SlideShare ist ein Scribd-Unternehmen logo
1 von 16
An Adaptive Speed-Call List Algorithm and Its Evaluation with ESM Seunghwan Lee, JungsukSeo, Geehyuk Lee HCI Lab @				 Darling Mother Professor Kim This time Frequent Thursday Evening
Motivation Can a mobile phonepredictand recommenda number to call in a certain context?
Overview We … Designed a method to generate an adaptive speed-call list Online survey with 75 participants Using 3 month call log of 20 users Verified the effectiveness  Using 3 month call log of 20 users Implemented our method and conducted ESM study With 10 users for 3 months Darling Mother Professor Kim This time Frequent Thursday Evening
How do you make a phone call? Online survey with 75 participants (16-63 years old, 5+ years experience)  USER SURVEY
Any regular call pattern? 67% have a number to call regularly Parents (51%)  Lover / spouse (44%) Friends (21%),  Brothers & sisters (15%) USER SURVEY my husbandto get home every night.  my husbandto get home every night.  my colleagues to remind meeting. my lazy boyfriend for morning call. my lazy boyfriend for morning call. my friendsfor having lunch together. my parentat weekend. my parentat weekend. my friends after my work hours Chinese restaurant for food delivery every weekend my wife every lunch time. my wife every lunch time. my girlfriend while leaving my office. my girlfriend while leaving my office.
Call pattern analysis Collected 3 month call log of 20 users Call map: Visualize outgoing calls on a graph  X-axis: Date in a month(1~30/31) Y-axis: Time-of-day (0~24) Red circles: outgoing calls / Label: call destination CALL ANALYSIS USER SURVEY Can you see the difference in calling pattern?
Reasons for recommendation ALGORITHM CALL ANALYSIS USER SURVEY Phone calculates “the probabilities of Bernoulli trials” for each reasonfor each person in the current context 5 reasons for recommendation
Algorithm designAn example ALGORITHM CALL ANALYSIS USER SURVEY 2008/3/25 1:30 AM(Late night) Tuesday (weekday) pChicken,D= 015   pChicken,D′= 015−stdDay   pChicken,D′= 015− stdDay24hour=0   pChicken,W= 315   pChicken,W′= 315−stdWeekday   pChicken,W′= 315 − stdWeekday5×24hour=0.002   pChicken,T= 590   pChicken,T′= 590−stdTime   pChicken,T′=590 − stdTime3hour=0.017   pChicken,DP= 690   pChicken,DP′= 690−stdDaypart   pChicken,DP′=690 − stdDaypart6hour=0.01   pChicken,1H= 690×24   pChicken,1H′= 690×24−std1−h slot   pChicken,1H′= 690×24 − std1−h slot1hour=0.003   590   015   Day of week Time of day  1-hour slots of a day 690×24   Weekend / Weekday spans Dayparts of a day nc=15   kic3   690  
Speed-call lists for the same user / at different time ALGORITHM CALL ANALYSIS USER SURVEY
Algorithm verification using 3 month call logs of 20 users ALGORITHM CALL ANALYSIS USER SURVEY < < >
Callers were different Our speed call list was effective for some while not for others.  Different recommendation method for different calling type. ALGORITHM CALL ANALYSIS USER SURVEY
Our ESM study To collect in-situ user feedback and experience in the real situations With 10 users for 3 months No more comparison with existing methods ESM FIELD STUDY CALL ANALYSIS USER SURVEY ALGORITHM ESM system Real-time addition/modification of questions Easy self report with a screenshot  Application logging for evaluating recommendation performance
Recommendation performance Using adaptive speed-call list (19%) Mean Rank: 2.3 / Median rank: 2 ESM FIELD STUDY CALL ANALYSIS USER SURVEY ALGORITHM The rank lower than 5th Include calleeswithout recent call & Newly dialed numbers
In-situ feedback right after call Was it helpful?  Can you tell me more? I tried to call my boyfriend, and he was on the top of the list. The person I wanted to call was the 1st. It always shows a similar list, but is quite helpful. I could see "home" when I was about to call home. I called someone many times recently, and the list helped me.  The time to call was reduced due to the list. I called him after a long pause, and the list was not useful. ( - ) I looked up the number to call from phonebook. ( - ) … ESM FIELD STUDY CALL ANALYSIS USER SURVEY ALGORITHM
Questionnaire summary ESM FIELD STUDY CALL ANALYSIS USER SURVEY ALGORITHM 80% reported the speed-call list was helpful Because of adequate candidates on the list No need to type for search Complained about … Extra button clicks to invoke a speed-call list Time delay due to calculating a speed-call list New requirements Fast switching to a search page or recent call list when recommendation fails  Fast calculation    Accuracy
Conclusion We … Studied mobile phone users’ calling patterns Designed a call recommendation algorithm for an adaptive speed-call list Evaluated it with an ESM system, as well as call logs   Adaptive speed call list  Effective in terms of  Hit ratio on Top-rank  Hit ratio on first page The number of button presses  Helpful in the real situation Positive in-situ responses Thank you  for your attention. sh.lee@kaist.ac.kr Because I’m curious

Weitere ähnliche Inhalte

Andere mochten auch

Weee trace weee-forum 20111125
Weee trace weee-forum 20111125Weee trace weee-forum 20111125
Weee trace weee-forum 20111125Enrique Redondo
 
Section 504 shorter version 2008
Section 504 shorter version 2008Section 504 shorter version 2008
Section 504 shorter version 2008Susan E. Myers
 
SES Spring 2015 Legal Update
SES Spring 2015 Legal UpdateSES Spring 2015 Legal Update
SES Spring 2015 Legal UpdateKelli Moors
 
Introduction to section 504 09 08
Introduction to section 504 09 08Introduction to section 504 09 08
Introduction to section 504 09 08Susan E. Myers
 
Spring 2014 Special Education in the Modern Age: Parent Participation in the...
Spring 2014  Special Education in the Modern Age: Parent Participation in the...Spring 2014  Special Education in the Modern Age: Parent Participation in the...
Spring 2014 Special Education in the Modern Age: Parent Participation in the...Fagen Friedman & Fulfrost
 
SES Fall 2012 The Shifting Categories of ED, OHI and SLD
SES Fall 2012 The Shifting Categories of ED, OHI and SLDSES Fall 2012 The Shifting Categories of ED, OHI and SLD
SES Fall 2012 The Shifting Categories of ED, OHI and SLDFagen Friedman & Fulfrost
 
SES spring 2015: The ABCs of ESY (Extended School Year)
SES spring 2015: The ABCs of ESY (Extended School Year)SES spring 2015: The ABCs of ESY (Extended School Year)
SES spring 2015: The ABCs of ESY (Extended School Year)Fagen Friedman & Fulfrost
 
TheJaroMeterPromisesPersonalGoals
TheJaroMeterPromisesPersonalGoalsTheJaroMeterPromisesPersonalGoals
TheJaroMeterPromisesPersonalGoalsJaron Brandon
 
413 ch1 powerpoint
413 ch1 powerpoint413 ch1 powerpoint
413 ch1 powerpointJenYork
 
Bullying and the Special Education Student
Bullying and the Special Education StudentBullying and the Special Education Student
Bullying and the Special Education StudentLynn Prado
 
SES FALL 2014: Child Find Under the Microscope
SES FALL 2014: Child Find Under the MicroscopeSES FALL 2014: Child Find Under the Microscope
SES FALL 2014: Child Find Under the MicroscopeFagen Friedman & Fulfrost
 

Andere mochten auch (20)

Weee trace weee-forum 20111125
Weee trace weee-forum 20111125Weee trace weee-forum 20111125
Weee trace weee-forum 20111125
 
Section 504 shorter version 2008
Section 504 shorter version 2008Section 504 shorter version 2008
Section 504 shorter version 2008
 
376387 1
376387 1376387 1
376387 1
 
SES Spring 2015 Legal Update
SES Spring 2015 Legal UpdateSES Spring 2015 Legal Update
SES Spring 2015 Legal Update
 
Marcy gutierrez 2
Marcy gutierrez 2Marcy gutierrez 2
Marcy gutierrez 2
 
A hoonanian
A hoonanianA hoonanian
A hoonanian
 
Sloan simmons
Sloan simmonsSloan simmons
Sloan simmons
 
Introduction to section 504 09 08
Introduction to section 504 09 08Introduction to section 504 09 08
Introduction to section 504 09 08
 
AB 1266 Webinar
AB 1266 WebinarAB 1266 Webinar
AB 1266 Webinar
 
Spring 2014 Special Education in the Modern Age: Parent Participation in the...
Spring 2014  Special Education in the Modern Age: Parent Participation in the...Spring 2014  Special Education in the Modern Age: Parent Participation in the...
Spring 2014 Special Education in the Modern Age: Parent Participation in the...
 
The Implications of AB 1266
The Implications of AB 1266The Implications of AB 1266
The Implications of AB 1266
 
SES Fall 2012 The Shifting Categories of ED, OHI and SLD
SES Fall 2012 The Shifting Categories of ED, OHI and SLDSES Fall 2012 The Shifting Categories of ED, OHI and SLD
SES Fall 2012 The Shifting Categories of ED, OHI and SLD
 
SES Spring 2013 - All Things Considered
SES Spring 2013 - All Things ConsideredSES Spring 2013 - All Things Considered
SES Spring 2013 - All Things Considered
 
SES spring 2015: The ABCs of ESY (Extended School Year)
SES spring 2015: The ABCs of ESY (Extended School Year)SES spring 2015: The ABCs of ESY (Extended School Year)
SES spring 2015: The ABCs of ESY (Extended School Year)
 
Marcy gutierrez
Marcy gutierrezMarcy gutierrez
Marcy gutierrez
 
SES Fall 2015: Legal Update
SES Fall 2015: Legal UpdateSES Fall 2015: Legal Update
SES Fall 2015: Legal Update
 
TheJaroMeterPromisesPersonalGoals
TheJaroMeterPromisesPersonalGoalsTheJaroMeterPromisesPersonalGoals
TheJaroMeterPromisesPersonalGoals
 
413 ch1 powerpoint
413 ch1 powerpoint413 ch1 powerpoint
413 ch1 powerpoint
 
Bullying and the Special Education Student
Bullying and the Special Education StudentBullying and the Special Education Student
Bullying and the Special Education Student
 
SES FALL 2014: Child Find Under the Microscope
SES FALL 2014: Child Find Under the MicroscopeSES FALL 2014: Child Find Under the Microscope
SES FALL 2014: Child Find Under the Microscope
 

Ähnlich wie An Adaptive Speed-Call List Algorithm and Its Evaluation with ESM

Quality Journey --Sampling Process.pdf
Quality Journey --Sampling Process.pdfQuality Journey --Sampling Process.pdf
Quality Journey --Sampling Process.pdfNileshJajoo2
 
Practical Statistical Testing
Practical Statistical TestingPractical Statistical Testing
Practical Statistical TestingAdrian Cuyugan
 
Examples of working with streaming data
Examples of working with streaming dataExamples of working with streaming data
Examples of working with streaming dataYi-Shin Chen
 
Rinse and Repeat : The Spiral of Applied Machine Learning
Rinse and Repeat : The Spiral of Applied Machine LearningRinse and Repeat : The Spiral of Applied Machine Learning
Rinse and Repeat : The Spiral of Applied Machine LearningAnna Chaney
 
Did Something Change? - Using Statistical Techniques to Interpret Service and...
Did Something Change? - Using Statistical Techniques to Interpret Service and...Did Something Change? - Using Statistical Techniques to Interpret Service and...
Did Something Change? - Using Statistical Techniques to Interpret Service and...Joao Galdino Mello de Souza
 
Nova Southeastern University H. Wayne Huizenga School of Bus.docx
Nova Southeastern University H. Wayne Huizenga School of Bus.docxNova Southeastern University H. Wayne Huizenga School of Bus.docx
Nova Southeastern University H. Wayne Huizenga School of Bus.docxhenrymartin15260
 
The t Test for Related.docx
The t Test for Related.docxThe t Test for Related.docx
The t Test for Related.docxchristalgrieg
 
VOICE PASSWORD BASED SPEAKER VERIFICATION SYSTEM USING VOWEL AND NON VOWEL RE...
VOICE PASSWORD BASED SPEAKER VERIFICATION SYSTEM USING VOWEL AND NON VOWEL RE...VOICE PASSWORD BASED SPEAKER VERIFICATION SYSTEM USING VOWEL AND NON VOWEL RE...
VOICE PASSWORD BASED SPEAKER VERIFICATION SYSTEM USING VOWEL AND NON VOWEL RE...niranjan kumar
 
sampling and statiscal inference
sampling and statiscal inferencesampling and statiscal inference
sampling and statiscal inferenceShruti MISHRA
 
Monte Carlo Simulation
Monte Carlo SimulationMonte Carlo Simulation
Monte Carlo SimulationDeepti Singh
 
Rv defcon25 burner phone challenge - dakota nelson
Rv defcon25   burner phone challenge - dakota nelsonRv defcon25   burner phone challenge - dakota nelson
Rv defcon25 burner phone challenge - dakota nelsonreconvillage
 
Recent and Robust Query Auto-Completion - WWW 2014 Conference Presentation
Recent and Robust Query Auto-Completion - WWW 2014 Conference PresentationRecent and Robust Query Auto-Completion - WWW 2014 Conference Presentation
Recent and Robust Query Auto-Completion - WWW 2014 Conference Presentationstewhir
 
Gender voice recognition.pptx
Gender voice recognition.pptxGender voice recognition.pptx
Gender voice recognition.pptxRohith572864
 

Ähnlich wie An Adaptive Speed-Call List Algorithm and Its Evaluation with ESM (20)

Quality Journey --Sampling Process.pdf
Quality Journey --Sampling Process.pdfQuality Journey --Sampling Process.pdf
Quality Journey --Sampling Process.pdf
 
Practical Statistical Testing
Practical Statistical TestingPractical Statistical Testing
Practical Statistical Testing
 
Examples of working with streaming data
Examples of working with streaming dataExamples of working with streaming data
Examples of working with streaming data
 
Quality of data
Quality of dataQuality of data
Quality of data
 
Rinse and Repeat : The Spiral of Applied Machine Learning
Rinse and Repeat : The Spiral of Applied Machine LearningRinse and Repeat : The Spiral of Applied Machine Learning
Rinse and Repeat : The Spiral of Applied Machine Learning
 
Work sampling
Work samplingWork sampling
Work sampling
 
Did Something Change? - Using Statistical Techniques to Interpret Service and...
Did Something Change? - Using Statistical Techniques to Interpret Service and...Did Something Change? - Using Statistical Techniques to Interpret Service and...
Did Something Change? - Using Statistical Techniques to Interpret Service and...
 
Nova Southeastern University H. Wayne Huizenga School of Bus.docx
Nova Southeastern University H. Wayne Huizenga School of Bus.docxNova Southeastern University H. Wayne Huizenga School of Bus.docx
Nova Southeastern University H. Wayne Huizenga School of Bus.docx
 
The t Test for Related.docx
The t Test for Related.docxThe t Test for Related.docx
The t Test for Related.docx
 
Sampling brm chap-4
Sampling brm chap-4Sampling brm chap-4
Sampling brm chap-4
 
Chap014
Chap014Chap014
Chap014
 
VOICE PASSWORD BASED SPEAKER VERIFICATION SYSTEM USING VOWEL AND NON VOWEL RE...
VOICE PASSWORD BASED SPEAKER VERIFICATION SYSTEM USING VOWEL AND NON VOWEL RE...VOICE PASSWORD BASED SPEAKER VERIFICATION SYSTEM USING VOWEL AND NON VOWEL RE...
VOICE PASSWORD BASED SPEAKER VERIFICATION SYSTEM USING VOWEL AND NON VOWEL RE...
 
Sampling methods
Sampling methodsSampling methods
Sampling methods
 
135-Ch5.ppt
135-Ch5.ppt135-Ch5.ppt
135-Ch5.ppt
 
Lecture1.pdf
Lecture1.pdfLecture1.pdf
Lecture1.pdf
 
sampling and statiscal inference
sampling and statiscal inferencesampling and statiscal inference
sampling and statiscal inference
 
Monte Carlo Simulation
Monte Carlo SimulationMonte Carlo Simulation
Monte Carlo Simulation
 
Rv defcon25 burner phone challenge - dakota nelson
Rv defcon25   burner phone challenge - dakota nelsonRv defcon25   burner phone challenge - dakota nelson
Rv defcon25 burner phone challenge - dakota nelson
 
Recent and Robust Query Auto-Completion - WWW 2014 Conference Presentation
Recent and Robust Query Auto-Completion - WWW 2014 Conference PresentationRecent and Robust Query Auto-Completion - WWW 2014 Conference Presentation
Recent and Robust Query Auto-Completion - WWW 2014 Conference Presentation
 
Gender voice recognition.pptx
Gender voice recognition.pptxGender voice recognition.pptx
Gender voice recognition.pptx
 

Kürzlich hochgeladen

"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii SoldatenkoFwdays
 
How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.Curtis Poe
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Mattias Andersson
 
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxUse of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxLoriGlavin3
 
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024BookNet Canada
 
The State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptxThe State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptxLoriGlavin3
 
Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 3652toLead Limited
 
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxLoriGlavin3
 
Gen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdfGen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdfAddepto
 
SAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptxSAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptxNavinnSomaal
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024Lonnie McRorey
 
A Deep Dive on Passkeys: FIDO Paris Seminar.pptx
A Deep Dive on Passkeys: FIDO Paris Seminar.pptxA Deep Dive on Passkeys: FIDO Paris Seminar.pptx
A Deep Dive on Passkeys: FIDO Paris Seminar.pptxLoriGlavin3
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek SchlawackFwdays
 
Moving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfMoving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfLoriGlavin3
 
DevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platformsDevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platformsSergiu Bodiu
 
WordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your BrandWordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your Brandgvaughan
 
Nell’iperspazio con Rocket: il Framework Web di Rust!
Nell’iperspazio con Rocket: il Framework Web di Rust!Nell’iperspazio con Rocket: il Framework Web di Rust!
Nell’iperspazio con Rocket: il Framework Web di Rust!Commit University
 
Connect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationConnect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationSlibray Presentation
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxLoriGlavin3
 

Kürzlich hochgeladen (20)

"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko
 
How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?
 
DMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special EditionDMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special Edition
 
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxUse of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
 
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
 
The State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptxThe State of Passkeys with FIDO Alliance.pptx
The State of Passkeys with FIDO Alliance.pptx
 
Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365
 
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
 
Gen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdfGen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdf
 
SAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptxSAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptx
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024
 
A Deep Dive on Passkeys: FIDO Paris Seminar.pptx
A Deep Dive on Passkeys: FIDO Paris Seminar.pptxA Deep Dive on Passkeys: FIDO Paris Seminar.pptx
A Deep Dive on Passkeys: FIDO Paris Seminar.pptx
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
 
Moving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfMoving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdf
 
DevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platformsDevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platforms
 
WordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your BrandWordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your Brand
 
Nell’iperspazio con Rocket: il Framework Web di Rust!
Nell’iperspazio con Rocket: il Framework Web di Rust!Nell’iperspazio con Rocket: il Framework Web di Rust!
Nell’iperspazio con Rocket: il Framework Web di Rust!
 
Connect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationConnect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck Presentation
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
 

An Adaptive Speed-Call List Algorithm and Its Evaluation with ESM

  • 1. An Adaptive Speed-Call List Algorithm and Its Evaluation with ESM Seunghwan Lee, JungsukSeo, Geehyuk Lee HCI Lab @ Darling Mother Professor Kim This time Frequent Thursday Evening
  • 2. Motivation Can a mobile phonepredictand recommenda number to call in a certain context?
  • 3. Overview We … Designed a method to generate an adaptive speed-call list Online survey with 75 participants Using 3 month call log of 20 users Verified the effectiveness Using 3 month call log of 20 users Implemented our method and conducted ESM study With 10 users for 3 months Darling Mother Professor Kim This time Frequent Thursday Evening
  • 4. How do you make a phone call? Online survey with 75 participants (16-63 years old, 5+ years experience) USER SURVEY
  • 5. Any regular call pattern? 67% have a number to call regularly Parents (51%) Lover / spouse (44%) Friends (21%), Brothers & sisters (15%) USER SURVEY my husbandto get home every night. my husbandto get home every night. my colleagues to remind meeting. my lazy boyfriend for morning call. my lazy boyfriend for morning call. my friendsfor having lunch together. my parentat weekend. my parentat weekend. my friends after my work hours Chinese restaurant for food delivery every weekend my wife every lunch time. my wife every lunch time. my girlfriend while leaving my office. my girlfriend while leaving my office.
  • 6. Call pattern analysis Collected 3 month call log of 20 users Call map: Visualize outgoing calls on a graph X-axis: Date in a month(1~30/31) Y-axis: Time-of-day (0~24) Red circles: outgoing calls / Label: call destination CALL ANALYSIS USER SURVEY Can you see the difference in calling pattern?
  • 7. Reasons for recommendation ALGORITHM CALL ANALYSIS USER SURVEY Phone calculates “the probabilities of Bernoulli trials” for each reasonfor each person in the current context 5 reasons for recommendation
  • 8. Algorithm designAn example ALGORITHM CALL ANALYSIS USER SURVEY 2008/3/25 1:30 AM(Late night) Tuesday (weekday) pChicken,D= 015   pChicken,D′= 015−stdDay   pChicken,D′= 015− stdDay24hour=0   pChicken,W= 315   pChicken,W′= 315−stdWeekday   pChicken,W′= 315 − stdWeekday5×24hour=0.002   pChicken,T= 590   pChicken,T′= 590−stdTime   pChicken,T′=590 − stdTime3hour=0.017   pChicken,DP= 690   pChicken,DP′= 690−stdDaypart   pChicken,DP′=690 − stdDaypart6hour=0.01   pChicken,1H= 690×24   pChicken,1H′= 690×24−std1−h slot   pChicken,1H′= 690×24 − std1−h slot1hour=0.003   590   015   Day of week Time of day 1-hour slots of a day 690×24   Weekend / Weekday spans Dayparts of a day nc=15   kic3   690  
  • 9. Speed-call lists for the same user / at different time ALGORITHM CALL ANALYSIS USER SURVEY
  • 10. Algorithm verification using 3 month call logs of 20 users ALGORITHM CALL ANALYSIS USER SURVEY < < >
  • 11. Callers were different Our speed call list was effective for some while not for others. Different recommendation method for different calling type. ALGORITHM CALL ANALYSIS USER SURVEY
  • 12. Our ESM study To collect in-situ user feedback and experience in the real situations With 10 users for 3 months No more comparison with existing methods ESM FIELD STUDY CALL ANALYSIS USER SURVEY ALGORITHM ESM system Real-time addition/modification of questions Easy self report with a screenshot Application logging for evaluating recommendation performance
  • 13. Recommendation performance Using adaptive speed-call list (19%) Mean Rank: 2.3 / Median rank: 2 ESM FIELD STUDY CALL ANALYSIS USER SURVEY ALGORITHM The rank lower than 5th Include calleeswithout recent call & Newly dialed numbers
  • 14. In-situ feedback right after call Was it helpful? Can you tell me more? I tried to call my boyfriend, and he was on the top of the list. The person I wanted to call was the 1st. It always shows a similar list, but is quite helpful. I could see "home" when I was about to call home. I called someone many times recently, and the list helped me. The time to call was reduced due to the list. I called him after a long pause, and the list was not useful. ( - ) I looked up the number to call from phonebook. ( - ) … ESM FIELD STUDY CALL ANALYSIS USER SURVEY ALGORITHM
  • 15. Questionnaire summary ESM FIELD STUDY CALL ANALYSIS USER SURVEY ALGORITHM 80% reported the speed-call list was helpful Because of adequate candidates on the list No need to type for search Complained about … Extra button clicks to invoke a speed-call list Time delay due to calculating a speed-call list New requirements Fast switching to a search page or recent call list when recommendation fails Fast calculation Accuracy
  • 16. Conclusion We … Studied mobile phone users’ calling patterns Designed a call recommendation algorithm for an adaptive speed-call list Evaluated it with an ESM system, as well as call logs Adaptive speed call list Effective in terms of Hit ratio on Top-rank Hit ratio on first page The number of button presses Helpful in the real situation Positive in-situ responses Thank you for your attention. sh.lee@kaist.ac.kr Because I’m curious

Hinweis der Redaktion

  1. Nice to meet you. I’m Seunghwan Lee from KAIST and I’m honored to present our work “ ” in CHI Let’s start with very short demo. ###OK. That’s a speed-call list, a kind of call recommendation.
  2. We were curious about this given question.
  3. We were curious about this given question.To answer this question, We …Let me show you a very very short demo again, If a user press the send key, #####then the adaptive speed-call list is given with adequate recommendation reasons.
  4. To find out which calling method is used to call and whether there is any regular call pattern or not,We conducted online survey with 75 users.As you see in the table, Searching and selecting from recent call comprised more than sixty percent of responses. Therefore we adopted pressing the send key for the start event to see our speed call list.
  5. To analyze call pattern, We collected … visualized With a vertical time-of-day axis (0 to 24 hour ) and a horizontaldate axis. (1 to 30 or 31)Red circles and labels indicate call destinationYes. These call maps are from different users. Left from a graduate student.he loves to eat a late-night snack and sleep late. Right shows calls from an married man. He has regular life pattern and call pattern. Majority of calls are concentrated on his wife.
  6. To give a speed-call list, we selected .. The possible values for each variables are shown in the right column. For example, the phone can recommend Mom “Because it is Sunday” by the day of week,Or recommend a colleague Because it is afternoon by daypart of day .
  7. Let me explain the algorithm with this example call map. A user press the send key to make a call ## at this red point, one thirty AM (1:30) on Tuesday.Consider the probability for chicken delivery ### for weekend or weekday.The phone counts number of weekdays for the last 3 months. And also count number of weekdays with a phone call to chicken delivery.In the same manner, the phone calculates probabilities for different variables, too.Subtract std from these values for pessimistic estimation. Normalize these values by dividing by the individual time span. Then, take the highest P for chicken delivery, Sort all the Ps for each person. Finally select 5 of them. /// Q: because of different nc, all the Ps have Different confidence level. So And we Normalized them into probability in unit time because of Different time scales of 5 independent variablesWe can show you an example with this call map. If a user press the send button on the mobile phone at this red point, AM 1:30, the late night in TuesdayThe phone know that it’s late night of Tuesday, one of the weekdays then count number of Tuesdayduring last 3 months. And also count number of Tuesday with a phone call to each person. In the same manner, the phone calculate probabilities for each variablesBecause of Different confidence level from different 𝑛_𝑐, we adjusted p with pessimistic inference
  8. The result will be like this. Delivery for chicken is on the top because of time /and Call for family is recommended based on the call frequency. For the same user, The same algorithm can give a different result like this ### for different time, Monday afternoon.
  9. Then We compared the performance of the algorithm with other methods. …It would be the best if the number is on the top of the list. Its probability Is highest in our case. Otherwise it would be still good if it appears on the first page. Its probability Is highest in our case, too. The number of button presses for calling was smallest.The average result showed our adaptive-speed call list is best among these methods.
  10. Thegraphs shows individual differences and We could classify 20 users into 3 groups based on recommendation suitability. The red line group are most suitable for call prediction. We can classify the user group with some features of call logs like call concentration rate. ###Therefore if we determine the user group first, then selecting more effective variables would be possible and it is expected to bring better recommendation result.
  11. We implemented our method on samsung blackjack and conducted a ESM study to … ###The ESM system we used support questionnaire, self report, application logging###We replaced usualcall log page into our speed call list like this. If the recommendation is failed, then user can change the list to usual recent call log page or can access phone book easily.
  12. 19 % of all outgoing calls were made using … seventy percent of cases were success and others failed. In this case, the mean and median rank was about 2. The failed cases include recommendation impossible cases like callee without recent call, and the newly dialed numbers. /// Q: we used blackjack, a smartphone, so some people could use photo shortcut or other plugins installed by themselves, And another reason for low rate is that after they press the send key, they had to click ‘call log’ soft button on the screen again to see the list So some of them didn’t use the list in some cases. The page replacement was not perfect. I feel sad about it too. Actually, we could calculate all the callees with call log again, and we got 60% of success for those missed cases.
  13. The ESM system presented a window right after call to ask user feedback. About sixty percent answered it was helpful at that moment. And some positive or negative descriptions were also gathered.
  14. On the questionnaire after field study,### ###they had to press the send key and click ‘call log’ soft button on the screen again to see the list ###We requested themto give us some requirement for speed-call list, and fast switching to existing method was elected as the most important requirement, than other 2 items.Contrary to our expectation, accuracy was ranked at 3rdWe thought that people considered it as a supplemental tool, and an occasional failure did not seem to be very critical.
  15. I skipped some details due to short time,If you are curious, please contact me. /// Q: on average, 25 users were called by participants for 3 months