Pairw ise Variability Index: Variability Index: Pairw ise Evaluating the Cognitive Evaluating the Cognitive Difficulty of Using Mobile Text Difficulty of Using Mobile Text Entry Systems Entry Systems Frode Eika SANDNES Frode Eika SANDNES Faculty of Engineering Faculty of Engineering Oslo University University College, Oslo, College, Oslo, Norw ay Norw ay Oslo Hua - -Li JIAN Li JIAN Hua Faculty of Arts Arts Faculty of National Cheng Kung Kung University University, , Tainan Tainan, Taiw an , Taiw an National Cheng
Evaluating mobile mobile text entry text entry Evaluating • Performance oriented • Performance oriented – Inter keystroke delays Inter keystroke delays – – Characters – Characters per per second second/ /minute minute – KSPC ( KSPC (keystrokes keystrokes per per characters characters) ) – – … … – • Error oriented • Error oriented – Error Error rate rate – – … … – • • Cognitive Cognitive – NASA TLX ( NASA TLX (task load index task load index ) ) – – … … –
Cognitive processing and and Cognitive processing typing rhythm typing rhythm • Some relationship betw een typing speed, speed, • Some relationship betw een typing error rate and rate and cognitive difficulty of typing cognitive difficulty of typing error task task • Duration of inter- -keystroke delay related keystroke delay related to to • Duration of inter cognitive processing load load cognitive processing • Regular typing rythm - -> > cognitively cognitively • Regular typing rythm balanced text entry strategy balanced text entry strategy • Use median median based measures based measures to to avoid avoid • Use outliers – – thought w onders thought w onders and and resting resting outliers breaks breaks
The pairw ise variability index The pairw ise variability index − d d + = i 1 i pvi + i d d + i 1 i Select the median of The difference betw een all the pvi i tw o consecutive values timestamps d i = b i+1 – b i
Experiment Experiment • Subjects Subjects: : • – 3 3 volunteer volunteer undergraduate undergraduate students at OUC students at OUC – • Equipment Equipment : : • – Desktop Desktop computer computer w ith w ith full full keyboard keyboard, , – monitor, mouse mouse monitor, • 5 5- -keys pammed keys pammed to to keys keys a, s, d, f and [ a, s, d, f and [space space] ] • – Text entry implemented Text entry implemented as as apples apples, , – running in in brow ser brow ser running • Procedure Procedure for for each method each method • – 5 5 minutes practice minutes practice – – 15 15 minutes typing session minutes typing session ( (source text source text – on screen) screen) on – questionnaire questionnaire – • Measurements Measurements • – Timestamped keystroke events Timestamped keystroke events –
3 multi multi- -keystroke keystroke 3 strategies strategies • Multitap Multitap ( (similar similar to old to old • handsets) handsets) • Tree Tree - -based based • • One One - -stroke ( stroke (similar similar to T9) to T9) •
3 multi multi- -keystroke keystroke 3 Key-1: a, b, c, d, e, f strategies strategies Key-2: g, h, i, j, k, l Key-3: m, n, o, p, q, r Key-4: s, t, u, v, w , x, y, z • Multitap Multitap ( (similar similar to old to old • Key-5: [break-key] handsets) handsets) • Tree Tree - -based based Characters retreived by • repeadedly pressing the • One One - -stroke ( stroke (similar similar to T9) to T9) • key labelled w ith the desired character until it appear, then the break key.
Step 1: Chose category 3 multi multi- -keystroke keystroke 3 Key-1: a, b, c, d, e Key-2: f, g, h, i, j strategies strategies Key-3: k, l, m, n, o Key-4: p-q, r, s, t, u, • Multitap Multitap ( (similar similar to old to old • Key-5: v, w , x, y-z, [ ] handsets) handsets) Step 2: Chose w ithin cat. • Tree Tree - -based based • Key-1: a • One One - -stroke ( stroke (similar similar to T9) to T9) • Key-2: b Key-3: c Key-4: d Key-5: e
3 multi multi- -keystroke keystroke 3 strategies strategies Key-1: a, b, d, e • Multitap Multitap ( (similar similar to old to old • Key-2: f, g, h, i, j ,k, l, m Key-3: n, o, p, q, r handsets) handsets) Key-4: s, t, u, v, w , x, y, z • Tree Tree - -based based • Key-5: [space] • One One - -stroke ( stroke (similar similar to T9) to T9) • Key w ith desired character once. Word appear after w ord completed– or resolve ambiguities
Results Results Subject measure MultiTap Tree-based One-stroke Subject 1 Median ikd 0.5 1.4 1.2 Mean ch/min 22.5 13.0 28.5 Median npvi 0.29 0.55 0.41 Preverence 3/5 0/5 4/5 Subject 2 Median ikd 0.52 1.02 1.94 Mean ch/min 27.2 18.6 31.1 Median npvi 0.25 0.46 0.32 Preverence 1/5 5/5 4/5 Subject 3 Median ikd 0.24 2.13 0.55 Mean ch/min 26.5 7.7 26.2 Median npvi 0.39 0.49 0.46 Preverence 4/5 0/5 3/5
Results Results Subject measure MultiTap Tree-based One-stroke Subject 1 Median ikd 0.5 1.4 1.2 Mean ch/min 22.5 13.0 28.5 Median npvi 0.29 0.55 0.41 Preverence 3/5 0/5 4/5 Subject 2 Median ikd 0.52 1.02 1.94 Mean ch/min 27.2 18.6 31.1 Median npvi 0.25 0.46 0.32 Preverence 1/5 5/5 4/5 The tree based method Subject 3 Median ikd 0.24 2.13 0.55 is the slow est method, Mean ch/min 26.5 7.7 26.2 the least popular and the one w ith the Median npvi 0.39 0.49 0.46 largest npvi value Preverence 4/5 0/5 3/5
Results Results The one-stroke method is the fastest to use Subject measure MultiTap Tree-based One-stroke and therefore prefered Subject 1 Median ikd 0.5 1.4 1.2 by subects, but the Mean ch/min 22.5 13.0 28.5 high npvi value Median npvi 0.29 0.55 0.41 indicate cognitive difficulty Preverence 3/5 0/5 4/5 Subject 2 Median ikd 0.52 1.02 1.94 Mean ch/min 27.2 18.6 31.1 Median npvi 0.25 0.46 0.32 Preverence 1/5 5/5 4/5 Subject 3 Median ikd 0.24 2.13 0.55 Mean ch/min 26.5 7.7 26.2 Median npvi 0.39 0.49 0.46 Preverence 4/5 0/5 3/5
Results Results Subject measure MultiTap Tree-based One-stroke Subject 1 Median ikd 0.5 1.4 1.2 Mean ch/min 22.5 13.0 28.5 Median npvi 0.29 0.55 0.41 Preverence 3/5 0/5 4/5 Subject 2 Median ikd 0.52 1.02 1.94 Mean ch/min 27.2 18.6 31.1 Median npvi 0.25 0.46 0.32 The MultiTap method Preverence 1/5 5/5 4/5 is easy to use and has Subject 3 Median ikd 0.24 2.13 0.55 the low est npvi value Mean ch/min 26.5 indicating cognitive 7.7 26.2 ease. But is is slow er Median npvi 0.39 0.49 0.46 to use than one-stroke Preverence 4/5 0/5 3/5
Conclusions – – median median npvi npvi Conclusions • Useful Useful measure measure for for comparing various comparing various • text entry strategies text entry strategies • Gives Gives an an indication of cognitive indication of cognitive • difficulty difficulty • Can Can be be applied directly applied directly to to keystroke keystroke • logs, independent , independent of strategy of strategy logs • No No normalisation normalisation or or scaling necessary scaling necessary • • Easy Easy to to compute compute • • Roboust Roboust to to outliers outliers • • Should Should not be not be used exclusively used exclusively, , but but in in • conjunction w ith other quantities conjunction w ith other quantities
you!!! !!! Thank you Thank
Recommend
More recommend
