game  touch  data  from  91,731  

installaJons  determined  error  rates        and  systemaJc  skew  

derived    shid  funcJon  validated  in  a  controlled  

experiment        

 

   LimitaJons  know  ligle  about  the  players  experimental  manipulaJon  only  has  a  small  effect  unlikely  that  the  funcJon  works  for  other  tasks  

Page 34: 100,000,000 Taps: Analysis and Improvement of Touch Performance in the Large

Niels  Henze  niels.henze@uni-­‐oldenburg.de  

University  of  Oldenburg    

Enrico  Rukzio  enrico.rukzio@uni-­‐due.de    

University  of  Duisburg-­‐Essen    

Susanne  Boll  susanne.boll@uni-­‐oldenburg.de  

University  of  Oldenburg  

100,000,000  Taps    

Analysis  and  Improvement  of  

Touch  Performance  in  the  Large  

ContribuJon    

“App  Store  study”  to  build  a  model  for  an  interacJve  task  

public  between-­‐groups  experiment  for  the  task  

compensaJon  funcJon  for  touches  beyond  specific  

devices,  users,  and  contexts  

Interested  in  the  data?  hgp://nhenze.net/?page_id=673