Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
������
Interaction Beyond the Individual: ���A Lecture on HCI-Oriented Collaborative and Social Computing
Hao-‐Chuan Wang . 王浩全 Department of Computer Science Ins3tute of Informa3on Systems and Applica3ons Na3onal Tsing Hua University, Taiwan hAp://www.cs.nthu.edu.tw/~haochuan
@ Department of Computer Science and Informa3on Engineering, Na3onal Taiwan University. Oct 12 2012.
������
������
Wang
Background Hao-‐Chuan Wang 王浩全 �Assistant Professor, NTHU (Feb 2012 –) PhD & Postdoc, Cornell (3.5 years) PhD student, Carnegie Mellon
(2 years; transferred to Cornell) Other work and educaDonal experiences: Academia Sinica, NCCU, NTNU
2
������
������
Wang
NTHU Collaborative and Social Computing (CSC) Lab
3
������
������
Wang
Agenda
• What: Social compuDng from an HCI perspecDve • Why: Value of social compuDng • How: Design of social compuDng systems • How: Research in social compuDng. CHI & CSCW. • ReflecEon
4
������
������
Wang
Some References
Thomas Erickson’s Tutorial on InteracDon-‐Design.org hVp://www.interacDon-‐design.org/encyclopedia/
social_compuDng.html Panos IpeiroDs’ WWW 2011 Tutorial hVp://www.slideshare.net/ipeiroDs/managing-‐crowdsourced-‐
human-‐computaDon
5
������
������
Wang
What is Social Computing
6
������
������
Wang
HCI: Studying the Existing and Possible Relationships between Computers and People
7
hVp://old.sigchi.org/cdg/figure_1.gif
ACM SIGCHI Curricula 1996 (15 years ago)
������
������
Wang
Observation from Today
Nothing wrong, but slightly outdated. What’s changing today?
-‐ Much emphasis is on the context of use
-‐ Computers are more powerful and can look and work very differently
-‐ Not necessarily “one human, one computer”
-‐ Computer-‐mediated human-‐human interacDon becomes commonplace
8
������
������
Wang
Examples: MSN, QQ
9
������
������
Wang
Skype
10
������
������
Wang
Twitter, Plurk
11
������
������
Wang
Facebook, Google+
12
������
������
Wang
Amazon.com
13
������
������
Wang
Wikipedia
14
������
������
Wang
What’s common among these systems?���1. Technology-mediation���
2. ?
15
������
������
Wang
What’s common among these systems?���1. Social Interaction���
16
������
������
Wang
What’s common among these systems?���1. Social Interaction���
2. Technology Mediation
17
������
������
Wang
The Invisible Computers
QuesEon: Consider your recent experience of online communicaDon (email, IM, Skype, Facebook), rank the salience of the following targets:
(A) Computers (B) People you talk to (C) Tasks you do with people
18
������
������
Wang
The Invisible Computers
QuesEon: Consider your recent experience of online communicaDon (email, IM, Skype, Facebook), rank the salience of the following targets:
(A) Computers (B) People you talk to (C) Tasks you do with people
Most likely orderings: B, C, A or C, B, A. Computers play more of mediaDng roles, and can be
invisible to users. Social interacEon can maRer more.
19
������
������
Wang
Computing Systems with Significant “Social Layers”
“The social layer” as what disDnguishes these systems from other compuDng systems • Email, MSN, Skype are valuable because they support remote human communicaDon
• Facebook won’t be as rich and aVracDve if we did not have many friends using it
• Wikipedia becomes another content-‐less website if people are not moDvated or interested in making contribuDons.
An emerging category: Social CompuEng Not all technical, not all social, but “socio-‐technical”
20
������
������
Wang
Defining Social Computing
“Social compuDng refers to systems that support the gathering, processing and disseminaDon of informaDon that is distributed across social collecDves.
Furthermore, the informaDon in quesDon is not independent of people, but rather is significant precisely because it linked to people, who are in turn associated with other people.”
– Thomas Erickson, IBM Research
21
http://www.interaction-design.org/encyclopedia/social_computing.html
������
������
Wang
Why Social Computing?
22
������
������
Wang
Value of Social Computing
Enabling mechanism • Breaking exisDng constraints
Efficiency of processing • IntegraDon of collecDve efforts
Quality of outcomes • Social input, synergy
Human-‐machine collaboraDon • Leveraging unique human processing abiliDes • AugmenDng human processing with machines
Unique value emerges from using technologies to couple
people & enable interpersonal communicaEon.
23
http://www.interaction-design.org/encyclopedia/social_computing.html
������
������
Wang
Enabling Mechanism: Breaking the Constraints
24
Ex. Computer-‐mediated communicaDon tools enable remote communicaDon and distributed collaboraDon. Ex. Social networking sites (e.g., Facebook) make it possible to develop and maintain social connecDons at a different scale and intensity, and with different organizaDonal properDes (e.g., denser network).
������
������
Wang
Efficiency of Processing
25
CollecDve efforts can lead to efficient processing. Aner the 311 Earthquake, over 1500 edits on the Wikipedia arDcle in one day, producing a well-‐formed arDcle with rich text, photos and maps.
������
������
Wang
311 Earthquake Wikipedia Editing History
26
������
������
Wang
Quality of Outcomes Bounded raEonality: For problem solving and decision making,
people are with limited processing resources and cannot search the problem space thoroughly for more opDmal soluDons and decisions.
Ex. Technology-‐mediated social recommendaEon may help.
27
http://www.interaction-design.org/images/encyclopedia/social_computing/fig1_social_computing_research_social_media.jpg
������
������
Wang
Human-Machine Collaboration Human computaEon:
leveraging unique human processing capabiliDes, such as image and natural language understanding for content analysis and labeling.
Ex. DigiDzing old ediDons of
the New York Times with reCAPTCHA.
28
������
������
Wang
ESP Game
29
hVp://www.slideshare.net/ipeiroDs/managing-‐crowdsourced-‐human-‐computaDon
������
������
Wang 30 hVp://www.slideshare.net/ipeiroDs/managing-‐crowdsourced-‐human-‐computaDon
������
������
Wang 31
hVp://www.slideshare.net/ipeiroDs/managing-‐crowdsourced-‐human-‐computaDon
������
������
Wang
“Games With A Purpose” (GWAP)
Why are people doing the work (image labeling) for free? • Because it’s fun! • Image labeling as a by-‐product of gaming
People don’t necessarily want to do free work even when the task is simple. Need to moDvate or incenDvize people. • Good experience (gaming, GWAP) • Monetary incenDve (Amazon Mechanical Turk) • EducaDon (learning, Duolingo)
32
hRp://www.gwap.com/gwap/ Games with a Purpose
������
������
Wang
Duolingo
TranslaDng the whole web while people learn a second language.
33
Duolingo IntroducEon Video http://www.youtube.com/watch?v=WyzJ2Qq9Abs
http://duolingo.com/ Sign up Duolingo to learn a second language
������
������
Wang
Human-Machine Collaboration
AugmenEng Human Processing: People can be bad at doing some work, and machines can possibly help out.
Ex. IdeaExpander-‐ SupporDng idea generaDon by visualizing ongoing conversaDons as relevant pictures.
34
[Wang et al., CSCW 2010] hVp://www.cs.cornell.edu/~haochuan/manuscripts/WangCosleyFussell_CSCW_10.pdf
������
������
Wang
Enhanced machine translaDon with keyword highlighDng for cross-‐language chat (e.g., Mandarin-‐English communicaDon)
35
Human-Machine Collaboration
[Gao, Wang, Fussell, Cosley. under review]
������
������
Wang
How to Design Social Computing Systems?
36
������
������
Wang
Designing Social Computing Systems Ideally from an HCI design perspecDve:
Study -‐> Design -‐> Prototype -‐> Study -‐> Redesign …
37
hVp://pages.cpsc.ucalgary.ca/~saul/hci_topics/pdf_files/introducDon_481.pdf Saul Greenberg
Human Computer Interaction
A discipline concerned with the
of interactive computing systems for human use �
design implementation
evaluation
������
������
Wang
Designing Social Computing Systems (cont.)
RealisDcally, designers onen are not very clear what lead to successful social compuDng • Facebook changes all the Dme, but hard to say it’s always becoming “beVer”
• Usable interfaces do not necessarily imply useful social compuDng, and vice versa
• A strong “studier” culture: Studying how people collaborate offline and online
• Borrowing from mulEple disciplines: CommunicaDon, social psychology, sociology, STS, urban planning etc.
38
Saul Greenberg
Human Computer Interaction
A discipline concerned with the
of interactive computing systems for human use �
design implementation
evaluation
������
������
Wang
More about Social Computing Design
“Best pracDces and pixalls in social compuDng”: Interview with Thomas Erickson (IBM Research) on InteracDon-‐Design.org
39
Best pracEces (– 6’10’’): hRp://www.youtube.com/watch?v=gnsRuXaZCNA
������
������
Wang
Summary about Best Design Practices by Thomas Erickson In short: It’s not trivial.
• Learning from face-‐to-‐face interacDon and emulaDng aspects of it online may help
• Close, in-‐context observaDon may help • Don’t over-‐trust designers’ intuiDon • Be comfortable with contradicDons (acknowledge that it’s complex)
• Prototype the system and push it into the context as soon as possible
Conceptually, social compuDng design is sDll “user-‐centered”,
but onen there is no good method or heurisDc, and the outcome can be more unpredictable than regular user interface design.
40
������
������
Wang
Learning From Everyday Social Interaction
Mobilizing the crowd with socially shared knowledge, feeling and value.
41
http://www.youtube.com/watch?v=je1KOcBYGjM
������
������
Wang
How: Research in Social Computing
42
������
������
Wang
Invention-Driven and Understanding-Driven Research
CompuDng academics are with a strong tradiDon of invenDon • Invent an arDfact (e.g., algorithm) and study its properDes thoroughly. InvenDon takes a lead.
Good but don’t always work great
• Academics didn’t invent Facebook. Zuckerberg and colleagues invented more of the tool, but less of the social structure and social interacDon out there.
• Not all clear how to iniDate and sustain social networking sites, online communiDes etc. yet.
43
������
������
Wang
Invention-Driven and Understanding-Driven Research (cont.)
Understanding-‐driven strategy • PragmaEsm: Doesn’t maVer who invented it. Accept that it’s there and many users like or use it.
• What’s important is not to reinvent it, but to gain deeper understanding of the phenomena.
• Richer understanding may contribute to improvement and new invenDon later.
Studying offline and online social interacDons in different domains and situaDons is relevant and valuable.
44
������
������
Wang
Some Elements in Social Computing Research
Computer-‐mediated communicaDon Computer-‐supported cooperaDve work Social media Social networking Online community Human computaDon Crowdsourcing ComputaDonal social sciences (e-‐social sciences) Computer-‐supported collaboraDve learning etc.
45
������
������
Wang
Example: What Twitter Tells Us
Computer-‐mediated communicaDon Computer-‐supported cooperaDve work Social media Social networking Online community Human computaDon Crowdsourcing ComputaDonal social sciences (e-‐social sciences) Computer-‐supported collaboraDve learning etc.
46
������
������
Wang
“Twitterology: A New Science?”
TwiVer as a micro-‐blogging service records hundreds of millions public comments from hundreds of millions of people worldwide. • TwiVer messages can possibly help us understand people’s behaviors and answer some social science quesDons
• Sampling bias: Need to keep in mind the gap between online and offline behaviors
47
hVp://www.nyDmes.com/2011/10/30/opinion/sunday/ twiVerology-‐a-‐new-‐science.html
������
������
Wang
Using Twitter Data to Study Mood Variation
Use a validated mood dicDonary to analyze TwiVer data and present paVerns of mood variaDon across hours of a day and days of a week. Show that posiDve and negaDve moods correlate with paVerns of work, sleep and daylength change.
48
Scott A. Golder and Michael W. Macy. (2011) Diurnal and Seasonal Mood Vary with Work, Sleep and Daylength Across Diverse Cultures. Science.
“Global mood swing” reflected on TwiRer. hRp://www.youtube.com/ watch?v=wp98_R1YieY
������
������
Wang
More on Research Tool How to measure posiDve and negaDve moods?
• Check the language that people use! LIWC-‐ LinguisDc Inquiry and Word Count hVp://www.liwc.net/
• Developed by UT AusDn Psychologist, J. Pennebaker • Controlled dicDonary of words that capture psychological states (e.g., 184 anger-‐related words for measuring Anger)
• Broadly accepted for social science and social compuDng research.
Word count versus NLP? • Precision versus recall • Hypothesis tesDng versus predicDon • Using LIWC features for machine learning purposes can be powerful. (e.g., Yi-‐Chia Wang et al., CSCW 2012)
49
http://kraut.hciresearch.org/sites/kraut.hciresearch.org/files/articles/yichiaw-support-revision-final-v2.1.pdf
������
������
Wang
Test It Out- The First Two Minutes of the US Presidential Debate on Oct 3, 2012
50
Transcript: http://edition.cnn.com/2012/10/03/politics/debate-transcript/index.html LIWC dimensions: http://www.kovcomp.co.uk/wordstat/LIWC.html
������
������
Wang
The Social Aspect of Research
CommuniEes of PracEce: A profession can be defined socially, including shared understanding, experience and belief that people possess and things that people do in a community. [Jean Lave & EDenne Wenger]
51
hVp://en.wikipedia.org/wiki/Community_of_pracDce
������
������
Wang
The Social Aspect of Research (cont.)
Social compuDng research is also shaped by communiDes. Different communiDes can have somewhat different views. • Choosing a community, and knowing and parDcipaDng it deeply
• Things look new, different outside of the community may look old, familiar inside the community
ACM Special Interest Group on Computer-‐Human InteracDon (SIGCHI) • Two major SIGCHI conferences: CHI and CSCW.
52
������
������
Wang
Some Major HCI Communities (Grudin, 2011)
53
������
������
Wang
CHI (Human Factors in Computing Systems)
CHI (pronounced like “Kai”) is the umbrella conference of SIGCHI • One of the oldest, starDng from 1982 (30 years) • Covering all topics in HCI • One of the largest ACM conferences, 2000-‐3000 parDcipants; more than 10 parallel sessions
• One of the hardest for paper acceptance, 20-‐25% acceptance rate
• Review process: external reviewers & AC (Associate Chair) reviewers; Face-‐to-‐face PC meeDngs for final paper selecDon.
54
������
������
Wang
CHI (Human Factors in Computing Systems) 2012 ���Paper Subcommittees
1. Usability, Accessibility and User Experience 2. Specific ApplicaDon Areas 3. InteracDon Beyond the Individual 4. Design 5. InteracDon Using Specific ModaliDes 6. Understanding People: Theory, Concepts, Methods 7. InteracDon Techniques and Devices 8. Expanding InteracDon through Technology, Systems and
Tools
55
������
������
Wang
Some CHI 2012 Photos
56
������
������
Wang
Some CHI 2012 Photos
57
������
������
Wang
CSCW (Computer-Supported Cooperative Work)
CSCW is one SIGCHI conference specialized for collaboraDve technologies and social compuDng. • Held every other year (biennially) from 1986 to 2008
• Interleaving with ECSCW • Held annually since 2010. Slight change of Dtle to “ACM Conference on Computer Supported CooperaDve Work and Social CompuDng” starDng 2013.
• Similar quality and difficulty to CHI. The first SIGCHI conference adopts a two-‐phase review process (similar to journal) since 2012.
• Smaller in size, about 600+ parDcipants. More focused, easier to socialize. Common “I liked CSCW more than CHI” comment from CSCW and social compuDng folks.
58
������
������
Wang
Reflections
59
������
������
Wang
Be Aware of the “Because It’s New” Thinking
IntuiDvely, it seems straighxorward to consider social compuDng and HCI in general are new • Facebook, TwiVer … are new
However, many relevant ideas and systems are not new • Email, instant messaging, BBS are useful but not new • The underlying technical components and ideas have much overlap
CommuniDes are not new • CHI, CSCW have been there for 30 years • Understandings of social interacDon and technical know-‐hows are accumulaDng and influencing subsequent work.
Doing it because it’s valuable but not just because it’s new.
60
������
������
Wang
The Invisible Designers “Social design”-‐ the Social ConstrucDon of Technology (SCOT)
• A sociological response to technological determinism • Social shaping of technologies (rejecDon, acceptance etc.) may also play a role. • Usability, usefulness, markeDng, social norm, culture etc.
61
http://ilikeinnovation.com/wp-content/uploads/2010/04/Picture-31.png
������
������
Wang
The Role of Culture
Social compuDng cannot work without people, and people’s thoughts and behaviors are shaped by culture (e.g., Western versus Eastern). • Important to ask how cultures differ and what’s the implicaDon to social compuDng. “One size may not fit all”
• More, perhaps we can leverage cultural characterisDcs and differences to enable useful social compuDng.
62
������
������
Wang
Finally, Revisiting “The Two Cultures”
C.P. Snow, BriDsh scienDst and writer, argued that there exists an intellectual and communicaDve gap between “the sciences” and “the humaniDes” • ScienDsts don’t know Shakespeare • Humanists don’t know Thermaldynamics • But (let’s be naive), are there any pracDcal, funcDonal reasons that the gap should be bridged?
63
http://www.scientificamerican.com/article.cfm?id=an-update-on-cp-snows-two-cultures
������
������
Wang
Bridging the Gap Creates Value
Social compuEng as a proof-‐of-‐concept that combining compuDng and social research, technologies and humaniDes can lead to concrete, beneficial outcomes
Social studies and analyses are as useful as computer
programming in social compuDng design • Viewing them as problem solving tools; creaDvely and thoughxully ge�ng value out of them
• Merging the two cultures into one problem solving culture-‐ Responding to social problems, and increasing the social contribuDons of work at both sides.
64
������
������
Wang
Thank You
65
清華大學人機合作與社群運算實驗室 NTHU CollaboraDve and Social CompuDng Lab (CSC Lab)
hVp://www.cs.nthu.edu.tw/~haochuan/ http://daoofstrategy.blogspot.com/2010/08/sign-of-times-robotics-trend-is-here.html