際際滷

際際滷Share a Scribd company logo

Practical trigger action programming in the smart home

J
Jehwan Seo

Practical Trigger-Action Programming in the Smart Home

1 of 30
Download to read offline
Practical Trigger- Action Programming in the Smart Home + CHI 2014 -Blase Ur / x 2014 Autumn
Practical Trigger-Action Programming in the Smart Home 2014. 10. 10 UX Lab
碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University) 豢豌 CHI '14 Proceedings of the SIGCHI Conference on Human Factors in Computing Systems
AUTHORS Blase Ur : PhD, Carnegie Mellon University Elyse McManus : Assistant Researcher, Brown University Melwyn Pak Yong Ho : Research Assistant, Brown University Michael L. Littman : Professor of Computer Science, Brown University 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
ABSTRACT . 朱 れ trigger-action (if, then) programming る 蠍郁鍵襯 譟一 蟆 ろ奄 譟一 . . trigger-action (if, then) programming 殊 郁規 谿瑚れ 豢 覈 蟲 る 蟆 谿場. . 讌 襭 trigger 覿 襾語 蟲 trigger 譬襯襯 蟲覿 . . IFTTT 螻旧 67,169 trigger-action 襦蠏碁 螻煙 螳 朱 れ れ 襷 螻 trigger-action interactions襯 燕 . . 226 谿瑚れ ろ碁ゼ 朱 覩語 る 覲旧 triggers actions 襦蠏碁 襷 覯 觜襴 覦一 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
KEYWORDS End-User Programming Home Automation Smart Home Internet of Things Condition-Action Programming 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
INTRODUCTION . 螳 覦 る 螳ロ蟆 襷 蠍一れ 郁規 讌 讌襷 蠏 蠍一れ 蟲 觜 襷 り 覲旧″ 蟆 伎. . 豕蠏 2 伎 蠍一れ 朱 れ 觜 る 蠍郁鍵れ 企螻,mass-market pervasive computing ロ 覈覃 襷り . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
INTRODUCTION . 朱語 IFTTT = if trigger, then action 詩螻 trigger-action programming企 覈覈. . ろ IFTTT 蠍磯レ 朱 れ 螳 レ襯 trigger-action programming 蟆 . Philips Hue lights Belkin WeMo family of outlets, switches, motion sensors Crowd-funded devices : Twine, WigWag 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
INTRODUCTION . る 襦蠏碁覦 蟲 ｸ炎骸 伎 譟壱襯 企至 襷豢 蟆瑚? 郁規 讌 . 朱語 IFTTT 蠍磯レ れ蟆 single trigger, single actions 襷 豬蟆 螻 . Q1) 朱 れ 襷 一煙 襦 讌 蟾? Q2) 朱 れ 襷 一煙 る０ 讌 蟾? Q3) 豢螳 襦蠏碁覦 蟆渚 覦蠖 讌 蟾? Q4) 覈語 譴 螳 暑讌 螳? . れ 3螳讌 蟯 郁規襯 牛 讌覓碁れ 譟一 . ease of use expressibility or 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
1. trigger-action programming 螳 る (smart-home behavior) 伎企讌 蠍 螳れ 譟一螳 . H1: trigger-action programming 螳 る . H2: 覈覈 る れ 覲旧 triggers or actions 襦 . . ろ 覦覯 1)Mechanical Turk (MTurk) 318覈蟆 蠏碁れ 螳 る 蟆れ 5螳讌 讌 . . ろ 蟆郁骸 1)豢 伎れ trigger-action programming朱 蟲 螳ロ 覲伎. 2) 伎 襦蠏碁覦 蟆渚 襦蠏碁覦 螻 蠏碁れ 螻 蟆 郁 . 3)IFTTT affordances 襴 襦蠏碁覦 22% 1螳 伎 trigger action 蟲 . 4)谿瑚れ triggerれ 3螳讌 豢 襦 螻, 螳 machine learning natural application. INTRODUCTION 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
2. 襷豢ろ る 襦蠏碁覦 襦 讌 test 讌 . H3: れ襦, れ れ 麹 覦朱 triggers and actions 蟆壱 蟆企. . ろ 覦覯 1)MTturk れ 伎 襷 麹 譟壱 覦蟆 . 2)豕譬 (End User)螳 IFTTT 螻旧 67,170螳 襦蠏碁 蠍 覈. . ろ 蟆郁骸 1)IFTT 讌 襦蠏碁 覿 蟆 れ焔れ 覦蟆 . INTRODUCTION 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
INTRODUCTION 3. trigger-action programming 煙 Test襯 讌 . H4: 谿瑚れ simple or complex interface襯 燕 ろ朱襯 覲伎 蟆企. H5: 伎 襦蠏碁覦 蟆渚螻 trigger-action programming 蟆渚 ろ朱 襯 讀螳. . ろ 覦覯 1)襦 IFTTT interaction design 覲伎譯手, 226覈 MTTurker襯 朱 10螳 襦蠏碁覦 螻殊(Task)襯 焔 . 2)谿瑚れ蟆 れ朱 single trigger and a single action (simple interface) or multiple triggers and multiple actions (complex interface) 螻殊螳 . . ろ 蟆郁骸 1)simple and complex interfaces 伎 谿瑚れ 螻 襷譟焔 谿願 覦蟆 .(supporting H4) 2)complex interface襯 谿瑚れ 觜訣 炎概襯襦 complex tasks襯 襭 . 3)H5 螳れ 覿覿朱 語 螳ロ. 1)襦蠏碁覦 蟆渚 伎 螻牛讌 . 2)谿瑚れ 豌 覯讌 螻殊 語朱, 2覯讌 螻殊 覿磯 焔レ 蠏狩. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
INTRODUCTION Overall 1)multiple triggers and multiple actions trigger-action programming る 襦蠏碁覦 れ 蠏殊 讀蟇磯ゼ 覦蟆 . 2) Trigger襯 蟲燕 覦覯訖襷 Trigger螳 語 machine learning 蠏殊 蟲 蟆 殊襯 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
BACKGROUND AND RELATED WORK Trigger-action Programming . IFTTT 69螳 るジ 豈 螻牛讌襷, single trigger -> single action 襦蠏碁覦 れ 覲語朱 . . 2螳讌 覯 prototype interface襯 襷れ 螻 . 1) one trigger and one action channel襷 simple interface 2) multiple trigger and action channels complex interface ( 覈 triggerれ and 襦 蟲焔 讌) 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
Related Work . る 豕譬 (end-user) 襦蠏碁覦 伎 郁規 譟一 Newman: 豕譬 蟲 ルレ る 襴貅伎 る手 譯殊 螻, 貉るる磯ゼ 牛 螻旧 旧旭レ 麹誤. Newman, M. W. Now were cooking: Recipes for end-user service composition in the digital home. Position Paper CHI 2006 Workshop IT@Home, 2006. Dey et al: 語 襦蠏碁 蟯 20覈 谿瑚れ 誤磯薫 . trigger-action programming螻 螳 if-then 蠏豺 蠍磯 螳 朱 覃 覈語 覦蟆 . Dey, A. K., Sohn, T., Streng, S., and Kodama, J. iCAP: Interactive prototyping of context-aware applications. In Proc. Pervasive (2006) . trigger-action programming 轟煙 螻旧 る 誤壱伎 伎 郁規 譟一 . trigger-action programming 譴 碁 れ 螳 伎 郁規 譟一 . trigger-action programming 覓語 覿螳狩蟆 覦 螳 覯蟾 蟆 企給る手 譯殊ロ 伎 郁規 譟一 BACKGROUND AND RELATED WORK 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
STUDY.1 . 覓朱Μ 蠍郁鍵襯 貉碁， trigger-action programming れ 譟一 . 螻殊螳 trigger-action programming朱 蟲 螳ロ讌 覿 . 螳螳 るジ trigger actionれ 蟆壱 覦覯 螳讌襯 (れ煙 る trigger-action programming 豕譬 蟆 り 覺) Methodology 1)る 螻 ろ覃伎 譟一襯 MTurk workerれ 覈讌 . 2)MTurk workerれ蟆 讌覓語 讌. 3)谿瑚れ 覦 trigger-action programming 蟯 伎 る企 覦覯 れ殊 . 轟 襦蠏碁覦 れ 譯 麹螻 ル碁 蠍磯ルれ 讌 蟾 蟠蠍. 4)谿瑚れ 覦 trigger-action programming 譴觜螳 襦 ろ覃伎 襯 覲伎 譴 襷れ 覩碁 讌レ (る 襦覺 讌) 讌 蟾 蟇煙 . . (1) 襷 願 覲給襯 覦れ 蟇語願る 願 蟇語願 蟆暑 覿 れ 蟆 企. (2) 襷 願 讌 螳覃 覓手唄れ 蟶殊 蟆企. (3) 願 ろ襷企 り螳 覓殊 螻 苦 (蠏碁Μ螻 願 讌螳覃 れ ろ襷企螳 貅讌). (4) 壱ｸ覓殊 る 蟆 れ. 襷 轟 讌 誤磯血朱 覈 郁屋 伎 企 蟆 讌襯 る, 轟 讌 朱 蟆 れ 螳讌螳 覘? 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
STUDY.1 Methodology 5) 谿瑚れ , 焔螻 襦蠏碁覦 蟆渚豺(覓 蟆渚 / 所 蟆渚 / 蟆渚) 覲企ゼ 讌. 谿瑚れ 瑚規 糾 轟煙 磯 殊 蟆語 蟠蠍蠍 覓. 6) 覈 襴曙 貊れ 襾殊 譯殊牡 螳 覓伎語 蟲覿 . 7)Pilot-study 一危磯 覿 螻給朱 螳覦 豺危螻襴 譴 . 襴曙朱 蟆郁骸 89.1% 螻 襾語 蟆れ 伎 殊襯 讌 覈 危れ 伎 蟆 殊襯 覺. Programming : 豕譬 螳 るジ 覦る 譟壱 蟆 螻ろ 蠍磯蓋 蠍磯レ 譟壱 (ex: 覦蟾レ 企讌覃 朱 る企煙 貅) Self-regulation : 讌 朱 碁襯 誤螻 譟一襯 豬. (ex: 願 誤 讌 襯 譟一企) Remote control : 豕譬 れ譴襷 蟆 讀 蠍郁鍵襯 貉碁， 蠍磯ゼ . (ex: 佒 覯殊 覿 貅) Specialized functionality : 襦蠏碁覦 讌襷 轟 願 . 豕譬 螳 覈 . (ex: 豺 襷 蠍郁) 8) 21螳 trigger 豈螻 46螳 action channel襦 企讌 貊覿 螳覦 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
STUDY.1 Result . MTturk る れ 覦朱 る れ 覿覿 襦蠏碁覦 螳ロ 蟆. . 襦蠏碁覦 螳ロ 覈 れ multiple triggers and actions 讌蠍磯 る trigger-action programming朱 覈 蟲 . <螳 覯譯殊 れ 觜 蟆郁骸> 1) Programming(62.6%) : 螳蟾 覩碁 襦蠏碁覦 I want the fan in my room to turn on when it is hot. Notify me if my pet gets out of the backyard. 2) Remote control(13.5%) : る誤, 貉危, , れ襦 螳 蠍郁鍵襯 貉碁， る 蟆れ. start the coffee pot from my bedroom, lock and unlock the doors on command with your voice, 3) Specialized functionality(10.8%) : 譟伎讌 蟆 YouTube on bathroom mirror, I want my mail to be delivered by a robot. 4) self-regulation(10.1%) temperatures that adjust automatically to the perfect temperature. 5) unclear(2.8%) Smart Light 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
STUDY.1 Result Trigger-action programming structure . trigger-action programming multiple triggers and multiple actions 讌伎 . (22% 襯 襦蠏碁覦 蠍 ) < 襦蠏碁覦 (trigger, action) 觜 蟆郁骸 > 1) single trigger, single action(1,1): 77.9% close the blinds when the sun is too bright call to let me know when the kids get home. 2)single trigger with multiple actions(1,2+): 5.2% 3)multiple triggers for a single action(2+,1): 8.5% 4)multiple triggers and multiple actions(2+,2+): 8.4% When I get up at night, I would want my lights to turn on and off as I enter and exit the room. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
STUDY.1 Result Triggers level of abstraction . the lowest level of abstraction : engineering sense, レ 覲(企襖 碁, 蠍郁鍵 off), dates/times, moisture, sound, and light : 谿瑚れ 讌朱 殊襯 瑚讌 蟆渚レ 覦蟆 . . Ex: 蟲郁 覦朱 蟇語 れ伎 覈 殊 蠍壱 讌 覦 讌 . . The second category : , , レ, 覓朱Μ 殊 I want the sink to turn on when I pick up my toothbrush. I would like my home to automatically clean the floors on a daily basis while no one is in the room. . The third category : 覿覿覈 trigger : Fuzzy(覿覿覈) trigger machine learning I would like to be notified when my pool chemicals drop lower than normal. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
STUDY.1 Result Diversity of behaviors . れ 螻 襯 る 螳 豕譬 襦蠏碁覦 伎 譯朱 企 蟆. . 襦蠏碁覦朱 覿襯 995螳讌 れ煙 * Good-Turing estimation 覿 讌 . 1)21螳 Trigger channels(47螳 譟壱): 伎 譟壱 伎語 Trigger 譟壱 覦 襯 2% 2)46螳 Action channels(77螳 譟壱): 伎 譟壱 伎語 Action 譟壱 覦 襯 4% 3)Trigger channel + Action Channels: 伎 譟壱 伎語 Trigger譟壱+Action 譟壱 覦 襯 13% 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University) * Good-Turing frequency estimation: 讌蠍蟾讌 譬 螳豌 覦 襯 豢 糾 蠍磯
STUDY.2 . ろ覃伎 豕譬 襦蠏碁覦 讌 襴 殊讌 讌襷, IFTTT 轟危語 朱Μ 企ロ磯れ 企 れ trigger-action programming 螻 . . れ襦 trigger-action programming 譟壱襦 危危蠍 伎 IFTTT 螻旧 襦蠏碁れ ろ. Methodology 1)IFTTT 轟危語 螻糾 67,169 朱ゼ 覈 れ 覦 2)れ 覦 螻糾 朱 譴 覓朱Μ 蠍郁鍵れ 貉碁，蠍 6螳 豈れ 貉れるゼ 螻 . : 覓朱Μ 蠍郁鍵 (blink(1), Philips Hue, Up by Jawbone, WeMo Motion, WeMo Switch, and Withings) 3)企 triggerれ 企 actionり骸 郁 讌襯 Good-Turing estimation 覿 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University) blink(1) Up by Jawbone Withings
STUDY.2 Result . 蠍壱 6螳 覓朱Μ 蠍郁鍵襯 朱れ 貉れ煙 . 1) 覓朱Μ 蠍郁鍵れ 16螳讌 襦 るジ Trigger 企欧語 18螳讌 襦 るジ Actionれ 讌 . 2) 豌 ろ 譴 513螳 (0.8%)螳 覓朱Μ 蠍郁鍵 Trigger襦 . 3) 豌 ろ 譴 858螳 (1.3%) 覓朱Μ 蠍郁鍵 Action朱 . 4) 92 case 襷 覓朱Μ 蠍郁鍵 Trigger + Action 譟壱朱 . . Good-Turing estimation 覿 1)伎 譟壱 伎語 覓朱Μ 蠍郁鍵 Trigger 譟壱 覦 襯 11% 2)伎 譟壱 伎語 覓朱Μ 蠍郁鍵 Action 譟壱 覦 襯 9% 3)蟆郁骸 襷 螳 れ ろ 蟆れ る, 蟆 1/10 觜 殊 蠍 覩 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
STUDY.3 . 襷讌襷 郁規 trigger-action programming 訖襷 瑚規 糾 襯 譟一 . . multiple triggers/actions螻 螳 蟯 蟯螻襯 覺 Methodology 1)る 蠍一 郁規 譟一襯 MTurk workerれ 螻 . 2)谿瑚れ蟆 30覿 蟇碁Μ 郁規 谿瑚 覲伎朱 碁 $2.00襯 讌覿 . 3)螳螳 谿瑚れ蟆 single trigger and single action (simple interface) triggers and multiple actions (complex interface) 襯 讌 誤壱伎るゼ 蟆 れ朱 . 4)谿瑚れ 螻殊襯 蟇碁Μ 螳螻 襦蠏碁 襯 豸′ . 5)瑚規 糾 讌覓語 牛 谿瑚れ Table.1 企 10螳 螻殊襯 蠍 誤壱伎るゼ 螻, 襷讌襷朱 *Likert-scale 襷譟焔 譟一襯 襭 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University) *Likert-scale (襴貉ろ 豌) : 覦れ 譯殊伎 覓語レ 朱 讌襯 豌 襦 轟 譯殊 覦 襯 覲企 豌.
STUDY.3 Result Task completion . 覦朱 襦蠏碁 襷 伎 simple or complex interface 炎概朱 . . Figure 8 覲 覿覿 螻殊 炎概襯 80% 伎 . . Task B 炎概襯 伎 谿瑚れ RSS 豺讌 覈詩 蟆 覓語企手 螳 . Get all updates from the website www.xkcd.com via email. . Task G H 蟲 覿螳 螻殊 磯殊 Skip 覃 炎概朱 螳譯 : Task G 蟆曙磯 殊 譟伎 or れ伎り 螳螻 螳螳 Trigger 蟆 襦 . The lighting in my bedroom should be on when I am there and off when I am not there. . Task H 蟆曙磯 貉 覲 旧 If it begins to rain then change the light colors to blue. . Task I, J Simple Interface 谿瑚 炎概襯 伎 Skip 蟆 炎概朱 螳譯朱蠍 覓語. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
STUDY.3 Result Time to complete a task . 谿瑚れ 一 螻殊襯 襭朱 谿瑚れ 焔レ 觜襯願 レ 螻朱ゼ 蟯谿壱. 1) 谿瑚れ 螻殊 A-F 襯 蠍 朱 襷 螳 蟇碁Μ讌襯 譟一 . 2) 譬 覲襦 誤壱伎 襦 覿 襭 豢 蟾讌 螳朱 狩 覈語 襷. 3) 誤壱伎る 覿螳ロ 螻殊 覈視 襭 一危磯 誤. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
STUDY.3 Result Satisfaction . 覦朱 谿瑚れ simple interface complex interface 煙 伎 襷譟炎 . 谿瑚れ interface 3螳讌 7-point 襴貉ろ 豌襦 牛. . 他 伎 伎 . 1) 朱ゼ 襷 蟆 所 讌蟯企. 2) 殊 企 譬襯 朱ゼ 襷 蟆 蟯 螳蟆 蟆企. 3) 覲旧″ 襦蠏碁覦 誤壱伎るゼ る０ 蟆企. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
Discussion . 朱 れ multiple triggers and actions 螳讌螻 炎概朱 trigger-action programming襯 蠏手碓襯 螻牛. . 蠍磯レ 谿瑚れ 襾碁Μ 覦襦 るゴ 襷 る れ 蟲 . . れ 誤 蠍 trigger-action programming 覯 襷 覓語れ 褐 蟆. . multiple trigger襯 襦襦 蟲燕 覦覯 覩瑚 覲旧″る 蟆 . . Trigger 企欧, 譟郁唄 譟壱 . . 覓瑚骸 螳 螳豌企 企欧 (the moment the door closes), 譟郁唄(is the door closed?), 蠏碁Μ螻 (close the door) 讌 蟆 . . 谿瑚れ Triggerれ 蠍 豢 覯 麹 ロ 郁規襯 . . 豢 覯讌 螻(ex: 覓殊 覓 蟆) 襾語 蠍磯朱 れ 碁襯 蠏碁 れる 襷豢 殊襯 襷 覦覯 郁規 蠍壱襯 . . る 朱 襾語 豢 覲企 讌レ朱 願屋 蟆. . 螳 豕譬 襦蠏碁覦 伎 trigger-action programming レ 螻る 螳豺螳 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
Limitations . trigger-action programming 螳螻 曙 螳 . . trigger-action programming 朱 螳 譬 覦覯企手 螳蟆 譯殊レ . . . 朱語 一危郁 覦 覈 れ 讌 . IFTTT community 朱Μ 企牛磯れ 蠍 . . IFTTT MTurk れ 朱 る慨 蠍一 燕レ 螳讌螻 . . MTurk る れ蟆 覈 覲企る 螳蠍 螳 豌 5螳讌 燕手 豌 . . trigger-action programming 覈 る 轟 覿覿 谿り 譯殊ロ . . 朱 れ trigger-action programming 譟一 郁規(field study) 螻. . 襾語 蠍磯 Trigger 殊 麹 蟯 蟆. . 郁規 讌蠍蟾讌 殊 覩碁 伎れ レ 覈襭蟆 蟆. . 襦蠏碁 蟲燕 覦覯 覃 覈語 豢螳 郁規襯 螳豺螳 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University) Thank you!!

More Related Content

Practical trigger action programming in the smart home

  • 1. Practical Trigger- Action Programming in the Smart Home + CHI 2014 -Blase Ur / x 2014 Autumn
  • 2. Practical Trigger-Action Programming in the Smart Home 2014. 10. 10 UX Lab
  • 3. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University) 豢豌 CHI '14 Proceedings of the SIGCHI Conference on Human Factors in Computing Systems
  • 4. AUTHORS Blase Ur : PhD, Carnegie Mellon University Elyse McManus : Assistant Researcher, Brown University Melwyn Pak Yong Ho : Research Assistant, Brown University Michael L. Littman : Professor of Computer Science, Brown University 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 5. ABSTRACT . 朱 れ trigger-action (if, then) programming る 蠍郁鍵襯 譟一 蟆 ろ奄 譟一 . . trigger-action (if, then) programming 殊 郁規 谿瑚れ 豢 覈 蟲 る 蟆 谿場. . 讌 襭 trigger 覿 襾語 蟲 trigger 譬襯襯 蟲覿 . . IFTTT 螻旧 67,169 trigger-action 襦蠏碁 螻煙 螳 朱 れ れ 襷 螻 trigger-action interactions襯 燕 . . 226 谿瑚れ ろ碁ゼ 朱 覩語 る 覲旧 triggers actions 襦蠏碁 襷 覯 觜襴 覦一 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 6. KEYWORDS End-User Programming Home Automation Smart Home Internet of Things Condition-Action Programming 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 7. INTRODUCTION . 螳 覦 る 螳ロ蟆 襷 蠍一れ 郁規 讌 讌襷 蠏 蠍一れ 蟲 觜 襷 り 覲旧″ 蟆 伎. . 豕蠏 2 伎 蠍一れ 朱 れ 觜 る 蠍郁鍵れ 企螻,mass-market pervasive computing ロ 覈覃 襷り . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 8. INTRODUCTION . 朱語 IFTTT = if trigger, then action 詩螻 trigger-action programming企 覈覈. . ろ IFTTT 蠍磯レ 朱 れ 螳 レ襯 trigger-action programming 蟆 . Philips Hue lights Belkin WeMo family of outlets, switches, motion sensors Crowd-funded devices : Twine, WigWag 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 9. INTRODUCTION . る 襦蠏碁覦 蟲 ｸ炎骸 伎 譟壱襯 企至 襷豢 蟆瑚? 郁規 讌 . 朱語 IFTTT 蠍磯レ れ蟆 single trigger, single actions 襷 豬蟆 螻 . Q1) 朱 れ 襷 一煙 襦 讌 蟾? Q2) 朱 れ 襷 一煙 る０ 讌 蟾? Q3) 豢螳 襦蠏碁覦 蟆渚 覦蠖 讌 蟾? Q4) 覈語 譴 螳 暑讌 螳? . れ 3螳讌 蟯 郁規襯 牛 讌覓碁れ 譟一 . ease of use expressibility or 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 10. 1. trigger-action programming 螳 る (smart-home behavior) 伎企讌 蠍 螳れ 譟一螳 . H1: trigger-action programming 螳 る . H2: 覈覈 る れ 覲旧 triggers or actions 襦 . . ろ 覦覯 1)Mechanical Turk (MTurk) 318覈蟆 蠏碁れ 螳 る 蟆れ 5螳讌 讌 . . ろ 蟆郁骸 1)豢 伎れ trigger-action programming朱 蟲 螳ロ 覲伎. 2) 伎 襦蠏碁覦 蟆渚 襦蠏碁覦 螻 蠏碁れ 螻 蟆 郁 . 3)IFTTT affordances 襴 襦蠏碁覦 22% 1螳 伎 trigger action 蟲 . 4)谿瑚れ triggerれ 3螳讌 豢 襦 螻, 螳 machine learning natural application. INTRODUCTION 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 11. 2. 襷豢ろ る 襦蠏碁覦 襦 讌 test 讌 . H3: れ襦, れ れ 麹 覦朱 triggers and actions 蟆壱 蟆企. . ろ 覦覯 1)MTturk れ 伎 襷 麹 譟壱 覦蟆 . 2)豕譬 (End User)螳 IFTTT 螻旧 67,170螳 襦蠏碁 蠍 覈. . ろ 蟆郁骸 1)IFTT 讌 襦蠏碁 覿 蟆 れ焔れ 覦蟆 . INTRODUCTION 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 12. INTRODUCTION 3. trigger-action programming 煙 Test襯 讌 . H4: 谿瑚れ simple or complex interface襯 燕 ろ朱襯 覲伎 蟆企. H5: 伎 襦蠏碁覦 蟆渚螻 trigger-action programming 蟆渚 ろ朱 襯 讀螳. . ろ 覦覯 1)襦 IFTTT interaction design 覲伎譯手, 226覈 MTTurker襯 朱 10螳 襦蠏碁覦 螻殊(Task)襯 焔 . 2)谿瑚れ蟆 れ朱 single trigger and a single action (simple interface) or multiple triggers and multiple actions (complex interface) 螻殊螳 . . ろ 蟆郁骸 1)simple and complex interfaces 伎 谿瑚れ 螻 襷譟焔 谿願 覦蟆 .(supporting H4) 2)complex interface襯 谿瑚れ 觜訣 炎概襯襦 complex tasks襯 襭 . 3)H5 螳れ 覿覿朱 語 螳ロ. 1)襦蠏碁覦 蟆渚 伎 螻牛讌 . 2)谿瑚れ 豌 覯讌 螻殊 語朱, 2覯讌 螻殊 覿磯 焔レ 蠏狩. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 13. INTRODUCTION Overall 1)multiple triggers and multiple actions trigger-action programming る 襦蠏碁覦 れ 蠏殊 讀蟇磯ゼ 覦蟆 . 2) Trigger襯 蟲燕 覦覯訖襷 Trigger螳 語 machine learning 蠏殊 蟲 蟆 殊襯 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 14. BACKGROUND AND RELATED WORK Trigger-action Programming . IFTTT 69螳 るジ 豈 螻牛讌襷, single trigger -> single action 襦蠏碁覦 れ 覲語朱 . . 2螳讌 覯 prototype interface襯 襷れ 螻 . 1) one trigger and one action channel襷 simple interface 2) multiple trigger and action channels complex interface ( 覈 triggerれ and 襦 蟲焔 讌) 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 15. Related Work . る 豕譬 (end-user) 襦蠏碁覦 伎 郁規 譟一 Newman: 豕譬 蟲 ルレ る 襴貅伎 る手 譯殊 螻, 貉るる磯ゼ 牛 螻旧 旧旭レ 麹誤. Newman, M. W. Now were cooking: Recipes for end-user service composition in the digital home. Position Paper CHI 2006 Workshop IT@Home, 2006. Dey et al: 語 襦蠏碁 蟯 20覈 谿瑚れ 誤磯薫 . trigger-action programming螻 螳 if-then 蠏豺 蠍磯 螳 朱 覃 覈語 覦蟆 . Dey, A. K., Sohn, T., Streng, S., and Kodama, J. iCAP: Interactive prototyping of context-aware applications. In Proc. Pervasive (2006) . trigger-action programming 轟煙 螻旧 る 誤壱伎 伎 郁規 譟一 . trigger-action programming 譴 碁 れ 螳 伎 郁規 譟一 . trigger-action programming 覓語 覿螳狩蟆 覦 螳 覯蟾 蟆 企給る手 譯殊ロ 伎 郁規 譟一 BACKGROUND AND RELATED WORK 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 16. STUDY.1 . 覓朱Μ 蠍郁鍵襯 貉碁， trigger-action programming れ 譟一 . 螻殊螳 trigger-action programming朱 蟲 螳ロ讌 覿 . 螳螳 るジ trigger actionれ 蟆壱 覦覯 螳讌襯 (れ煙 る trigger-action programming 豕譬 蟆 り 覺) Methodology 1)る 螻 ろ覃伎 譟一襯 MTurk workerれ 覈讌 . 2)MTurk workerれ蟆 讌覓語 讌. 3)谿瑚れ 覦 trigger-action programming 蟯 伎 る企 覦覯 れ殊 . 轟 襦蠏碁覦 れ 譯 麹螻 ル碁 蠍磯ルれ 讌 蟾 蟠蠍. 4)谿瑚れ 覦 trigger-action programming 譴觜螳 襦 ろ覃伎 襯 覲伎 譴 襷れ 覩碁 讌レ (る 襦覺 讌) 讌 蟾 蟇煙 . . (1) 襷 願 覲給襯 覦れ 蟇語願る 願 蟇語願 蟆暑 覿 れ 蟆 企. (2) 襷 願 讌 螳覃 覓手唄れ 蟶殊 蟆企. (3) 願 ろ襷企 り螳 覓殊 螻 苦 (蠏碁Μ螻 願 讌螳覃 れ ろ襷企螳 貅讌). (4) 壱ｸ覓殊 る 蟆 れ. 襷 轟 讌 誤磯血朱 覈 郁屋 伎 企 蟆 讌襯 る, 轟 讌 朱 蟆 れ 螳讌螳 覘? 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 17. STUDY.1 Methodology 5) 谿瑚れ , 焔螻 襦蠏碁覦 蟆渚豺(覓 蟆渚 / 所 蟆渚 / 蟆渚) 覲企ゼ 讌. 谿瑚れ 瑚規 糾 轟煙 磯 殊 蟆語 蟠蠍蠍 覓. 6) 覈 襴曙 貊れ 襾殊 譯殊牡 螳 覓伎語 蟲覿 . 7)Pilot-study 一危磯 覿 螻給朱 螳覦 豺危螻襴 譴 . 襴曙朱 蟆郁骸 89.1% 螻 襾語 蟆れ 伎 殊襯 讌 覈 危れ 伎 蟆 殊襯 覺. Programming : 豕譬 螳 るジ 覦る 譟壱 蟆 螻ろ 蠍磯蓋 蠍磯レ 譟壱 (ex: 覦蟾レ 企讌覃 朱 る企煙 貅) Self-regulation : 讌 朱 碁襯 誤螻 譟一襯 豬. (ex: 願 誤 讌 襯 譟一企) Remote control : 豕譬 れ譴襷 蟆 讀 蠍郁鍵襯 貉碁， 蠍磯ゼ . (ex: 佒 覯殊 覿 貅) Specialized functionality : 襦蠏碁覦 讌襷 轟 願 . 豕譬 螳 覈 . (ex: 豺 襷 蠍郁) 8) 21螳 trigger 豈螻 46螳 action channel襦 企讌 貊覿 螳覦 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 18. STUDY.1 Result . MTturk る れ 覦朱 る れ 覿覿 襦蠏碁覦 螳ロ 蟆. . 襦蠏碁覦 螳ロ 覈 れ multiple triggers and actions 讌蠍磯 る trigger-action programming朱 覈 蟲 . <螳 覯譯殊 れ 觜 蟆郁骸> 1) Programming(62.6%) : 螳蟾 覩碁 襦蠏碁覦 I want the fan in my room to turn on when it is hot. Notify me if my pet gets out of the backyard. 2) Remote control(13.5%) : る誤, 貉危, , れ襦 螳 蠍郁鍵襯 貉碁， る 蟆れ. start the coffee pot from my bedroom, lock and unlock the doors on command with your voice, 3) Specialized functionality(10.8%) : 譟伎讌 蟆 YouTube on bathroom mirror, I want my mail to be delivered by a robot. 4) self-regulation(10.1%) temperatures that adjust automatically to the perfect temperature. 5) unclear(2.8%) Smart Light 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 19. STUDY.1 Result Trigger-action programming structure . trigger-action programming multiple triggers and multiple actions 讌伎 . (22% 襯 襦蠏碁覦 蠍 ) < 襦蠏碁覦 (trigger, action) 觜 蟆郁骸 > 1) single trigger, single action(1,1): 77.9% close the blinds when the sun is too bright call to let me know when the kids get home. 2)single trigger with multiple actions(1,2+): 5.2% 3)multiple triggers for a single action(2+,1): 8.5% 4)multiple triggers and multiple actions(2+,2+): 8.4% When I get up at night, I would want my lights to turn on and off as I enter and exit the room. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 20. STUDY.1 Result Triggers level of abstraction . the lowest level of abstraction : engineering sense, レ 覲(企襖 碁, 蠍郁鍵 off), dates/times, moisture, sound, and light : 谿瑚れ 讌朱 殊襯 瑚讌 蟆渚レ 覦蟆 . . Ex: 蟲郁 覦朱 蟇語 れ伎 覈 殊 蠍壱 讌 覦 讌 . . The second category : , , レ, 覓朱Μ 殊 I want the sink to turn on when I pick up my toothbrush. I would like my home to automatically clean the floors on a daily basis while no one is in the room. . The third category : 覿覿覈 trigger : Fuzzy(覿覿覈) trigger machine learning I would like to be notified when my pool chemicals drop lower than normal. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 21. STUDY.1 Result Diversity of behaviors . れ 螻 襯 る 螳 豕譬 襦蠏碁覦 伎 譯朱 企 蟆. . 襦蠏碁覦朱 覿襯 995螳讌 れ煙 * Good-Turing estimation 覿 讌 . 1)21螳 Trigger channels(47螳 譟壱): 伎 譟壱 伎語 Trigger 譟壱 覦 襯 2% 2)46螳 Action channels(77螳 譟壱): 伎 譟壱 伎語 Action 譟壱 覦 襯 4% 3)Trigger channel + Action Channels: 伎 譟壱 伎語 Trigger譟壱+Action 譟壱 覦 襯 13% 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University) * Good-Turing frequency estimation: 讌蠍蟾讌 譬 螳豌 覦 襯 豢 糾 蠍磯
  • 22. STUDY.2 . ろ覃伎 豕譬 襦蠏碁覦 讌 襴 殊讌 讌襷, IFTTT 轟危語 朱Μ 企ロ磯れ 企 れ trigger-action programming 螻 . . れ襦 trigger-action programming 譟壱襦 危危蠍 伎 IFTTT 螻旧 襦蠏碁れ ろ. Methodology 1)IFTTT 轟危語 螻糾 67,169 朱ゼ 覈 れ 覦 2)れ 覦 螻糾 朱 譴 覓朱Μ 蠍郁鍵れ 貉碁，蠍 6螳 豈れ 貉れるゼ 螻 . : 覓朱Μ 蠍郁鍵 (blink(1), Philips Hue, Up by Jawbone, WeMo Motion, WeMo Switch, and Withings) 3)企 triggerれ 企 actionり骸 郁 讌襯 Good-Turing estimation 覿 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University) blink(1) Up by Jawbone Withings
  • 23. STUDY.2 Result . 蠍壱 6螳 覓朱Μ 蠍郁鍵襯 朱れ 貉れ煙 . 1) 覓朱Μ 蠍郁鍵れ 16螳讌 襦 るジ Trigger 企欧語 18螳讌 襦 るジ Actionれ 讌 . 2) 豌 ろ 譴 513螳 (0.8%)螳 覓朱Μ 蠍郁鍵 Trigger襦 . 3) 豌 ろ 譴 858螳 (1.3%) 覓朱Μ 蠍郁鍵 Action朱 . 4) 92 case 襷 覓朱Μ 蠍郁鍵 Trigger + Action 譟壱朱 . . Good-Turing estimation 覿 1)伎 譟壱 伎語 覓朱Μ 蠍郁鍵 Trigger 譟壱 覦 襯 11% 2)伎 譟壱 伎語 覓朱Μ 蠍郁鍵 Action 譟壱 覦 襯 9% 3)蟆郁骸 襷 螳 れ ろ 蟆れ る, 蟆 1/10 觜 殊 蠍 覩 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 24. STUDY.3 . 襷讌襷 郁規 trigger-action programming 訖襷 瑚規 糾 襯 譟一 . . multiple triggers/actions螻 螳 蟯 蟯螻襯 覺 Methodology 1)る 蠍一 郁規 譟一襯 MTurk workerれ 螻 . 2)谿瑚れ蟆 30覿 蟇碁Μ 郁規 谿瑚 覲伎朱 碁 $2.00襯 讌覿 . 3)螳螳 谿瑚れ蟆 single trigger and single action (simple interface) triggers and multiple actions (complex interface) 襯 讌 誤壱伎るゼ 蟆 れ朱 . 4)谿瑚れ 螻殊襯 蟇碁Μ 螳螻 襦蠏碁 襯 豸′ . 5)瑚規 糾 讌覓語 牛 谿瑚れ Table.1 企 10螳 螻殊襯 蠍 誤壱伎るゼ 螻, 襷讌襷朱 *Likert-scale 襷譟焔 譟一襯 襭 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University) *Likert-scale (襴貉ろ 豌) : 覦れ 譯殊伎 覓語レ 朱 讌襯 豌 襦 轟 譯殊 覦 襯 覲企 豌.
  • 25. STUDY.3 Result Task completion . 覦朱 襦蠏碁 襷 伎 simple or complex interface 炎概朱 . . Figure 8 覲 覿覿 螻殊 炎概襯 80% 伎 . . Task B 炎概襯 伎 谿瑚れ RSS 豺讌 覈詩 蟆 覓語企手 螳 . Get all updates from the website www.xkcd.com via email. . Task G H 蟲 覿螳 螻殊 磯殊 Skip 覃 炎概朱 螳譯 : Task G 蟆曙磯 殊 譟伎 or れ伎り 螳螻 螳螳 Trigger 蟆 襦 . The lighting in my bedroom should be on when I am there and off when I am not there. . Task H 蟆曙磯 貉 覲 旧 If it begins to rain then change the light colors to blue. . Task I, J Simple Interface 谿瑚 炎概襯 伎 Skip 蟆 炎概朱 螳譯朱蠍 覓語. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 26. STUDY.3 Result Time to complete a task . 谿瑚れ 一 螻殊襯 襭朱 谿瑚れ 焔レ 觜襯願 レ 螻朱ゼ 蟯谿壱. 1) 谿瑚れ 螻殊 A-F 襯 蠍 朱 襷 螳 蟇碁Μ讌襯 譟一 . 2) 譬 覲襦 誤壱伎 襦 覿 襭 豢 蟾讌 螳朱 狩 覈語 襷. 3) 誤壱伎る 覿螳ロ 螻殊 覈視 襭 一危磯 誤. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 27. STUDY.3 Result Satisfaction . 覦朱 谿瑚れ simple interface complex interface 煙 伎 襷譟炎 . 谿瑚れ interface 3螳讌 7-point 襴貉ろ 豌襦 牛. . 他 伎 伎 . 1) 朱ゼ 襷 蟆 所 讌蟯企. 2) 殊 企 譬襯 朱ゼ 襷 蟆 蟯 螳蟆 蟆企. 3) 覲旧″ 襦蠏碁覦 誤壱伎るゼ る０ 蟆企. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 28. Discussion . 朱 れ multiple triggers and actions 螳讌螻 炎概朱 trigger-action programming襯 蠏手碓襯 螻牛. . 蠍磯レ 谿瑚れ 襾碁Μ 覦襦 るゴ 襷 る れ 蟲 . . れ 誤 蠍 trigger-action programming 覯 襷 覓語れ 褐 蟆. . multiple trigger襯 襦襦 蟲燕 覦覯 覩瑚 覲旧″る 蟆 . . Trigger 企欧, 譟郁唄 譟壱 . . 覓瑚骸 螳 螳豌企 企欧 (the moment the door closes), 譟郁唄(is the door closed?), 蠏碁Μ螻 (close the door) 讌 蟆 . . 谿瑚れ Triggerれ 蠍 豢 覯 麹 ロ 郁規襯 . . 豢 覯讌 螻(ex: 覓殊 覓 蟆) 襾語 蠍磯朱 れ 碁襯 蠏碁 れる 襷豢 殊襯 襷 覦覯 郁規 蠍壱襯 . . る 朱 襾語 豢 覲企 讌レ朱 願屋 蟆. . 螳 豕譬 襦蠏碁覦 伎 trigger-action programming レ 螻る 螳豺螳 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 29. Limitations . trigger-action programming 螳螻 曙 螳 . . trigger-action programming 朱 螳 譬 覦覯企手 螳蟆 譯殊レ . . . 朱語 一危郁 覦 覈 れ 讌 . IFTTT community 朱Μ 企牛磯れ 蠍 . . IFTTT MTurk れ 朱 る慨 蠍一 燕レ 螳讌螻 . . MTurk る れ蟆 覈 覲企る 螳蠍 螳 豌 5螳讌 燕手 豌 . . trigger-action programming 覈 る 轟 覿覿 谿り 譯殊ロ . . 朱 れ trigger-action programming 譟一 郁規(field study) 螻. . 襾語 蠍磯 Trigger 殊 麹 蟯 蟆. . 郁規 讌蠍蟾讌 殊 覩碁 伎れ レ 覈襭蟆 蟆. . 襦蠏碁 蟲燕 覦覯 覃 覈語 豢螳 郁規襯 螳豺螳 . 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University)
  • 30. 碁蟲 牛螻狩蠍一 (Graduate School of Convergence Science and Technology. Seoul National University) Thank you!!