UIUC; Dorothy Espelage, UIUC; Margaret Fleck, UIUC;

wonderfuldistinctΤεχνίτη Νοημοσύνη και Ρομποτική

16 Οκτ 2013 (πριν από 3 χρόνια και 9 μήνες)

84 εμφανίσεις

Scott Poole, UIUC;
Noshir

Contractor, Northwestern;
Mark Hasegawa
-
Johnson, UIUC;
Feniosky

Pena
-
Mora,
Columbia; David Forsyth, UIUC; Kenton McHenry,
UIUC; Dorothy
Espelage
, UIUC; Margaret Fleck, UIUC;
Alex
Yahja
, National Center for Supercomputing Apps

The Story behind Cultural Artifacts

Challenges


Socio
-
cultural consequences of group decisions


Inability to collect, analyze, and manage


High resolution,


High quality,


High volume
interaction

network data


Effective computer
-
aided collaboration among


Scholars


Scientists


Students


Volunteers


Stakeholders

Scientific Challenges


We
understand

small teams co
-
located (1
-
6 persons)
and we
think

we understand large aggregations of
1000s


We don’t understand large teams: 8
-
25, 25
-
70, 50
-
300,
350
-
500, 400
-
1000

the sweet spot of scholarly
collaborations and conferences


Current studies are surveys and case studies, not direct
observation, the gold standard


No tech to study these even though we coalesce in
natural groups of size 2, 5, 15,…


S
patial dispersion and movement make big difference

Importance of the Problem


Many critical groups are of this size:


Design Teams


Scholarly Collaborations


Cultural Studies


Legislative Bodies


Disaster Response Teams


Archaeology Teams


Medical Teams


Military Units


“Swarming” Disaster Response

Supported By


Cyber
-
enabled Discovery and Innovation (CDI)
program, National Science Foundation


Two Million Dollars Grant


National Center for Supercomputing Apps


Office of the Vice Chancellor for Research, University
of Illinois



Year 2 of Five Year Project


Project “
GroupScope


Approach


End
-
to
-
end system from data capture to analysis to user
and team engagement


Video cameras to capture video and audio, of


Study subjects such as children on playground


Scholars and researchers executing the study

in team and
individually


Synchronization of video and audio data


Annotation of video and audio


Coding of video and audio


Management of video and audio data


Analysis of video and audio; scenario simulation and
machine learning


Community involvement

Data
Acquisition
(cameras,
Kinect, audio
recorders, GPS,
iPhones,
iPads
)

First
-
order Data (audios,
photos, videos, sensor
data)

Data
Management
(Medici content
management,
ELAN
transcription)

Second
-
order Data (visual,
audio and text annotations,
coding and metadata)

Network
analysis, Group
identification,
Interaction
categorization

What
-
if
Scenario
Simulation and
Machine
Learning

Circle of Continuous Improvement

2D Face Tracking (
Kalal

TLD)

Depth
-
image “Kinect” Skeleton
Tracking

Human Movement Recognition

Social Interaction Recognition

Community Engagement


Professors and graduate students as primary research
participants


Students help annotate videos and audios of


s
tudy objects and artifacts


research activities of professors and research assistants


Interested folks help transcribe, translate, and annotate
videos and annotate


Multi
-
lingual collaboration enabled


Scenario “what
-
if” analyses of interactions and events


Annotated videos will “live” across time and place


Insights, inspirations, and moments are recorded and not lost
to time and place




In Closing



GroupScope
” tool is designed to provide


Computer
-
assisted collaboration among human teams


Natural and native human and professional social
-
networking

synergistic human machine effort


Scholarly collaboration tool with native domain
-
specific
design and interfaces


Natural collaboration space


By your consent, putting up video cameras to get PNC
2017 networking?


Will put up video cameras for NSF Radical Innovation
Summit 2013