Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out

SocioXensor Measuring User Experience and User Behaviour in

VIEWS: 3 PAGES: 32

									      SocioXensor
          Measuring
     user behaviour and
      user eXperience
          in ConteXt
     with mobile devices


Ingrid Mulder, Henri ter Hofte,
      Telematica Instituut
          Joke Kort
           TNO ICT
Outline
• Context of SocioXensor:
  Freeband User eXperience (FRUX) project
• Why SocioXensor?
   • Example: ESM study into interruptability
   • Design and evaluation of “we-centric” services
• SocioXensor, a tool for measuring:
   • user eXperience
   • application usage
   • user behaviour and (social) conteXt
• SocioXensor architecture and some sensors
• Q&A
Focus in the FRUX project:
Dynamic personal social context
We-centric services/applications
“.. sense, mediate, visualize and in some cases interpret social context
     information as part of their service provided to end-users”

          Ad                                     Theo




             Mediate/Aggregate context

                    We-centric service
                           Interpret context
We-centric R&D issues
• Requirements oriented research
   • In which social contexts do certain
     information/communication needs arise?
   • How often? Which patterns?
• Design oriented research
   •   Which (combination of) social context information
           Gives best predictive power
           At lowest cost
           Is relevant to mediate, aggregate and/or interpret
• Evaluation oriented research
   •   Did the user experience improve after introduction of
       a we-centric service?
           In which context(s) is it used? Which features?
           How often? Which patterns?
Example: ESM study
into availabilty for interruptions
• How does
  a user experience
   •   interruptability
• Depend on
  context
   • conversation
   • location
   • proximity
     (nearness)
• Study setup:
   • 9 persons
   • 1 week,
   • 8-22h
ESM study; some results:
How are you in conversation?

              via mobile                   via Instant
                           other; 0,5%
            phone; 0,9%                  Messaging; 0,4%
via fixed phone;
       1,1%


 face to face;
    32,3%




                                             not; 65%
ESM study; some results:
correlation with interruptability

          Question and answer                   r
          Are you in conversation?
             yes                           -0,369
          How are you in conversation?
            face to face                   -0,343
          With how many people are you?
             (incl. yourself)              -0,293
          Where at the <company> office?
            meeting room                   -0,222
          Where at the <company> office?
            My own office room             0,193
          Where at the <company> office?
            office room of a colleague     -0,172
          How are you in transit?
            with public transport          0,123
          Where are you now?
            in transit                     -0,115
Living Lab approach
“Bring the lab to the people”
                                          experience
                                           sampling
                        logging                            ethnography
high
                            SocioXensor
   Situatedness




                                   diary



                        survey        interview
                                                       lab experiment
low

                                    Obtrusiveness
                  low                                           high
SocioXensor Research targets:
    Method (book)
                                Instrument (tool)




      Measuring
 User Experience and
   User Behaviour
      in Context
                                  SocioXensor


                    Used and tested!
SocioXensor:
an in-situ measurement instrument

Phenomenon     User           Application   Social
               Experience     Usage         Context

Type of data   self-report,   log,          log,
               subjective     objective     objective

Examples       needs,         application   proximity,
               emotions,      usage         communication,
               frustrations                 relation

Measurement Experience        Usage         Context
Software    sampler           sensor        sensor
SocioXensor high-level architecture
                                                 User domain                Researcher domain




 (Mobile) Client


          Application    SocioXensor

                                        Cradle
                                                                              Researcher
                         GPRS/UMTS
 Operator Network

           Application                             internet
                          SocioXensor
            (network)
                                                           Context Data
                                                            Usage Data
                                                          Experience Data
 (Application) Server                                                        SocioXensor
                                                                              repository
           Application
                          SocioXensor
            (server)
SocioXensor client architecture

                                                                                      ?
                                                     Blue
                                                     Tooth
                                                              Audio       GPS         ...
(Mobile) Client
                                       Calendar



                          Experience   Context     Context   Context     Context Context
                           Sampler     Sensor      Sensor    Sensor      Sensor Sensor




                                                                                            Cradle
                  API   Usage                     SocioXensor
  Application
                        Sensor                    data manager       Context Data
                                                                      Usage Data
                                                                    Experience Data


                        SocioXensor              Local Repository
                                                                                            GPRS/
                                                                                            UMTS
Proximity context sensors:
Who is near?
•   Infrared (e.g. ‘Sociometer’)
•   Bluetooth (e.g. ‘BlueAware’)
•   WLAN
•   Audio
•   …
•   and correlation of e.g.,
    • GPS,
    • GSM cell-ID,
    • WLAN access point ID,
    • usage of computers at fixed locations.
Communication context sensors:
Who communicates with whom, how long, which media?
•   audio (for f2f conversations, e.g. ‘sociometer’),
•   mobile telephony
•   fixed telephony
•   Instant Messaging
•   E-mail
•   …
Relation context sensors:
Is A on the contact list of B? Do A and B have shared contacts?

•   E-mail contact lists
•   Instant Messaging contact lists
•   Speed dial entries in (mobile) telephones
•   …
Usage sensors:
Which application functions are used, how often?
•   Keyboard, mouse events
•   Screenshots
•   Application events
•   …
Experience samplers
•   Notify user
     • Pager (~semafoon)
     • Reminder on PDA/Smartphone
     • Phone
     • SMS
     • Instant Messaging
     • …
•   Collect answers
     • Paper form
     • PDA/Smartphone application
     • Voice (-response system / voicemail diary)
     • website
     • i-mode site (wapsite)
     • Instant Messaging (survey “bot”)
     • SMS
     • …
Bring the lab to the people with SocioXensor!
• Questions?




• More information
   • SocioXensor: http://socioxensor.freeband.nl
   • FRUX: http://frux.freeband.nl
   • Henri.terHofte at telin.nl
     Ingrid.Mulder at telin.nl
     Joke.Kort at tno.nl
Freeband
Facts Freeband

       Freeband Knowlegde Impuls
       •   April 2002 - April 2007
       •   24 Million Euro
       •   20 partners, 15 projects
       •   Management: Telematica Instituut
       •   More info www.freeband.n/kennisimpuls


       Freeband Communication
       •   April 2004 - December 2008
       •   61 Million Euro
       •   26 partners, 9 projects, 300 researchers
       •   Management: Telematica Instituut
       •   More info www.freeband.nl
Freeband Communication: projects
    Example Research: use and utility of
    real-time context information
    for mobile availability/interruptions
                                            Pilot:
                                            2 organisations
               Demonstrator     Validator   25 desktop users
                prototype      proptotype   10 mobile user
                   (lab)          (pilot)   3 months

                  Mockup
                 prototype     (mobile)
                 (papier)
                               office-
Interviews                     workers
     Logs (e-mail, IM)
                                                 Literature
        Surveys (N=2, N=82)

                              Workshop
Example prototype:
Live Contacts
•   From:
     • mobile phone
     • desktop PC (work/home)
•   In 4 clicks contact
     • Who is available?
     • When?
     • How?
•   Real-time context information
     • calendar
     • messenger status
     • location (home/work/mobile)
•   Availability profiles
     • red/orange/green
     • work phone / home phone
        / mobile phone / SMS / IM /
        e-mail / f2f
We-centric service:
design issues
• In which social contexts do certain
  information/communication needs arise?
• How often?
We-centric service:
design issues
• Which (combination of) social context information of
  others is most relevant to..
   • Mediate
     (don’t interpret, just transfer and visualize)
   • Aggregate
     (simple multi-factor /multi-person visualisation)
   • Interpret
     (e.g. filter incoming calls)
         Best predictive power
          (better than human observer?)
         At lowest cost (e.g. cost of building/using sensors
          and/or training time for learning algorithms)
We-centric service:
Evaluation issues
• Did the user experience improve
  after introduction of a we-centric service?
• In which context(s) is the we-centric service used?
  Which features?
• How often/how long is the we-centric service used?
  Which features?
We-centric service:
Evaluation issues
• Construct/develop models/theories/algorithms that
  explain/predict an experience,
  e.g. interruptability
   • (linear) regression models
          How strong do various (context) factors influence a
           certain experience?
          Are these effects statistically significant?
   • (learning) sensor-based algorithms that predict a
     certain experience
• Test hypotheses regarding
  value of we-centric services
  (e.g. user experience, utility and usability)
Sociometer, a tool for logging
Physical Proximity and Conversations
Sociometer (Choudhury et al). : a tool for logging
physical proximity and conversations
BlueAware (Eagle et al.):
logging/visualizing physical proximity

								
To top