CSCL 2011 Keynote on Social Computing and eLearning

CSCL 2011 | Keynote Augmented Social Cognition: How SocialComputing is Changing eLearning Ed H. Chi GoogleResearch Work done whileat Palo AltoResearch Center(PARC) 2008-05-13 CSCL 2011 Keynote

2008-05-13 CSCL 2011 Keynote

Prelude: A personal learning story To: chi@acm.org From: Brad Barrish <brad@…removed.for.privacy….com> Subject: Pancreatic cancer Date: Thu, 1 Feb 2007 21:37:55 PST Hey Ed. Im a fellow del.icio.us user and noticed you bookmark a lot of pancreatic cancer stuﬀ. Im at home with my dad who was diagnosed a little over a year ago and is now at the tale end of things. Ive learned a lot through his treatments and about whats out there. I dunno if its something you or a family member has, but just wanted to drop you an email. Be well. Brad 2008-05-13 CSCL 2011 Keynote

Talk in 3 Acts The Importance of Social Signals in eLearning n  Act I: The Invisible –  Social Search n  Act II: The Visible –  Shared Annotations n  Act III: The Abstracted –  Shared Knowledge Space 2008-05-13 CSCL 2011 Keynote

Act I: Invisible Social Signals from the Crowd Joint work w/ Todd Mytkowicz, Rowan Nairn, Lawrence Lee [Chi and Mytkowicz, Hypertext2008] [Kammerer et al., CHI2009] 2008-05-13 CSCL 2011 Keynote

Using Information Theory to Model Social Tagging [Ed H. Chi, Todd Mytkowicz, ACM Hypertext 2008] Concepts Topics Users Documents Noise Tags Decoding Encoding T1…Tn 2008-05-13 CSCL 2011 Keynote

Tagging Behavior H(Tag) shows tag saturation H(Doc | Tag), browsability 2008-05-13 CSCL 2011 Keynote

Implication I(Doc; Tag) Mutual Information Raise in avg. tag / bookmark 2008-05-13 CSCL 2011 Keynote

TagSearch: MapReduce Implementation Tags URLs P(URL|Tag) P(Tag|URL) n  Spreading Activation in a bi-‐graph n  Computation over a very large data set –  150 Million+ bookmarks 2008-05-13 CSCL 2011 Keynote

TagSearch: Use Semantic Analysis to Reduce Noise http://mrtaggy.com Semantic Similarity Graph Web Tools Reference Guide Howto Tutorial Tips Help Tip Tutorials Tricks 2008-05-13 CSCL 2011 Keynote

2008-05-13 CSCL 2011 Keynote

Experiment Design [Kammerer et al. CHI2009] n  2 interface x 3 task domain design –  2 Interface (between-‐subjects) n  Exploratory vs. Baseline –  3 task domains (within-‐subjects) n  Future Architecture, Global Warming, Web Mashups n  30 Subjects (22 male, 8 female) –  Intermediate or advanced computer and web search skills –  Half assigned Exploratory, half Baseline. n  For each domain, single block with 3 task types: –  Easy and Diﬃcult Page Collection Task [6min each] –  Summarization Task [12min] –  Keyword Generation Task [2min] 2008-05-13 CSCL 2011 Keynote

Evauation Results [Kammerer et al., CHI2009] n  Exploratory interface users: –  performed more queries, –  took more time, –  wrote better summaries (in 2/3 domains), –  generated more relevant keywords (in 2/3 domains), and –  had a higher cognitive load. n  Suggestive of deeper engagement and better learning. n  Some evidence of scaﬀolding for novices in the keyword generation and summarization tasks. 2008-05-13 CSCL 2011 Keynote

Act II: Visible Social Signals from Shared Highlighting Kudos to Lichan Hong, Les Nelson [Hong et al, AVI2008] [Nelson et al., HCII 2009] 2008-05-13 CSCL 2011 Keynote

Finding a Restaurant n  Appropriate for the occasion 2008-05-13 CSCL 2011 Keynote

Heuristics Poor heuristic Good heuristic 2008-05-13 CSCL 2011 Keynote

“Hints” Solo Cooperative (“good hints”) 2008-05-13 CSCL 2011 Keynote

SparTag.us: Social Highlighting 2008-05-13 CSCL 2011 Keynote

SparTag.us: Social Highlighting n  In situ tagging while reading n  Highlighting n  Shared notebooking n  Sharing! 2008-05-13 CSCL 2011 Keynote

Highlighting as Importance Indicator recall first-visit 2008-05-13 CSCL 2011 Keynote

Evaluation Task & Metric [Nelson et al., HCII2009] n  Sensemaking task –  Find and read material about Enterprise 2.0 mashups in order to write two essays n  Seeds: expert content for scaﬀolding –  Tags from del.icio.us –  URLs from Google/PageRank –  Constructed and then shared through social mechanisms (i.e., a SparTag.us friend ) n  Performance Measures –  Learning gain: Pre/Post Knowledge Test Posttest score - Pretest score Gain = Max score - Pretest score 2008-05-13 CSCL 2011 Keynote

Procedure SparTag.us SF with Friend SparTag.us Demographics SO Only Posttest & Pretest Without WS SparTag.us 2008-05-13 CSCL 2011 Keynote

Results: Learning Gain N=18 SparTag.us + Friend superior to both individual conditions No diﬀerence between the two control conditions 2008-05-13 CSCL 2011 Keynote

URL Kind Code Blog B Observation Conference C Employment E My.Spartag.us M News N URL KindOpenSource Code O Blog Search B S Conference C Vendor V Employment E Wikipedia MySpartagus M W News Consultant N X OpenSource O Search S Vendor V Wikipedia W Consultant X 2008-05-13 CSCL 2011 Keynote

Von Restorff Isolation Effect [1933] n  As applied to highlights, the von Restorff isolation effect suggests that readers: n  (a) tend to focus on and n  (b) learn what is marked, n  whether the information is important or not. –  Nist and Hogrebe 87 2008-05-13 CSCL 2011 Keynote

Act III: Abstracted Knowledge: The Science of Understanding Wikipedia Kudos to Bongwon Suh, Niki Kittur [Kittur et al., CHI2007] [Suh et al., WikiSym 2009] 2008-05-13 CSCL 2011 Keynote

Exponential Growth of Wikipedia: an accepted ‘fact’ Number of Articles (Log Scale) http://en.wikipedia.org/wiki/Wikipedia:Modelling_Wikipedia’s_growth 2008-05-13 CSCL 2011 Keynote

Growth of Edits 2008-05-13 CSCL 2011 Keynote

Something happened in early 2007 2008-05-13 CSCL 2011 Keynote

Growth of Active Editors *In thousands 2008-05-13 CSCL 2011 Keynote

Slowing Growth in Global Activity *In thousands 2008-05-13 CSCL 2011 Keynote

Earlier Exponential Growth Model n  Preferential Attachment: Edits beget edits –  more number of previous edits, more number of new edits Growth rate depends on: N = current population r = growth rate of the population N(t) = N 0 " e rt dN = r" N dt Growth rate Current of population ! population! 2008-05-13 CSCL 2011 Keynote

Logistic Growth Model n  Ecological population growth model –  Also depend on environmental conditions –  K, carrying capacity (due to resource limitation) dN N = rN(1" ) dt K 2008-05-13 CSCL 2011 Keynote

Match to Data: # of New Articles n  Follows a logistic growth curve New Article 2008-05-13 CSCL 2011 Keynote

Struggle for Existence -‐ Darwin n  Biological system –  Competition increases as population hit the limits of the ecology –  Advantage go to members of the population that have competitive dominance over others n  Analogy –  Limited opportunities to make novel contributions –  Increased patterns of conﬂict and dominance 2008-05-13 CSCL 2011 Keynote

“Showering” Hypothesis What drives contributions to Wikipedia? Cooperation is not the main driver? n  Hypothesis: Conflicts drives most of the contributions. –  How do we measure conflicts? n  Conflicts cause coordination costs to go up. –  How to measure coordination costs? n  “negotiation is critical to helping multiple perspectives to converge on shared knowledge.” –  Stahl, Group Cognition, Ch8, 2004 2008-05-13 CSCL 2011 Keynote

Conﬂict/Coordination Eﬀects in Wikipedia (Kittur, Suh, Pendleton, Chi, CHI2007) 2008-05-13 CSCL 2011 Keynote

Ratio of Reverted Contributions Monthly Ratio of Reverted Edits 2008-05-13 CSCL 2011 Keynote

Visual Analytics over Wikipedia data Mediator Pattern -‐ Terri Schiavo [Suh, et al., VAST2007] Anonymous (vandals/ spammers) Sympathetic to husband Mediators Sympathetic to parents 2008-05-13 CSCL 2011 Keynote

WikiDashboard.com 2008-05-13 CSCL 2011 Keynote

Coda: A Challenge: A modiﬁed logistic model n  Carrying Capacity as a function of time. 2008-05-13 CSCL 2011 Keynote

What Did We Learn? n  The Common Thread: –  Utilization of Social Signals for Learning and Information Access –  Whether it is invisible, visible, and abstracted. n  The Establishment of Common Ground –  Implicit Coordination –  Explicit Coordination –  Negotiation n  “All collective actions are built on common ground and its accumulation.” –  Clark and Brennan, 1991 2008-05-13 CSCL 2011 Keynote

Research Vision Augmented Social Cognition n  Cognition: the ability to remember, think, and reason; the faculty of knowing. n  Social Cognition: the ability of a group to remember, think, and reason; the construction of knowledge structures by a group. –  (not quite the same as in the branch of psychology that studies the cognitive processes involved in social interaction, though included) n  Augmented Social Cognition: Supported by systems, the enhancement of the ability of a group to remember, think, and reason; the system-‐supported construction of knowledge structures by a group. Citation: Chi, IEEE Computer, Sept 2008 2008-05-13 CSCL 2011 Keynote

From Rote Learning to Interaction 2008-05-13 CSCL 2011 Keynote

Thank you! n  chi@acm.org n  http://edchi.net 2008-05-13 CSCL 2011 Keynote

What I will not talk about … n  Motivation –  Cultural and economic incentives –  Personal and societal values –  Psychology (e.g. cognitive, personality, social) n  Policy and Investment –  Resources –  Teacher training –  Technological investment n  With the Assumption of Motivation and Resources, how to make information universally accessible and useful in a Web2.0 world? 2008-05-13 CSCL 2011 Keynote

2008-05-13 CSCL 2011 Keynote

Lowering Participation / Interaction Costs n  Interaction costs # People willing to produce for “free” determine number of people who participate n  Surplus of attention & motivation at small transaction costs n  Therefore… n  Important to keep interaction costs low Cost of participation 2008-05-13 CSCL 2011 Keynote

Using Machine Learning to Detect Conﬂicts n  Counting ‘Controversial’ labels n  5x cross-‐validation, R2 = 0.897 10000 9000 Actual controversial revisions 8000 7000 6000 5000 4000 3000 2000 1000 0 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 Predicted controversial revisions 2008-05-13 CSCL 2011 Keynote

Collaborative Knowledge Building n  “They cannot even begin to coordinate on content without assuming a vast amount of shared information or common ground…. And to coordinate on process, they need to update their common ground moment by moment. All collective actions are built on common ground and its accumulation.” –  Clark and Brennan, 1991 n  At Web-‐scale social learning, what we know about the nature of conﬂict and negotiation is woefully inadequate. 2008-05-13 CSCL 2011 Keynote

Google Plus as a Research Platform 2008-05-13 CSCL 2011 Keynote

http://www.isls.org	166
http://www.scoop.it	16
http://paper.li	13
http://design.cite.hku.hk	12
https://wwwtest.ole.bris.ac.uk	12
http://bb-sandbox.isys.bris.ac.uk	7
http://webcache.googleusercontent.com	2
http://isls.org	2
http://translate.googleusercontent.com	1

CSCL 2011 Keynote on Social Computing and eLearning

by Ed Chi on Jul 05, 2011

Accessibility

Categories

Tags

Upload Details

Usage Rights

9 Embeds 231

Statistics

CSCL 2011 Keynote on Social Computing and eLearning — Presentation Transcript