SlideShare une entreprise Scribd logo
1  sur  36
Télécharger pour lire hors ligne
The Power of Known
Peers: A Study in Two
Domains
Peter Brusilovsky with
Danielle Lee and
Sharon I-Han Hsiao
Overview
•  The context
•  The problem
•  The goal
•  The system
•  The study
University of Pittsburgh - PAWS Lab 2
http://www.masternewmedia.org/news/2006/12/01/social_bookmarking_services_and_tools.htm
The Wisdom of Crowds
Social Information Access
•  Social Navigation
–  Social support of user browsing
•  Social Recommendation (Collaborative Filtering)
–  Proactive information access
•  Social Search
–  Social support of search
•  Social Visualization
–  Social support for visualization-based access to information
•  Social Bookmarking
–  Access to bookmarked/shared information facilitated with tags
Social Navigation: The Start
•  Natural tendency of people to follow
each other
–  Making use of direct and indirect
cues about the activities of others
–  Following trails
•  Footsteps in sand or snow
•  Worn-out carpet
–  Using dogears and annotations
–  Giving direction or guidance
•  Navigation driven by the actions
from one or more advice
providers
• The pioneer idea of
asynchronous indirect
social navigation
• Developed for
collaborating writing and
editing
• Indicated read/edited
places in a large document
Edit Wear and Read Wear (1992)
Social Information Access
•  History-enriched environments
–  Edit Wear and Read Wear (1992)
–  Social navigation systems
•  Footprints, Juggler, Kalas
•  Collaborative filtering
–  Manual push and pull
•  Tapestry, LN Recommender
–  Modern automatic CF recommender systemss
•  Social Search
–  Quest-based systems
•  AntWorld
–  Group-based search (i-Spy)
From People to Crowds
•  It started with people following other people
–  ReadWear, Tapestry, AntWorld
•  But we need to scale these ideas up!
•  Let’s move from people to faceless crowds
–  Follow-the-crowd social navigation
–  Collaborative filtering
–  Group-based on community-based social search
11/15/2011
University of Pittsburgh - PAWS Lab 8
We Lost People in Crowds…
•  Crowd-based approach does work, but there are
issues
•  Less trust to a faceless crowd
•  Less motivation to follow
•  Malicious users and attacks
•  Should we step back?
–  Start seeing people in crowds?
11/15/2011
University of Pittsburgh - PAWS Lab 9
Brusilovsky, P., Chavan, G., Farzan, R., Social Adaptive Navigation
Support for Open Corpus Electronic Textbooks, AH2004
	
10/19	
Knowledge Sea: Social Navigation
MovieLens: Collaborative Filtering
11
I-SPY: Community-Based Search
HOW TO IMPROVE
RECOMMENDATIONS USING
VARIOUS SOCIAL NETWORKS
Exploring Watching Networks, Group Co-members and
Research Collaborators as a source of Recommendation
Danielle Lee
Why to Use Online Social Networks?
•  Connection in social networks are typically
known to users
•  Connected people have reasonably similar
interests
•  People tend to trust their connections more than
faceless peers
•  People are easily get influenced by those they
know
•  Address “Cold Start” problem
•  Decrease the risk of misuse and attacks
Paper Domain (Dataset)
Trust-based
Networks
Avesani, et al (2005) Ski Resorts (Moleskiing.it)
Al-Sharawneh & Williams (2010) General Items (Epinions)
Jamali & Ester (2009) General Items (Epinions)
Jamali & Ester (2011) General Items (Epinions) & Movies (Flixster)
Ma, et al. (2008) General Items (Epinions)
Massa & Avesani (2007) General Items (Epinions)
Walter, et al. (2009) General Items (Epinions)
DuBois, et al. (2009) Movies (FilmTrust)
Golbeck & Hendler (2006) Movies (FilmTrust)
Matsuo & Yamamoto (2007) Cosmetics (@cosme)
Friendships Bonhard, et al. (2007) Movies (MovieMatch)
Bourke, et al. (2011) Movies/TV(Facebook)
Groh & Ehmig (2007) Local Clubs (A German Site)
Liu and Lee (2010) Online Products (Cyworld)
Pera & Ng (2011) Book (Amazon and LibraryThing)
Sinha & Swearingen (2001)
Books (Amazon, Sleeper & RatingZones) and
Movies (Amazon, Reel.com, and MovieCritics)
Konstas, et al. (2009) Music (Last.fm)
Colleagues Guy, et al. (2009) Bookmarks of Web Pages (Lotus Connections)
Group Member Yuan, et al., (2009) Music (Last.fm)
Recommendations Based on Watching
•  User-assigned unilateral connections based on their interests
–  Highly object-centered relations and low personal familiarity
–  Users concentrate on the usefulness of watched partners’ information
collections.
–  Meets the ‘Similarity Attraction theory’ and holds ‘transitive power’.
–  Mimics the process of bookmarking interesting items.
•  E.g. “following” on Twitter, “plus one” on Google, “watching” on
Citeulike, “network” on Delicious and “contacts” on Flickr.
•  This study is based on a Citeulike Data set provided by the system
–  97,712 Users, 3,297,156 articles, 3,869,993 bookmarks and 44,847 watching
relations
–  The data set contains publications, the metadata (titles, author names,
publication name, publication years, etc.), tags and users’ bookmarks
Homophily in Watching Networks
•  Users in watching relations have more common
information items, metadata & tags than random
pairs
–  The similarity was the largest for direct connections and decreased
with the increase of social distance between users.
–  In particular, users connected by watching relations tend to co-
bookmark the same items.
–  The items shared by two users in direct watching relations are
more rare and have similar contents and context.
Co-
bookmarks
Jaccard Popularity
Log-
Likelihood
Title
Vector
Author
Name
Vector
Tag Vector
Direct 1.80 0.21% 8.69 .204 .1440 .0149 .0505
1 Hop .39 0.04% 7.75 .097 .0814 .0033 .0168
2 Hops .16 0.02% 7.38 .061 .0626 .0020 .0114
No
Relation
.04 0.02% 6.92 .023 .0147 .0007 .0020
Recommendations in Watching Networks
•  Fusing watching relations with traditional collaborative
filtering recommendations improves the quality
Group-Based Link Homophily
•  A group of people who are interested in the same topic places
uses in a specific kind of social relationship that can be used
for improving recommendations
•  The homophily study based on a Citeulike Data set provided
by the system:
–  12,944 Users, 4,109 Groups and 18,793 Membership
•  Information overlap between group co-members is
significantly larger than the overlap between random pairs.
Co-
bookmark
s
Jaccard Popularity
Log-
Likelihoo
d
Title
Vector
Author
Name
Vector
Tag
Vector
Group
Co-
Member
s
.26 1.01% 8.00 .050 .1117 .0222 .0595
No
Relation
.04 0.02% 6.92 .023 .0147 .0007 .0020
Group-based Recommendations
•  Matrix Factorization Recommendations based on Group library and
Group Co-members’ library performed the best
CF – Collaborative Filtering; Gmem – Group Comembers-based; Group – Comembers & Group-based
.000
.005
.010
.015
.020
.025
.030
.035
.040
.045
.050
CF
Gmem
Group
CF
Gmem
Group
CF
Gmem
Group
CF
Gmem
Group
Jaccard Similarity Matrix
Factorization
Jaccard Similarity Matrix
Factorization
Top5 Top2_F1
Group-based Recommendations for
everyone?
•  The idea of group-based recommendations is to pick candidate items from
those that are not yet discovered by target users, but available in the group
library and the co-members’ repositories.
•  Therefore, users in the area A might not benefit from group-based
recommendation.
Group-based Recommendations
•  Different Performance of Group-based recommendations depending
to Users’ position.
–  For the dictators who dominated their group activities, the recommendations
based on group information didn’t perform well, compared with other user
clusters.
.000
.005
.010
.015
.020
.025
.030
.035
Top5_F1measure
CF
Gmem
Group
CF_SVD
Gmem_SVD
Group_SVD
.00
.01
.02
.03
.04
.05
.06
.07
.08
Top2_F1measure
CF
Gmem
Group
CF_SVD
Gmem_SVD
Group_SVD
Recommendations Based on Research
Collaborators
•  Users in research collaborations interact to each other
personally and their relations are centered on their
research topics and the relevant by-products.
–  Online social networks for professionals is to implement offline
referral chains on the Web.
•  This study is based on Conference Navigator (current
version 3; hence it is CN3, now), a social adaptive system
to support conference attendees.
–  464 users, 1000 conference talks of 15 conferences, 189
collaboration relations, 144 social connections on CN3, and
5,094 bookmarks
–  Data set contains conference talks, the metadata (titles, author
names, publication name, publication years, etc.), users’
bookmarks and users’ own publication records.
Recommendations Based on Research
Collaborators: Results
•  Social Network-based Recommendations utilizing
content information of objects were the most effective
recommendation approach.
.00
.05
.10
.15
.20
.25
.30
CF
Community
CFCW
Profile
SVD
SN_Colleagues
SN_CN3
SN_Both
SN_SVD
SNCW_Colleagues
SNCW_CN3
SNCW_Both
Baseline SN SNCW
Top5_F1Measure
.00
.05
.10
.15
.20
.25
.30
.35
.40
.45
CF
Community
CFCW
Profile
SVD
SN_Colleagues
SN_CN3
SN_Both
SN_SVD
SNCW_Colleagues
SNCW_CN3
SNCW_Both
Baseline SN SNCW
Top2_F1Measure
References
•  Watching Relation-based Recommendations
–  Lee, D. H. & Brusilovsky, P. (2011) Improving Recommendations using Watching Networks in a
Social Tagging System, Proceedings of iConference 2011, Seattle, WA, USA, February 8 ~ 11, 2011
–  Lee, D. H. & Brusilovsky, P. (2010) Social Networks and Interest Similarity: The Case of
CiteULike, Proceedings of the 21st ACM Conference on Hypertext and Hypermedia (Hypertext),
Toronto Canada, June 14 ~ 16, 2010
•  Group-based Recommendations
–  Lee, D. H. & Brusilovsky, P. (2010) Using Self-Defined Group Activities for Improving
Recommendations in Collaborative Tagging Systems, Proceedings of the 3rd ACM Conference on
Recommender Systems (Recsys), Barcelona, Spain, September 26 ~ 30, 2010
–  Lee, D. H., Brusilovsky, P. & Schleyer, T. (Under Review) Group-based Recommendations for
Individual Members, Proceedings of the 21st ACM International Conference on Information and
Knowledge Management (CIKM 2012), Maui, Hawaii,USA, October 29-November 2, 2012
•  Collaborator-based Recommendations
–  Lee, D. H. & Brusilovsky, P. (Under Review) Exploring Social Approach to Recommend Talks at
Research Conferences, Proceedings of the 8th IEEE International Conference on Collaborative
Computing: Networking Applications and Worksharing (CollaborateCom 2012)
HOW TO PROVIDE SOCIAL
GUIDANCE TO LEARNING
RESOURCES
Who guides us better – a crowd or peers?
Sharon I-Han Hsiao
A Quest to Building a Social QuizGuide
11/15/2011
University of Pittsburgh - PAWS Lab 27
28
Good personalized guidance: improved problem solving success!
The more the students compared to their peers, the higher post-quiz scores they
received (r= 0.34 p=0.004)
Parallel Introspective Views
29
• Pros: Liked OUM, interactivity with the content, social guidance
• Cons: dense and complicated with increasing activities
QuizMap
30
Progressor:
• Higher Engagement: Increased the questions attempts and topic coverage
• Increased problem solving success
• Significant positive correlations between the frequencies of peer model sorting and
question attempts and success rate, r= 0.75, p< .01; r= 0.76, p< .01.
Progressor
The Effect of Visible Peers
QuizJET w/ IV Progressor
Parameters n=18 n=30
Peers 6.83±2.25 !"#!$%"&%'
Topics 4.00±0.79 ("))$%"&('
Questions 4.67±1.36 *"&&$%"%+'
,-./' 01'
2.-3.455'
&6'
,-./'01'7894' +:'
,-./'01'-247' 6&'
'
• Adding additional collection did not sacrifice the usage
• Increased the engagement (Quiz =: 5 hours, Example: 5 hours 20 mins)
• Increased diversity helped increase problem solving success
• Mix collections resulted in uniform performance
Progressor+
80.81
125.5
205.73 190.42
0
50
100
150
200
250
300
Non-adaptive Non-social ANS Progressor Progressor+
Attempts
References
•  Hsiao, I-H. and Brusilovsky, P. (2012) Motivational Social Visualizations for
Personalized E-learning, In: Proceedings of 7th European Conference on Technology
Enhanced Education (ECTEL), ECTEL 2012, Saarbrücken, Germany, September 18-21,
2012, Springer-Verlag, (to be appeared)
•  Hsiao, I-H., Guerra, J., Parra, D., Bakalov, F., König-Ries, B., and Brusilovsky, P. (2012)
Comparative Social Visualization for Personalized E-Learning. International
Working Conference Advanced Visual Interfaces, AVI 2012, Capri, Italy, May 21-25, 2012,
Proceeding AVI '12 Proceedings of the International Working Conference on Advanced
Visual Interfaces, Pages: 303-307, ACM New York, NY, USA
•  Bakalov, F., Hsiao, I-H., Brusilovsky, P., and König-Ries, B. (2011) Progressor:
Personalized visual access to programming problems, IEEE Symposium on Visual
Languages and Human-Centric Computing, September 18-22, 2011, Pittsburgh, PA, USA
•  Hsiao, I-H., Bakalov, F., Brusilovsky, P., and König-Ries, B. (2011) Open Social Student
Modeling: Visualizing Student Models with Parallel IntrospectiveViews.
Proceedings of 19th International Conference on User Modeling, Adaptation, and
Personalization (UMAP 2011), Girona, Spain, July 11-15, 2011, Springer, pp.171-182
Eventur.us
University of Pittsburgh - PAWS Lab 34
CoMeT (http://halley.exp.sis.pitt.edu/comet/)
University of Pittsburgh - PAWS Lab 35
Conference Navigator III
University of Pittsburgh - PAWS Lab 36
http://halley.exp.sis.pitt.edu/cn3/legacy.php?conferenceID=85

Contenu connexe

Tendances

ACRL 2015 presentation: Models of Library Engagement with MOOCs
ACRL 2015 presentation: Models of Library Engagement with MOOCsACRL 2015 presentation: Models of Library Engagement with MOOCs
ACRL 2015 presentation: Models of Library Engagement with MOOCs
Laura O'Brien
 
The changing nature of scholarly communication - What does this mean for rese...
The changing nature of scholarly communication - What does this mean for rese...The changing nature of scholarly communication - What does this mean for rese...
The changing nature of scholarly communication - What does this mean for rese...
Research Information Network
 
Indexing presentation 2013 06-04
Indexing presentation 2013 06-04Indexing presentation 2013 06-04
Indexing presentation 2013 06-04
Louise Spiteri
 
Social Networks and Social Capital
Social Networks and Social CapitalSocial Networks and Social Capital
Social Networks and Social Capital
Giorgos Cheliotis
 

Tendances (17)

5-14-13 An Introduction to VIVO Presentation Slides
5-14-13 An Introduction to VIVO Presentation Slides5-14-13 An Introduction to VIVO Presentation Slides
5-14-13 An Introduction to VIVO Presentation Slides
 
6.25.14 Meet the VIVO Project Director Presentation Slides
6.25.14 Meet the VIVO Project Director Presentation Slides6.25.14 Meet the VIVO Project Director Presentation Slides
6.25.14 Meet the VIVO Project Director Presentation Slides
 
Alone Together: Patterns of collaboration in free and open source software de...
Alone Together: Patterns of collaboration in free and open source software de...Alone Together: Patterns of collaboration in free and open source software de...
Alone Together: Patterns of collaboration in free and open source software de...
 
2.24.16 Slides, “VIVO plus SHARE: Closing the Loop on Tracking Scholarly Acti...
2.24.16 Slides, “VIVO plus SHARE: Closing the Loop on Tracking Scholarly Acti...2.24.16 Slides, “VIVO plus SHARE: Closing the Loop on Tracking Scholarly Acti...
2.24.16 Slides, “VIVO plus SHARE: Closing the Loop on Tracking Scholarly Acti...
 
6-4-13 VIVO Case Studies Presentation Slides
6-4-13 VIVO Case Studies Presentation Slides6-4-13 VIVO Case Studies Presentation Slides
6-4-13 VIVO Case Studies Presentation Slides
 
Next Generation Systems
Next Generation SystemsNext Generation Systems
Next Generation Systems
 
Gogia mock prospectus
Gogia mock prospectusGogia mock prospectus
Gogia mock prospectus
 
ACRL 2015 presentation: Models of Library Engagement with MOOCs
ACRL 2015 presentation: Models of Library Engagement with MOOCsACRL 2015 presentation: Models of Library Engagement with MOOCs
ACRL 2015 presentation: Models of Library Engagement with MOOCs
 
Digital Commons Institutional Repository: Roles for Library Liaisons
Digital Commons Institutional Repository: Roles for Library LiaisonsDigital Commons Institutional Repository: Roles for Library Liaisons
Digital Commons Institutional Repository: Roles for Library Liaisons
 
The changing nature of scholarly communication - What does this mean for rese...
The changing nature of scholarly communication - What does this mean for rese...The changing nature of scholarly communication - What does this mean for rese...
The changing nature of scholarly communication - What does this mean for rese...
 
11.13.14 Slides, “SHARE: An Overview”
11.13.14 Slides, “SHARE: An Overview”11.13.14 Slides, “SHARE: An Overview”
11.13.14 Slides, “SHARE: An Overview”
 
The Future of Research Communications and e-Scholarship: Are we there yet?
The Future of Research Communications and e-Scholarship: Are we there yet?The Future of Research Communications and e-Scholarship: Are we there yet?
The Future of Research Communications and e-Scholarship: Are we there yet?
 
Indexing presentation 2013 06-04
Indexing presentation 2013 06-04Indexing presentation 2013 06-04
Indexing presentation 2013 06-04
 
An Introduction to NodeXL for Social Scientists
An Introduction to NodeXL for Social ScientistsAn Introduction to NodeXL for Social Scientists
An Introduction to NodeXL for Social Scientists
 
Social networking and its impact on libraries
Social networking and its impact on libraries  Social networking and its impact on libraries
Social networking and its impact on libraries
 
Social Networks and Social Capital
Social Networks and Social CapitalSocial Networks and Social Capital
Social Networks and Social Capital
 
Facilitating collaboration
Facilitating collaborationFacilitating collaboration
Facilitating collaboration
 

Similaire à The Power of Known Peers: A Study in Two Domains

Social information Access Tutorial at UMAP 2014
Social information Access Tutorial at UMAP 2014Social information Access Tutorial at UMAP 2014
Social information Access Tutorial at UMAP 2014
Peter Brusilovsky
 
#lak2013, Leuven, DC slides, #learninganalytics
#lak2013, Leuven, DC slides, #learninganalytics#lak2013, Leuven, DC slides, #learninganalytics
#lak2013, Leuven, DC slides, #learninganalytics
Soudé Fazeli
 
282 sharon mombru ssp meeting new business models online communities 2905
282 sharon mombru ssp meeting new business models online communities 2905282 sharon mombru ssp meeting new business models online communities 2905
282 sharon mombru ssp meeting new business models online communities 2905
Society for Scholarly Publishing
 
Global Redirective Practices
Global Redirective PracticesGlobal Redirective Practices
Global Redirective Practices
adjwilli
 

Similaire à The Power of Known Peers: A Study in Two Domains (20)

Social information Access2012
Social information Access2012Social information Access2012
Social information Access2012
 
Social information Access Tutorial at UMAP 2014
Social information Access Tutorial at UMAP 2014Social information Access Tutorial at UMAP 2014
Social information Access Tutorial at UMAP 2014
 
Presentation at School of Information and Library Science, UNC, USA
Presentation at School of Information and Library Science, UNC, USAPresentation at School of Information and Library Science, UNC, USA
Presentation at School of Information and Library Science, UNC, USA
 
Crowd Sourcing of Library Services
Crowd Sourcing of Library ServicesCrowd Sourcing of Library Services
Crowd Sourcing of Library Services
 
Csora, "2Collab, The Research Collaboration Tool"
Csora, "2Collab, The Research Collaboration Tool"Csora, "2Collab, The Research Collaboration Tool"
Csora, "2Collab, The Research Collaboration Tool"
 
#lak2013, Leuven, DC slides, #learninganalytics
#lak2013, Leuven, DC slides, #learninganalytics#lak2013, Leuven, DC slides, #learninganalytics
#lak2013, Leuven, DC slides, #learninganalytics
 
Lagace - Copyright Clearance Center April 2, 2015
Lagace - Copyright Clearance Center April 2, 2015Lagace - Copyright Clearance Center April 2, 2015
Lagace - Copyright Clearance Center April 2, 2015
 
ELIXIR Webinar: BioSharing
ELIXIR Webinar: BioSharingELIXIR Webinar: BioSharing
ELIXIR Webinar: BioSharing
 
Where Have We Been & Where Are We Going?
Where Have We Been & Where Are We Going?Where Have We Been & Where Are We Going?
Where Have We Been & Where Are We Going?
 
Global Redirective Practices
Global Redirective PracticesGlobal Redirective Practices
Global Redirective Practices
 
Harnessing the Potential of Social Networks: The ABCs of using social network...
Harnessing the Potential of Social Networks: The ABCs of using social network...Harnessing the Potential of Social Networks: The ABCs of using social network...
Harnessing the Potential of Social Networks: The ABCs of using social network...
 
FORCE11: Creating a data and tools ecosystem
FORCE11:  Creating a data and tools ecosystemFORCE11:  Creating a data and tools ecosystem
FORCE11: Creating a data and tools ecosystem
 
6 oct15 writing kmb plan edited
6 oct15 writing kmb plan edited6 oct15 writing kmb plan edited
6 oct15 writing kmb plan edited
 
Overview to: BBSRC Oxford Doctoral Training Partnership - Dr Sansone - July 2014
Overview to: BBSRC Oxford Doctoral Training Partnership - Dr Sansone - July 2014Overview to: BBSRC Oxford Doctoral Training Partnership - Dr Sansone - July 2014
Overview to: BBSRC Oxford Doctoral Training Partnership - Dr Sansone - July 2014
 
Working with Social Media Data: Ethics & good practice around collecting, usi...
Working with Social Media Data: Ethics & good practice around collecting, usi...Working with Social Media Data: Ethics & good practice around collecting, usi...
Working with Social Media Data: Ethics & good practice around collecting, usi...
 
282 mombru
282 mombru282 mombru
282 mombru
 
282 sharon mombru ssp meeting new business models online communities 2905
282 sharon mombru ssp meeting new business models online communities 2905282 sharon mombru ssp meeting new business models online communities 2905
282 sharon mombru ssp meeting new business models online communities 2905
 
Lern, jan 2015, digital media slides
Lern, jan 2015, digital media slidesLern, jan 2015, digital media slides
Lern, jan 2015, digital media slides
 
Global Redirective Practices
Global Redirective PracticesGlobal Redirective Practices
Global Redirective Practices
 
Exploring Generative Models of Tripartite Graphs for Recommendation in Social...
Exploring Generative Models of Tripartite Graphs for Recommendation in Social...Exploring Generative Models of Tripartite Graphs for Recommendation in Social...
Exploring Generative Models of Tripartite Graphs for Recommendation in Social...
 

Plus de Peter Brusilovsky

User Control in Adaptive Information Access
User Control in Adaptive Information AccessUser Control in Adaptive Information Access
User Control in Adaptive Information Access
Peter Brusilovsky
 
Two Brains are Better than One: User Control in Adaptive Information Access
Two Brains are Better than One: User Control in Adaptive Information AccessTwo Brains are Better than One: User Control in Adaptive Information Access
Two Brains are Better than One: User Control in Adaptive Information Access
Peter Brusilovsky
 
Personalized Online Practice Systems for Learning Programming
Personalized Online Practice Systems for Learning ProgrammingPersonalized Online Practice Systems for Learning Programming
Personalized Online Practice Systems for Learning Programming
Peter Brusilovsky
 
Stereotype Modeling for Problem-Solving Performance Predictions in MOOCs and ...
Stereotype Modeling for Problem-Solving Performance Predictions in MOOCs and ...Stereotype Modeling for Problem-Solving Performance Predictions in MOOCs and ...
Stereotype Modeling for Problem-Solving Performance Predictions in MOOCs and ...
Peter Brusilovsky
 

Plus de Peter Brusilovsky (20)

SANN: Programming Code Representation Using Attention Neural Network with Opt...
SANN: Programming Code Representation Using Attention Neural Network with Opt...SANN: Programming Code Representation Using Attention Neural Network with Opt...
SANN: Programming Code Representation Using Attention Neural Network with Opt...
 
Computer Science Education: Tools and Data
Computer Science Education: Tools and DataComputer Science Education: Tools and Data
Computer Science Education: Tools and Data
 
Personalized Learning: Expanding the Social Impact of AI
Personalized Learning: Expanding the Social Impact of AIPersonalized Learning: Expanding the Social Impact of AI
Personalized Learning: Expanding the Social Impact of AI
 
Action Sequence Mining and Behavior Pattern Analysis for User Modeling
Action Sequence Mining and Behavior Pattern Analysis for User ModelingAction Sequence Mining and Behavior Pattern Analysis for User Modeling
Action Sequence Mining and Behavior Pattern Analysis for User Modeling
 
User Control in Adaptive Information Access
User Control in Adaptive Information AccessUser Control in Adaptive Information Access
User Control in Adaptive Information Access
 
Human-Centered AI in AI-ED - Keynote at AAAI 2022 AI for Education workshop
Human-Centered AI in AI-ED - Keynote at AAAI 2022 AI for Education workshopHuman-Centered AI in AI-ED - Keynote at AAAI 2022 AI for Education workshop
Human-Centered AI in AI-ED - Keynote at AAAI 2022 AI for Education workshop
 
User Control in AIED (Artificial Intelligence in Education)
User Control in AIED (Artificial Intelligence in Education)User Control in AIED (Artificial Intelligence in Education)
User Control in AIED (Artificial Intelligence in Education)
 
The Return of Intelligent Textbooks - ITS 2021 keynote talk
The Return of Intelligent Textbooks - ITS 2021 keynote talkThe Return of Intelligent Textbooks - ITS 2021 keynote talk
The Return of Intelligent Textbooks - ITS 2021 keynote talk
 
Data-Driven Education 2020: Using Big Educational Data to Improve Teaching an...
Data-Driven Education 2020: Using Big Educational Data to Improve Teaching an...Data-Driven Education 2020: Using Big Educational Data to Improve Teaching an...
Data-Driven Education 2020: Using Big Educational Data to Improve Teaching an...
 
Two Brains are Better than One: User Control in Adaptive Information Access
Two Brains are Better than One: User Control in Adaptive Information AccessTwo Brains are Better than One: User Control in Adaptive Information Access
Two Brains are Better than One: User Control in Adaptive Information Access
 
An Infrastructure for Sustainable Innovation and Research in Computer Scienc...
An Infrastructure for Sustainable Innovation and Research in Computer Scienc...An Infrastructure for Sustainable Innovation and Research in Computer Scienc...
An Infrastructure for Sustainable Innovation and Research in Computer Scienc...
 
Personalized Online Practice Systems for Learning Programming
Personalized Online Practice Systems for Learning ProgrammingPersonalized Online Practice Systems for Learning Programming
Personalized Online Practice Systems for Learning Programming
 
Human Interfaces to Artificial Intelligence in Education
Human Interfaces to Artificial Intelligence in EducationHuman Interfaces to Artificial Intelligence in Education
Human Interfaces to Artificial Intelligence in Education
 
Interfaces for User-Controlled and Transparent Recommendations
Interfaces for User-Controlled and Transparent RecommendationsInterfaces for User-Controlled and Transparent Recommendations
Interfaces for User-Controlled and Transparent Recommendations
 
UMAP 2019 talk Evaluating Visual Explanations for Similarity-Based Recommenda...
UMAP 2019 talk Evaluating Visual Explanations for Similarity-Based Recommenda...UMAP 2019 talk Evaluating Visual Explanations for Similarity-Based Recommenda...
UMAP 2019 talk Evaluating Visual Explanations for Similarity-Based Recommenda...
 
Course-Adaptive Content Recommender for Course Authoring
Course-Adaptive Content Recommender for Course AuthoringCourse-Adaptive Content Recommender for Course Authoring
Course-Adaptive Content Recommender for Course Authoring
 
The User Side of Personalization: How Personalization Affects the Users
The User Side of Personalization: How Personalization Affects the UsersThe User Side of Personalization: How Personalization Affects the Users
The User Side of Personalization: How Personalization Affects the Users
 
Data driveneducationicwl2016
Data driveneducationicwl2016Data driveneducationicwl2016
Data driveneducationicwl2016
 
From Expert-Driven to Data-Driven Adaptive Learning
From Expert-Driven to Data-Driven Adaptive LearningFrom Expert-Driven to Data-Driven Adaptive Learning
From Expert-Driven to Data-Driven Adaptive Learning
 
Stereotype Modeling for Problem-Solving Performance Predictions in MOOCs and ...
Stereotype Modeling for Problem-Solving Performance Predictions in MOOCs and ...Stereotype Modeling for Problem-Solving Performance Predictions in MOOCs and ...
Stereotype Modeling for Problem-Solving Performance Predictions in MOOCs and ...
 

Dernier

Salient Features of India constitution especially power and functions
Salient Features of India constitution especially power and functionsSalient Features of India constitution especially power and functions
Salient Features of India constitution especially power and functions
KarakKing
 

Dernier (20)

Graduate Outcomes Presentation Slides - English
Graduate Outcomes Presentation Slides - EnglishGraduate Outcomes Presentation Slides - English
Graduate Outcomes Presentation Slides - English
 
On National Teacher Day, meet the 2024-25 Kenan Fellows
On National Teacher Day, meet the 2024-25 Kenan FellowsOn National Teacher Day, meet the 2024-25 Kenan Fellows
On National Teacher Day, meet the 2024-25 Kenan Fellows
 
Mehran University Newsletter Vol-X, Issue-I, 2024
Mehran University Newsletter Vol-X, Issue-I, 2024Mehran University Newsletter Vol-X, Issue-I, 2024
Mehran University Newsletter Vol-X, Issue-I, 2024
 
SOC 101 Demonstration of Learning Presentation
SOC 101 Demonstration of Learning PresentationSOC 101 Demonstration of Learning Presentation
SOC 101 Demonstration of Learning Presentation
 
Towards a code of practice for AI in AT.pptx
Towards a code of practice for AI in AT.pptxTowards a code of practice for AI in AT.pptx
Towards a code of practice for AI in AT.pptx
 
Salient Features of India constitution especially power and functions
Salient Features of India constitution especially power and functionsSalient Features of India constitution especially power and functions
Salient Features of India constitution especially power and functions
 
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx
 
This PowerPoint helps students to consider the concept of infinity.
This PowerPoint helps students to consider the concept of infinity.This PowerPoint helps students to consider the concept of infinity.
This PowerPoint helps students to consider the concept of infinity.
 
FSB Advising Checklist - Orientation 2024
FSB Advising Checklist - Orientation 2024FSB Advising Checklist - Orientation 2024
FSB Advising Checklist - Orientation 2024
 
Single or Multiple melodic lines structure
Single or Multiple melodic lines structureSingle or Multiple melodic lines structure
Single or Multiple melodic lines structure
 
TỔNG ÔN TẬP THI VÀO LỚP 10 MÔN TIẾNG ANH NĂM HỌC 2023 - 2024 CÓ ĐÁP ÁN (NGỮ Â...
TỔNG ÔN TẬP THI VÀO LỚP 10 MÔN TIẾNG ANH NĂM HỌC 2023 - 2024 CÓ ĐÁP ÁN (NGỮ Â...TỔNG ÔN TẬP THI VÀO LỚP 10 MÔN TIẾNG ANH NĂM HỌC 2023 - 2024 CÓ ĐÁP ÁN (NGỮ Â...
TỔNG ÔN TẬP THI VÀO LỚP 10 MÔN TIẾNG ANH NĂM HỌC 2023 - 2024 CÓ ĐÁP ÁN (NGỮ Â...
 
Spatium Project Simulation student brief
Spatium Project Simulation student briefSpatium Project Simulation student brief
Spatium Project Simulation student brief
 
How to Create and Manage Wizard in Odoo 17
How to Create and Manage Wizard in Odoo 17How to Create and Manage Wizard in Odoo 17
How to Create and Manage Wizard in Odoo 17
 
Wellbeing inclusion and digital dystopias.pptx
Wellbeing inclusion and digital dystopias.pptxWellbeing inclusion and digital dystopias.pptx
Wellbeing inclusion and digital dystopias.pptx
 
Fostering Friendships - Enhancing Social Bonds in the Classroom
Fostering Friendships - Enhancing Social Bonds  in the ClassroomFostering Friendships - Enhancing Social Bonds  in the Classroom
Fostering Friendships - Enhancing Social Bonds in the Classroom
 
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptx
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptxSKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptx
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptx
 
80 ĐỀ THI THỬ TUYỂN SINH TIẾNG ANH VÀO 10 SỞ GD – ĐT THÀNH PHỐ HỒ CHÍ MINH NĂ...
80 ĐỀ THI THỬ TUYỂN SINH TIẾNG ANH VÀO 10 SỞ GD – ĐT THÀNH PHỐ HỒ CHÍ MINH NĂ...80 ĐỀ THI THỬ TUYỂN SINH TIẾNG ANH VÀO 10 SỞ GD – ĐT THÀNH PHỐ HỒ CHÍ MINH NĂ...
80 ĐỀ THI THỬ TUYỂN SINH TIẾNG ANH VÀO 10 SỞ GD – ĐT THÀNH PHỐ HỒ CHÍ MINH NĂ...
 
UGC NET Paper 1 Mathematical Reasoning & Aptitude.pdf
UGC NET Paper 1 Mathematical Reasoning & Aptitude.pdfUGC NET Paper 1 Mathematical Reasoning & Aptitude.pdf
UGC NET Paper 1 Mathematical Reasoning & Aptitude.pdf
 
REMIFENTANIL: An Ultra short acting opioid.pptx
REMIFENTANIL: An Ultra short acting opioid.pptxREMIFENTANIL: An Ultra short acting opioid.pptx
REMIFENTANIL: An Ultra short acting opioid.pptx
 
Sociology 101 Demonstration of Learning Exhibit
Sociology 101 Demonstration of Learning ExhibitSociology 101 Demonstration of Learning Exhibit
Sociology 101 Demonstration of Learning Exhibit
 

The Power of Known Peers: A Study in Two Domains

  • 1. The Power of Known Peers: A Study in Two Domains Peter Brusilovsky with Danielle Lee and Sharon I-Han Hsiao
  • 2. Overview •  The context •  The problem •  The goal •  The system •  The study University of Pittsburgh - PAWS Lab 2
  • 4. Social Information Access •  Social Navigation –  Social support of user browsing •  Social Recommendation (Collaborative Filtering) –  Proactive information access •  Social Search –  Social support of search •  Social Visualization –  Social support for visualization-based access to information •  Social Bookmarking –  Access to bookmarked/shared information facilitated with tags
  • 5. Social Navigation: The Start •  Natural tendency of people to follow each other –  Making use of direct and indirect cues about the activities of others –  Following trails •  Footsteps in sand or snow •  Worn-out carpet –  Using dogears and annotations –  Giving direction or guidance •  Navigation driven by the actions from one or more advice providers
  • 6. • The pioneer idea of asynchronous indirect social navigation • Developed for collaborating writing and editing • Indicated read/edited places in a large document Edit Wear and Read Wear (1992)
  • 7. Social Information Access •  History-enriched environments –  Edit Wear and Read Wear (1992) –  Social navigation systems •  Footprints, Juggler, Kalas •  Collaborative filtering –  Manual push and pull •  Tapestry, LN Recommender –  Modern automatic CF recommender systemss •  Social Search –  Quest-based systems •  AntWorld –  Group-based search (i-Spy)
  • 8. From People to Crowds •  It started with people following other people –  ReadWear, Tapestry, AntWorld •  But we need to scale these ideas up! •  Let’s move from people to faceless crowds –  Follow-the-crowd social navigation –  Collaborative filtering –  Group-based on community-based social search 11/15/2011 University of Pittsburgh - PAWS Lab 8
  • 9. We Lost People in Crowds… •  Crowd-based approach does work, but there are issues •  Less trust to a faceless crowd •  Less motivation to follow •  Malicious users and attacks •  Should we step back? –  Start seeing people in crowds? 11/15/2011 University of Pittsburgh - PAWS Lab 9
  • 10. Brusilovsky, P., Chavan, G., Farzan, R., Social Adaptive Navigation Support for Open Corpus Electronic Textbooks, AH2004 10/19 Knowledge Sea: Social Navigation
  • 13. HOW TO IMPROVE RECOMMENDATIONS USING VARIOUS SOCIAL NETWORKS Exploring Watching Networks, Group Co-members and Research Collaborators as a source of Recommendation Danielle Lee
  • 14. Why to Use Online Social Networks? •  Connection in social networks are typically known to users •  Connected people have reasonably similar interests •  People tend to trust their connections more than faceless peers •  People are easily get influenced by those they know •  Address “Cold Start” problem •  Decrease the risk of misuse and attacks
  • 15. Paper Domain (Dataset) Trust-based Networks Avesani, et al (2005) Ski Resorts (Moleskiing.it) Al-Sharawneh & Williams (2010) General Items (Epinions) Jamali & Ester (2009) General Items (Epinions) Jamali & Ester (2011) General Items (Epinions) & Movies (Flixster) Ma, et al. (2008) General Items (Epinions) Massa & Avesani (2007) General Items (Epinions) Walter, et al. (2009) General Items (Epinions) DuBois, et al. (2009) Movies (FilmTrust) Golbeck & Hendler (2006) Movies (FilmTrust) Matsuo & Yamamoto (2007) Cosmetics (@cosme) Friendships Bonhard, et al. (2007) Movies (MovieMatch) Bourke, et al. (2011) Movies/TV(Facebook) Groh & Ehmig (2007) Local Clubs (A German Site) Liu and Lee (2010) Online Products (Cyworld) Pera & Ng (2011) Book (Amazon and LibraryThing) Sinha & Swearingen (2001) Books (Amazon, Sleeper & RatingZones) and Movies (Amazon, Reel.com, and MovieCritics) Konstas, et al. (2009) Music (Last.fm) Colleagues Guy, et al. (2009) Bookmarks of Web Pages (Lotus Connections) Group Member Yuan, et al., (2009) Music (Last.fm)
  • 16. Recommendations Based on Watching •  User-assigned unilateral connections based on their interests –  Highly object-centered relations and low personal familiarity –  Users concentrate on the usefulness of watched partners’ information collections. –  Meets the ‘Similarity Attraction theory’ and holds ‘transitive power’. –  Mimics the process of bookmarking interesting items. •  E.g. “following” on Twitter, “plus one” on Google, “watching” on Citeulike, “network” on Delicious and “contacts” on Flickr. •  This study is based on a Citeulike Data set provided by the system –  97,712 Users, 3,297,156 articles, 3,869,993 bookmarks and 44,847 watching relations –  The data set contains publications, the metadata (titles, author names, publication name, publication years, etc.), tags and users’ bookmarks
  • 17. Homophily in Watching Networks •  Users in watching relations have more common information items, metadata & tags than random pairs –  The similarity was the largest for direct connections and decreased with the increase of social distance between users. –  In particular, users connected by watching relations tend to co- bookmark the same items. –  The items shared by two users in direct watching relations are more rare and have similar contents and context. Co- bookmarks Jaccard Popularity Log- Likelihood Title Vector Author Name Vector Tag Vector Direct 1.80 0.21% 8.69 .204 .1440 .0149 .0505 1 Hop .39 0.04% 7.75 .097 .0814 .0033 .0168 2 Hops .16 0.02% 7.38 .061 .0626 .0020 .0114 No Relation .04 0.02% 6.92 .023 .0147 .0007 .0020
  • 18. Recommendations in Watching Networks •  Fusing watching relations with traditional collaborative filtering recommendations improves the quality
  • 19. Group-Based Link Homophily •  A group of people who are interested in the same topic places uses in a specific kind of social relationship that can be used for improving recommendations •  The homophily study based on a Citeulike Data set provided by the system: –  12,944 Users, 4,109 Groups and 18,793 Membership •  Information overlap between group co-members is significantly larger than the overlap between random pairs. Co- bookmark s Jaccard Popularity Log- Likelihoo d Title Vector Author Name Vector Tag Vector Group Co- Member s .26 1.01% 8.00 .050 .1117 .0222 .0595 No Relation .04 0.02% 6.92 .023 .0147 .0007 .0020
  • 20. Group-based Recommendations •  Matrix Factorization Recommendations based on Group library and Group Co-members’ library performed the best CF – Collaborative Filtering; Gmem – Group Comembers-based; Group – Comembers & Group-based .000 .005 .010 .015 .020 .025 .030 .035 .040 .045 .050 CF Gmem Group CF Gmem Group CF Gmem Group CF Gmem Group Jaccard Similarity Matrix Factorization Jaccard Similarity Matrix Factorization Top5 Top2_F1
  • 21. Group-based Recommendations for everyone? •  The idea of group-based recommendations is to pick candidate items from those that are not yet discovered by target users, but available in the group library and the co-members’ repositories. •  Therefore, users in the area A might not benefit from group-based recommendation.
  • 22. Group-based Recommendations •  Different Performance of Group-based recommendations depending to Users’ position. –  For the dictators who dominated their group activities, the recommendations based on group information didn’t perform well, compared with other user clusters. .000 .005 .010 .015 .020 .025 .030 .035 Top5_F1measure CF Gmem Group CF_SVD Gmem_SVD Group_SVD .00 .01 .02 .03 .04 .05 .06 .07 .08 Top2_F1measure CF Gmem Group CF_SVD Gmem_SVD Group_SVD
  • 23. Recommendations Based on Research Collaborators •  Users in research collaborations interact to each other personally and their relations are centered on their research topics and the relevant by-products. –  Online social networks for professionals is to implement offline referral chains on the Web. •  This study is based on Conference Navigator (current version 3; hence it is CN3, now), a social adaptive system to support conference attendees. –  464 users, 1000 conference talks of 15 conferences, 189 collaboration relations, 144 social connections on CN3, and 5,094 bookmarks –  Data set contains conference talks, the metadata (titles, author names, publication name, publication years, etc.), users’ bookmarks and users’ own publication records.
  • 24. Recommendations Based on Research Collaborators: Results •  Social Network-based Recommendations utilizing content information of objects were the most effective recommendation approach. .00 .05 .10 .15 .20 .25 .30 CF Community CFCW Profile SVD SN_Colleagues SN_CN3 SN_Both SN_SVD SNCW_Colleagues SNCW_CN3 SNCW_Both Baseline SN SNCW Top5_F1Measure .00 .05 .10 .15 .20 .25 .30 .35 .40 .45 CF Community CFCW Profile SVD SN_Colleagues SN_CN3 SN_Both SN_SVD SNCW_Colleagues SNCW_CN3 SNCW_Both Baseline SN SNCW Top2_F1Measure
  • 25. References •  Watching Relation-based Recommendations –  Lee, D. H. & Brusilovsky, P. (2011) Improving Recommendations using Watching Networks in a Social Tagging System, Proceedings of iConference 2011, Seattle, WA, USA, February 8 ~ 11, 2011 –  Lee, D. H. & Brusilovsky, P. (2010) Social Networks and Interest Similarity: The Case of CiteULike, Proceedings of the 21st ACM Conference on Hypertext and Hypermedia (Hypertext), Toronto Canada, June 14 ~ 16, 2010 •  Group-based Recommendations –  Lee, D. H. & Brusilovsky, P. (2010) Using Self-Defined Group Activities for Improving Recommendations in Collaborative Tagging Systems, Proceedings of the 3rd ACM Conference on Recommender Systems (Recsys), Barcelona, Spain, September 26 ~ 30, 2010 –  Lee, D. H., Brusilovsky, P. & Schleyer, T. (Under Review) Group-based Recommendations for Individual Members, Proceedings of the 21st ACM International Conference on Information and Knowledge Management (CIKM 2012), Maui, Hawaii,USA, October 29-November 2, 2012 •  Collaborator-based Recommendations –  Lee, D. H. & Brusilovsky, P. (Under Review) Exploring Social Approach to Recommend Talks at Research Conferences, Proceedings of the 8th IEEE International Conference on Collaborative Computing: Networking Applications and Worksharing (CollaborateCom 2012)
  • 26. HOW TO PROVIDE SOCIAL GUIDANCE TO LEARNING RESOURCES Who guides us better – a crowd or peers? Sharon I-Han Hsiao
  • 27. A Quest to Building a Social QuizGuide 11/15/2011 University of Pittsburgh - PAWS Lab 27
  • 28. 28 Good personalized guidance: improved problem solving success! The more the students compared to their peers, the higher post-quiz scores they received (r= 0.34 p=0.004) Parallel Introspective Views
  • 29. 29 • Pros: Liked OUM, interactivity with the content, social guidance • Cons: dense and complicated with increasing activities QuizMap
  • 30. 30 Progressor: • Higher Engagement: Increased the questions attempts and topic coverage • Increased problem solving success • Significant positive correlations between the frequencies of peer model sorting and question attempts and success rate, r= 0.75, p< .01; r= 0.76, p< .01. Progressor
  • 31. The Effect of Visible Peers QuizJET w/ IV Progressor Parameters n=18 n=30 Peers 6.83±2.25 !"#!$%"&%' Topics 4.00±0.79 ("))$%"&(' Questions 4.67±1.36 *"&&$%"%+' ,-./' 01' 2.-3.455' &6' ,-./'01'7894' +:' ,-./'01'-247' 6&' '
  • 32. • Adding additional collection did not sacrifice the usage • Increased the engagement (Quiz =: 5 hours, Example: 5 hours 20 mins) • Increased diversity helped increase problem solving success • Mix collections resulted in uniform performance Progressor+ 80.81 125.5 205.73 190.42 0 50 100 150 200 250 300 Non-adaptive Non-social ANS Progressor Progressor+ Attempts
  • 33. References •  Hsiao, I-H. and Brusilovsky, P. (2012) Motivational Social Visualizations for Personalized E-learning, In: Proceedings of 7th European Conference on Technology Enhanced Education (ECTEL), ECTEL 2012, Saarbrücken, Germany, September 18-21, 2012, Springer-Verlag, (to be appeared) •  Hsiao, I-H., Guerra, J., Parra, D., Bakalov, F., König-Ries, B., and Brusilovsky, P. (2012) Comparative Social Visualization for Personalized E-Learning. International Working Conference Advanced Visual Interfaces, AVI 2012, Capri, Italy, May 21-25, 2012, Proceeding AVI '12 Proceedings of the International Working Conference on Advanced Visual Interfaces, Pages: 303-307, ACM New York, NY, USA •  Bakalov, F., Hsiao, I-H., Brusilovsky, P., and König-Ries, B. (2011) Progressor: Personalized visual access to programming problems, IEEE Symposium on Visual Languages and Human-Centric Computing, September 18-22, 2011, Pittsburgh, PA, USA •  Hsiao, I-H., Bakalov, F., Brusilovsky, P., and König-Ries, B. (2011) Open Social Student Modeling: Visualizing Student Models with Parallel IntrospectiveViews. Proceedings of 19th International Conference on User Modeling, Adaptation, and Personalization (UMAP 2011), Girona, Spain, July 11-15, 2011, Springer, pp.171-182
  • 36. Conference Navigator III University of Pittsburgh - PAWS Lab 36 http://halley.exp.sis.pitt.edu/cn3/legacy.php?conferenceID=85