6. So: How can we measure social networks and
face-to-face interactions?
Wearable Sensors, of course
7. ContextAware : Proximity and Location Logs
• Proximity and Location Log
– Sits in the background
• Details
– Starts on boot
– Logs location, other BT devices in a
user’s local area
• Applications
– Match Making Applications
– Social Network Mapping
– Epidemiology & Health Monitoring
14. Face-to-face interaction is primary medium of communication
High Complexity Information
Within a Floor
Within a Building
Within a Site
Between Sites
0 20 40 60 80
Proportion of Contacts
Face-to-Face
Telephone
Allen, T., Architecture and Communication Among Product Development Engineers. 1997, Sloan School of Magement, MIT: Cambridge.
15. Social Signals: cues in tone of voice, body language
• Intention, importance, reflected
in speaking style, body language
• People accurately and
automatically perceive this
information, and can use it to
accurately predict behavior
• Even though they are unaware
of it!
16. Social Signaling
Conversations are more than words
Darwin: active vs passive
engaged vs distant
stress vs relaxed
Mirroring vs independent
17. Audio Mining
How are they speaking? Who’s speaking when? Who’s talking to who?
Auditory Features Speaker Segmentation Finding Conversations
Who’s holding the floor?
Finding,Predicting, and Describing What sort of conversation is it?
Conversational Scenes Classifying Conversation Types
18. Measurement of Vocal Social Signals
• 1700 hours, 23 people, two weeks, workplace
• 96 people in salary negotiation (ave: 35min each)
• 84 people in speed dating (5 min each)
• 84 people in hiring scenario (5min each)
• 114 people in two conference situations (4 hours each)
• 6 people hospitalized for depression (3 wks each)
• Measured vocal signals for first 5 min (`thin slice’)
• Measured objective outcome, subjective outcome
• R>0.65, p<0.01 for quoted results
22. Measuring Mood
Sung, Marci, Pentland
Charting Mood
Sleep
More sleep Mania
Lack of energy
Low sociability
`flat’ conversation
Low sleep
More movement
High sociability
Talkative
Depression
23. Will you ask for more information? Gips, Pentland
27. Focus Groups
Pentland, Eagle
P a r ti c ip a n t S p e a k in g A v g (sec ) N ea re st T ra n s itio n A vg G ro u p
ti m e ( % ) C om m ent N e ig h b o r (N a m e , % ) In te re s t In te re s t
Iv an 1 .5 4 .1 N a th a n N a th a n -2 7 .2 1 .4 4
Jon 2 .2 2 .2 Sandy S an dy-47 .1 3 .3 6
Jo ost 9 .9 3 .5 Sandy Jordan-22 .2 0 .2 2
Jordan 1 1 .4 9 .6 M ik e S M ik e O -2 3 .0 5 .3 0
L e o n e l le 1 2 .8 8 .8 M ik e S S an dy-37 .1 8 .3 3
M ik e O . 1 6 .9 6 .6 Jordan M ik e S - 2 8 .0 9 .2 1
M ik e S . 1 0 .1 6 .6 Jordan S an dy-30 .1 9 .2 4
N a th a n 1 0 .8 1 0 .9 Iv a n S an dy-26 .4 0 .3 2
Sandy 2 4 .4 6 .9 M ik e S M ik e O -2 2 .1 7 .2 5
28. Focus Groups Madan, Pentland
Observed (bad) effects
• Domination
• Free riding
• Polarization
We can fix this!