SlideShare a Scribd company logo
1 of 29
Download to read offline
Social Recommendation

     Yuan Quan (袁 泉)
   IBM Research - China
About me
• Yuan Quan
  – M.S. Computer Science and Engineering, Xi’an Jiaotong
    University, 2003-2006.
  – B.S. Computer Science and Engineering, Xi’an Jiaotong
    University, 1999-2003.
• 2006 ~ now IBM China Research Lab
• Research interest
  – Personalized recommendation
  – User modeling
  – Social network analysis
Agenda
• Social Recommendation
   – Categories & samples
   – Definition
• Concept-level Overview
• Effectiveness of Social Relationship
• Technologies on Social Fusion
   – Pair-wise similarity fusion
   – Graph-based fusion
       • Graph-based data models
       • Algorithms
Social Recommendation Categories

• Collaborative Filtering is a kind of social recommender
   – compare with traditional content-based approach
• Recommendation from friends
   – Offline: daily recommendation from friends
   – Online: news feeds from friends on Facebook, Re-tweet, 开心转帖
• Any recommendation using social data as input
   – Social relationship / social network
       • friendship, membership, trust/distrust, follow
   – Social tagging & bookmarking
• Recommendation over Social Media (Blog, YouTube)
Collaborative Filtering - Amazon
Friends’ Recommendation – Facebook
Social Recommendation based on massive
            people’s wisdom
Recommending Friends via Social Network
Music Recommendation based on Taste &
        Friendship/Membership
Agenda
• Social Recommendation
   – Categories & samples
   – Definition
• Concept-level Overview
• Effectiveness of Social Relationship
• Technologies on Social Fusion
   – Pair-wise similarity fusion
   – Graph-based fusion
       • 5 graph-based data models
       • Algorithms
           – Random walk
           – Class label propagation - adsorption
Social Recommendation Overview
         Input:                                         Output:

                                                        Information item
User-Item (Rating)             Algorithms
                                                        Merchandise/Ads
                      User/Item KNN; Clustering-based
Social Relations      Graph-based Algorithms            People
                      Matrix Factorization
 Social Tagging       Information Diffusion
                                                        Community
                      Probabilistic Model…




           Context:

                        Time      Location      Query
Effectiveness of Social Relationship

• CF vs SF                                                     Familiarity vs Similarity
•       Social filtering approach outperforms the              • Extensive user survey with 290 participants and a field study
                                                               including 90 users, indicates superiority of the familiarity network as
        CF approach in all variants of the
                                                               a basis for recommendations
        experiment                                             • Trustworthy




    G. Groh et.al, Recommendations in Taste Related Domains:      I.Guy, et.al Personalized Recommendation of Social Software Items
    Collaborative Filtering vs. Social Filtering, GROUP07         Based on Social Relations, ACM Recsys09
Agenda
• Social Recommendation
   – Categories & samples
   – Definition
• Concept-level Overview
• Effectiveness of Social Relationship
• Technologies on Social Fusion
   – Pair-wise similarity fusion
   – Graph-based fusion
       • 5 graph-based data models
       • Algorithms
           – Random walk
           – Class label propagation - adsorption
Fusing via weighted-similarity
                    friendship only
                             Item                                                   User
                       Ia      Ib      Ic                                     Ua      Ub      Uc
               Ua        1       0       1                            Ua        1       0       1
      User                                                   User
               Ub        0       1       0                            Ub        0       1       0
               Uc        1       1       0                            Uc        1       0       1
                  User-Item Matrix                                       Friendship Matrix



                       Simui                                                     Simfri

Neighborhood Similarity Formula:
                       Simui+fri(ua,ub) = λ *Simui(ua,ub) + (1-λ)*Simfri (ua,ub)
                              Optimal λ was learned by cross-validation
         Konstas, et, al. On social networks and collaborative recommendation, SIGIR09
         Yuan, et, al. Augmenting Collaborative Recommender by Fusing Explicit Social Relationships. ACM
         RecSys09, workshop of Social Recommender
Fusing via weighted-similarity
                  membership only
                        Item                                  Group
                   Ia      Ib   Ic                       Ga     Gb    Gc
             Ua     1       0    1                  Ua    0      0     1
      User                                   User
             Ub     0       1    0                  Ub    0      1     1
             Uc     1       1    0                  Uc    1      0     0
               User-Item Matrix                      Membership Matrix



                   Simui                                   Simmem

Neighborhood Similarity Formula:
                       Simui+mem(ua,ub) = λ *Simui(ua,ub) + (1-λ)*Simmem(ua,ub)
Fusing via weighted-similarity
              friendship + membership
                 Item                            User                           Group
            Ia     Ib   Ic                  Ua    Ub      Uc               Ga     Gb    Gc
       Ua    1      0       1          Ua    1        0    1          Ua    0      0     1
User                            User                           User
       Ub    0      1       0          Ub    0        1    0          Ub    0      1     1
       Uc    1      1       0          Uc    1        0    1          Uc    1      0     0
        User-Item Matrix                Friendship Matrix              Membership Matrix



                    Simui                    Simfri                         Simmem
 Neighborhood
     Similarity
      Formula:      Simui+fri+mem(ua,ub) = λSimui + (1-λ)[β Simmem + (1-
                    β)Simfri ]
                    Optimal λand β was learned by cross-validation
Experimental results cont.




•   The baseline is user-based CF on user-item matrix only by cosine similarity
Agenda
• Social Recommendation
   – Categories & samples
   – Definition
• Concept-level Overview
• Effectiveness of Social Relationship
• Technologies on Social Fusion
   – Pair-wise similarity fusion
   – Graph-based fusion
       • 5 graph-based data models
       • Algorithms
           – Random walk
           – Class label propagation - adsorption
Model 1: Classic user-item bipartite graph
             with attributes


       attributes     age      gender    loc




       item           i1       i2        i3




       user          u1         u2       u3


       attributes   category   color    price
Model 2: user-item bipartite graph with
                      social relationships
             user   item              Ga                          Ia
                                                     Ua
                     i1
               u1

                                      Gb                          Ib
                                                    Ub
friendship    u2          i2


                                      Gc                               Ic
                                                    Uc
              u3           i3


                                                U    user node
                      membership
                      friendship                I    item node
                      user’s behavior on item
                                                G    group node
Model 3: Triple models & Temporal models


             tag                 group




      user         item   user           item



      User-Item-Tag       User-Item-Group
Model 4: Temporal Models
• Information flow
   – u and r have 40 items in common
   – u and v have 40 items in common
                                               Session: a combinational
                                               node of user & item

                                                       session


      Fig.1 How adoption patterns affect the
      recommendations                           user             item



                                                 User-Item-Session



       Fig.2 illustration of Info. Flow
  X. Song et.al, Personalized
  Recommendation Driven by Information
  Flow, SIGIR 06
Model 5




 TrustWalker: RW on a trust network
M Jamali, TrustWalker: a random walk model for combining   A heterogeneous social network:
trust-based and item-based recommendation, SIGKDD09
                                                           User-Resource-Tag-Category

                                                           Zhang & Tang, Recommendation over a
                                                           Heterogeneous Social Network, WAIM08
Agenda
• Social Recommendation
   – Categories & samples
   – Definition
• Concept-level Overview
• Effectiveness of social relationship
• Technologies on fusing social relationships
   – Pair-wise similarity fusion
   – Graph-based fusion
       • 5 graph-based data models
       • Algorithms
           – Random walk
           – Class label propagation - adsorption
Random Walk
•   Random walk is a mathematical formalization of a trajectory that consists of taking successive
    random steps. Often, random walks are assumed to have Markov properties:


•   E.g. the path traced by a molecule as it travels in a liquid or a gas, the search path of a foraging
    animal, the financial status of a gambler can all be modeled as random walks




           One dimension RW
                                                                         Two dimension RW
Random Walk cont.
• RW on graph:                         PageRank is a random walk on graph




•   RW’s usage in recommendation
     – For each user, rank & recommend top-N unknown items
     – Calculate similarities between nodes
         • E.g. user-user nodes similarity for neighborhood
         • Similarity measures: Average Commute-Time, Average FPT, L+, etc.
•   Notice:
     – Transition probability matrix
     – Personalized vector
     – Damping factor
Class propagation - adsorption
                                                       Shadow
                                                       vertex
                                                  1




             1


                                                   1




Baluja, et.al, Video Suggestion and Discovery for YouTube: Taking Random Walks Through the View Graph, WWW08
Our work
•   Augmenting Collaborative Recommender by Fusing Explicit Social
    Relationships.
    – First work to discover membership as useful as friendship in
      recommendation.
        • ACM RecSys09, workshop of Social Recommender
•   Model Users’ Long-/short-term Preference on Graph for
    Recommendation.
    – First work to balance the influence of long-/short-term preference on
      graph
        • Submitted to SIGKDD10.
•   Temporal Dynamic of Social Trust for Recommendation
    – First work to study the temporal dynamics of social relations and its
      usage for recommendation
        • Draft for ACM Recsys10.
Thanks~!

More Related Content

What's hot

Recommender system introduction
Recommender system   introductionRecommender system   introduction
Recommender system introduction
Liang Xiang
 
Social Recommender Systems Tutorial - WWW 2011
Social Recommender Systems Tutorial - WWW 2011Social Recommender Systems Tutorial - WWW 2011
Social Recommender Systems Tutorial - WWW 2011
idoguy
 
Building a Recommendation Engine - An example of a product recommendation engine
Building a Recommendation Engine - An example of a product recommendation engineBuilding a Recommendation Engine - An example of a product recommendation engine
Building a Recommendation Engine - An example of a product recommendation engine
NYC Predictive Analytics
 

What's hot (20)

Social network analysis part ii
Social network analysis part iiSocial network analysis part ii
Social network analysis part ii
 
Community detection algorithms
Community detection algorithmsCommunity detection algorithms
Community detection algorithms
 
CS6010 Social Network Analysis Unit I
CS6010 Social Network Analysis Unit ICS6010 Social Network Analysis Unit I
CS6010 Social Network Analysis Unit I
 
Social Media Mining - Chapter 9 (Recommendation in Social Media)
Social Media Mining - Chapter 9 (Recommendation in Social Media)Social Media Mining - Chapter 9 (Recommendation in Social Media)
Social Media Mining - Chapter 9 (Recommendation in Social Media)
 
06 Community Detection
06 Community Detection06 Community Detection
06 Community Detection
 
Community Detection
Community Detection Community Detection
Community Detection
 
Recommender system introduction
Recommender system   introductionRecommender system   introduction
Recommender system introduction
 
Community Detection in Social Media
Community Detection in Social MediaCommunity Detection in Social Media
Community Detection in Social Media
 
Recommendation System Explained
Recommendation System ExplainedRecommendation System Explained
Recommendation System Explained
 
Diversity and novelty for recommendation system
Diversity and novelty for recommendation systemDiversity and novelty for recommendation system
Diversity and novelty for recommendation system
 
Recommendation system
Recommendation systemRecommendation system
Recommendation system
 
Social Media Mining - Chapter 3 (Network Measures)
Social Media Mining - Chapter 3 (Network Measures)Social Media Mining - Chapter 3 (Network Measures)
Social Media Mining - Chapter 3 (Network Measures)
 
Introduction to Recommendation Systems
Introduction to Recommendation SystemsIntroduction to Recommendation Systems
Introduction to Recommendation Systems
 
CS6010 Social Network Analysis Unit IV
CS6010 Social Network Analysis Unit IVCS6010 Social Network Analysis Unit IV
CS6010 Social Network Analysis Unit IV
 
Social Recommender Systems Tutorial - WWW 2011
Social Recommender Systems Tutorial - WWW 2011Social Recommender Systems Tutorial - WWW 2011
Social Recommender Systems Tutorial - WWW 2011
 
Cascading behavior in the networks
Cascading behavior in the networksCascading behavior in the networks
Cascading behavior in the networks
 
Building a Recommendation Engine - An example of a product recommendation engine
Building a Recommendation Engine - An example of a product recommendation engineBuilding a Recommendation Engine - An example of a product recommendation engine
Building a Recommendation Engine - An example of a product recommendation engine
 
Homophily and influence in social networks
Homophily and influence in social networksHomophily and influence in social networks
Homophily and influence in social networks
 
NE7012- SOCIAL NETWORK ANALYSIS
NE7012- SOCIAL NETWORK ANALYSISNE7012- SOCIAL NETWORK ANALYSIS
NE7012- SOCIAL NETWORK ANALYSIS
 
Tag based recommender system
Tag based recommender systemTag based recommender system
Tag based recommender system
 

Viewers also liked

Tag And Tag Based Recommender
Tag And Tag Based RecommenderTag And Tag Based Recommender
Tag And Tag Based Recommender
gu wendong
 
TurKit: A Toolkit for Human Computation Algorithms
TurKit: A Toolkit for Human Computation AlgorithmsTurKit: A Toolkit for Human Computation Algorithms
TurKit: A Toolkit for Human Computation Algorithms
Greg Little
 
CrowdSearcher. Reactive and multiplatform Crowdsourcing. keynote speech at DB...
CrowdSearcher. Reactive and multiplatform Crowdsourcing. keynote speech at DB...CrowdSearcher. Reactive and multiplatform Crowdsourcing. keynote speech at DB...
CrowdSearcher. Reactive and multiplatform Crowdsourcing. keynote speech at DB...
Search Computing
 
Social media recommendation based on people and tags (final)
Social media recommendation based on people and tags (final)Social media recommendation based on people and tags (final)
Social media recommendation based on people and tags (final)
es712
 
RSWEB 2013: A research platform for social recommendation
RSWEB 2013: A research platform for social recommendationRSWEB 2013: A research platform for social recommendation
RSWEB 2013: A research platform for social recommendation
Amit Sharma
 
It Only Takes a Minute
It Only Takes a MinuteIt Only Takes a Minute
It Only Takes a Minute
elliottofhook
 
Answering Search Queries with CrowdSearcher: a crowdsourcing and social netwo...
Answering Search Queries with CrowdSearcher: a crowdsourcing and social netwo...Answering Search Queries with CrowdSearcher: a crowdsourcing and social netwo...
Answering Search Queries with CrowdSearcher: a crowdsourcing and social netwo...
Marco Brambilla
 
Robert Rosenthal - Social Media & the 3Rs: Content Strategy Basics for Engagi...
Robert Rosenthal - Social Media & the 3Rs: Content Strategy Basics for Engagi...Robert Rosenthal - Social Media & the 3Rs: Content Strategy Basics for Engagi...
Robert Rosenthal - Social Media & the 3Rs: Content Strategy Basics for Engagi...
Social Media for Nonprofits
 
Introduction to the Social Dimension of Education (gamilla, vinson, sabelo)
Introduction to the Social Dimension of Education (gamilla, vinson, sabelo)Introduction to the Social Dimension of Education (gamilla, vinson, sabelo)
Introduction to the Social Dimension of Education (gamilla, vinson, sabelo)
Frezzy Vinson
 

Viewers also liked (20)

Trust and Recommender Systems
Trust and  Recommender SystemsTrust and  Recommender Systems
Trust and Recommender Systems
 
Tag And Tag Based Recommender
Tag And Tag Based RecommenderTag And Tag Based Recommender
Tag And Tag Based Recommender
 
Google Tech Talk on Social Recommendation
Google Tech Talk on Social RecommendationGoogle Tech Talk on Social Recommendation
Google Tech Talk on Social Recommendation
 
TurKit: A Toolkit for Human Computation Algorithms
TurKit: A Toolkit for Human Computation AlgorithmsTurKit: A Toolkit for Human Computation Algorithms
TurKit: A Toolkit for Human Computation Algorithms
 
Volunteer Anywhere
Volunteer AnywhereVolunteer Anywhere
Volunteer Anywhere
 
Fundchange and Koodonation Workshop Slides - Nov 23, 2011
Fundchange and Koodonation Workshop Slides - Nov 23, 2011Fundchange and Koodonation Workshop Slides - Nov 23, 2011
Fundchange and Koodonation Workshop Slides - Nov 23, 2011
 
CrowdSearcher. Reactive and multiplatform Crowdsourcing. keynote speech at DB...
CrowdSearcher. Reactive and multiplatform Crowdsourcing. keynote speech at DB...CrowdSearcher. Reactive and multiplatform Crowdsourcing. keynote speech at DB...
CrowdSearcher. Reactive and multiplatform Crowdsourcing. keynote speech at DB...
 
A Market In Your Social Network: The Effect of Extrinsic Rewards on Friendsou...
A Market In Your Social Network: The Effect of Extrinsic Rewards on Friendsou...A Market In Your Social Network: The Effect of Extrinsic Rewards on Friendsou...
A Market In Your Social Network: The Effect of Extrinsic Rewards on Friendsou...
 
Social media recommendation based on people and tags (final)
Social media recommendation based on people and tags (final)Social media recommendation based on people and tags (final)
Social media recommendation based on people and tags (final)
 
RSWEB 2013: A research platform for social recommendation
RSWEB 2013: A research platform for social recommendationRSWEB 2013: A research platform for social recommendation
RSWEB 2013: A research platform for social recommendation
 
It Only Takes a Minute
It Only Takes a MinuteIt Only Takes a Minute
It Only Takes a Minute
 
Answering Search Queries with CrowdSearcher: a crowdsourcing and social netwo...
Answering Search Queries with CrowdSearcher: a crowdsourcing and social netwo...Answering Search Queries with CrowdSearcher: a crowdsourcing and social netwo...
Answering Search Queries with CrowdSearcher: a crowdsourcing and social netwo...
 
Robert Rosenthal - Social Media & the 3Rs: Content Strategy Basics for Engagi...
Robert Rosenthal - Social Media & the 3Rs: Content Strategy Basics for Engagi...Robert Rosenthal - Social Media & the 3Rs: Content Strategy Basics for Engagi...
Robert Rosenthal - Social Media & the 3Rs: Content Strategy Basics for Engagi...
 
Fundchange Koodonation Social Media for Charities and Non-Profits
Fundchange Koodonation Social Media for Charities and Non-ProfitsFundchange Koodonation Social Media for Charities and Non-Profits
Fundchange Koodonation Social Media for Charities and Non-Profits
 
CSCW 2013 - Investigating the Appropriateness of Social Network Question Aski...
CSCW 2013 - Investigating the Appropriateness of Social Network Question Aski...CSCW 2013 - Investigating the Appropriateness of Social Network Question Aski...
CSCW 2013 - Investigating the Appropriateness of Social Network Question Aski...
 
Masters thesis defense talk
Masters thesis defense talkMasters thesis defense talk
Masters thesis defense talk
 
Introduction to the Social Dimension of Education (gamilla, vinson, sabelo)
Introduction to the Social Dimension of Education (gamilla, vinson, sabelo)Introduction to the Social Dimension of Education (gamilla, vinson, sabelo)
Introduction to the Social Dimension of Education (gamilla, vinson, sabelo)
 
Building Recommendation Systems on Social Data @KTH - FutureFriday - March 2014
Building Recommendation Systems on Social Data @KTH - FutureFriday - March 2014Building Recommendation Systems on Social Data @KTH - FutureFriday - March 2014
Building Recommendation Systems on Social Data @KTH - FutureFriday - March 2014
 
We Are Social's Guide To Building A Connected Strategy
We Are Social's Guide To Building A Connected StrategyWe Are Social's Guide To Building A Connected Strategy
We Are Social's Guide To Building A Connected Strategy
 
Crowd Agents: Interactive Crowd-Powered Systems in the Real World
Crowd Agents:  Interactive Crowd-Powered Systems in the Real WorldCrowd Agents:  Interactive Crowd-Powered Systems in the Real World
Crowd Agents: Interactive Crowd-Powered Systems in the Real World
 

Similar to Social Recommendation

Ronny lempelyahooindiabigthinkerapril2013
Ronny lempelyahooindiabigthinkerapril2013Ronny lempelyahooindiabigthinkerapril2013
Ronny lempelyahooindiabigthinkerapril2013
Muthusamy Chelliah
 
Recommender system algorithm and architecture
Recommender system algorithm and architectureRecommender system algorithm and architecture
Recommender system algorithm and architecture
Liang Xiang
 
Recommender Systems! @ASAI 2011
Recommender Systems! @ASAI 2011Recommender Systems! @ASAI 2011
Recommender Systems! @ASAI 2011
Ernesto Mislej
 
Random Walk by User Trust and Temporal Issues toward Sparsity Problem in Soci...
Random Walk by User Trust and Temporal Issues toward Sparsity Problem in Soci...Random Walk by User Trust and Temporal Issues toward Sparsity Problem in Soci...
Random Walk by User Trust and Temporal Issues toward Sparsity Problem in Soci...
Sc Huang
 
Digital Trails Dave King 1 5 10 Part 2 D3
Digital Trails   Dave King   1 5 10   Part 2   D3Digital Trails   Dave King   1 5 10   Part 2   D3
Digital Trails Dave King 1 5 10 Part 2 D3
Dave King
 

Similar to Social Recommendation (20)

Ronny lempelyahooindiabigthinkerapril2013
Ronny lempelyahooindiabigthinkerapril2013Ronny lempelyahooindiabigthinkerapril2013
Ronny lempelyahooindiabigthinkerapril2013
 
Enhance discovery Solr and Mahout
Enhance discovery Solr and MahoutEnhance discovery Solr and Mahout
Enhance discovery Solr and Mahout
 
Recommender system algorithm and architecture
Recommender system algorithm and architectureRecommender system algorithm and architecture
Recommender system algorithm and architecture
 
项亮 推荐系统实践 从入门到精通
项亮 推荐系统实践 从入门到精通 项亮 推荐系统实践 从入门到精通
项亮 推荐系统实践 从入门到精通
 
Managing Online Business Communities
Managing Online Business CommunitiesManaging Online Business Communities
Managing Online Business Communities
 
Rae
RaeRae
Rae
 
Recomendation system: Community Detection Based Recomendation System using Hy...
Recomendation system: Community Detection Based Recomendation System using Hy...Recomendation system: Community Detection Based Recomendation System using Hy...
Recomendation system: Community Detection Based Recomendation System using Hy...
 
Чираг Шах «Коллективный поиск, взаимодействие пользователей: подходы к изучен...
Чираг Шах «Коллективный поиск, взаимодействие пользователей: подходы к изучен...Чираг Шах «Коллективный поиск, взаимодействие пользователей: подходы к изучен...
Чираг Шах «Коллективный поиск, взаимодействие пользователей: подходы к изучен...
 
Social Computing Research
Social Computing ResearchSocial Computing Research
Social Computing Research
 
IntroductionRecommenderSystems_Petroni.pdf
IntroductionRecommenderSystems_Petroni.pdfIntroductionRecommenderSystems_Petroni.pdf
IntroductionRecommenderSystems_Petroni.pdf
 
Social Influence & Community Detection
Social Influence & Community DetectionSocial Influence & Community Detection
Social Influence & Community Detection
 
Overview of recommender system
Overview of recommender systemOverview of recommender system
Overview of recommender system
 
Recommender Systems! @ASAI 2011
Recommender Systems! @ASAI 2011Recommender Systems! @ASAI 2011
Recommender Systems! @ASAI 2011
 
Introduction to recommender systems
Introduction to recommender systemsIntroduction to recommender systems
Introduction to recommender systems
 
SRS presentation
SRS presentationSRS presentation
SRS presentation
 
Seams2016 presentation calikli_et_al
Seams2016 presentation calikli_et_alSeams2016 presentation calikli_et_al
Seams2016 presentation calikli_et_al
 
Tutorial on Relationship Mining In Online Social Networks
Tutorial on Relationship Mining In Online Social NetworksTutorial on Relationship Mining In Online Social Networks
Tutorial on Relationship Mining In Online Social Networks
 
Random Walk by User Trust and Temporal Issues toward Sparsity Problem in Soci...
Random Walk by User Trust and Temporal Issues toward Sparsity Problem in Soci...Random Walk by User Trust and Temporal Issues toward Sparsity Problem in Soci...
Random Walk by User Trust and Temporal Issues toward Sparsity Problem in Soci...
 
Lec7 collaborative filtering
Lec7 collaborative filteringLec7 collaborative filtering
Lec7 collaborative filtering
 
Digital Trails Dave King 1 5 10 Part 2 D3
Digital Trails   Dave King   1 5 10   Part 2   D3Digital Trails   Dave King   1 5 10   Part 2   D3
Digital Trails Dave King 1 5 10 Part 2 D3
 

More from gu wendong

宜信大数据金融云-CSDN
宜信大数据金融云-CSDN宜信大数据金融云-CSDN
宜信大数据金融云-CSDN
gu wendong
 
Pharos Social Map Based Recommendation For Content Centric Social Websites
Pharos Social Map Based Recommendation For Content Centric Social WebsitesPharos Social Map Based Recommendation For Content Centric Social Websites
Pharos Social Map Based Recommendation For Content Centric Social Websites
gu wendong
 
Resys China 创刊号
Resys China 创刊号Resys China 创刊号
Resys China 创刊号
gu wendong
 
孙超 - Recommendation Algorithm as a product
孙超 - Recommendation Algorithm as a product孙超 - Recommendation Algorithm as a product
孙超 - Recommendation Algorithm as a product
gu wendong
 
EdChang - Parallel Algorithms For Mining Large Scale Data
EdChang - Parallel Algorithms For Mining Large Scale DataEdChang - Parallel Algorithms For Mining Large Scale Data
EdChang - Parallel Algorithms For Mining Large Scale Data
gu wendong
 
王守崑 - 豆瓣在推荐领域的实践和思考
王守崑 - 豆瓣在推荐领域的实践和思考王守崑 - 豆瓣在推荐领域的实践和思考
王守崑 - 豆瓣在推荐领域的实践和思考
gu wendong
 
From Search To Discover by Wanght
From Search To Discover by WanghtFrom Search To Discover by Wanght
From Search To Discover by Wanght
gu wendong
 
Understanding Rbm by WangYuanTao
Understanding Rbm by WangYuanTaoUnderstanding Rbm by WangYuanTao
Understanding Rbm by WangYuanTao
gu wendong
 
Netflix Prize by Xlvector
Netflix Prize by XlvectorNetflix Prize by Xlvector
Netflix Prize by Xlvector
gu wendong
 

More from gu wendong (9)

宜信大数据金融云-CSDN
宜信大数据金融云-CSDN宜信大数据金融云-CSDN
宜信大数据金融云-CSDN
 
Pharos Social Map Based Recommendation For Content Centric Social Websites
Pharos Social Map Based Recommendation For Content Centric Social WebsitesPharos Social Map Based Recommendation For Content Centric Social Websites
Pharos Social Map Based Recommendation For Content Centric Social Websites
 
Resys China 创刊号
Resys China 创刊号Resys China 创刊号
Resys China 创刊号
 
孙超 - Recommendation Algorithm as a product
孙超 - Recommendation Algorithm as a product孙超 - Recommendation Algorithm as a product
孙超 - Recommendation Algorithm as a product
 
EdChang - Parallel Algorithms For Mining Large Scale Data
EdChang - Parallel Algorithms For Mining Large Scale DataEdChang - Parallel Algorithms For Mining Large Scale Data
EdChang - Parallel Algorithms For Mining Large Scale Data
 
王守崑 - 豆瓣在推荐领域的实践和思考
王守崑 - 豆瓣在推荐领域的实践和思考王守崑 - 豆瓣在推荐领域的实践和思考
王守崑 - 豆瓣在推荐领域的实践和思考
 
From Search To Discover by Wanght
From Search To Discover by WanghtFrom Search To Discover by Wanght
From Search To Discover by Wanght
 
Understanding Rbm by WangYuanTao
Understanding Rbm by WangYuanTaoUnderstanding Rbm by WangYuanTao
Understanding Rbm by WangYuanTao
 
Netflix Prize by Xlvector
Netflix Prize by XlvectorNetflix Prize by Xlvector
Netflix Prize by Xlvector
 

Recently uploaded

+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
?#DUbAI#??##{{(☎️+971_581248768%)**%*]'#abortion pills for sale in dubai@
 

Recently uploaded (20)

2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...
 
Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...
 
Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...Apidays New York 2024 - The value of a flexible API Management solution for O...
Apidays New York 2024 - The value of a flexible API Management solution for O...
 
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdfUnderstanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
 
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
 
TrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data DiscoveryTrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
 
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemkeProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processors
 
GenAI Risks & Security Meetup 01052024.pdf
GenAI Risks & Security Meetup 01052024.pdfGenAI Risks & Security Meetup 01052024.pdf
GenAI Risks & Security Meetup 01052024.pdf
 
A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century education
 
Real Time Object Detection Using Open CV
Real Time Object Detection Using Open CVReal Time Object Detection Using Open CV
Real Time Object Detection Using Open CV
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
What Are The Drone Anti-jamming Systems Technology?
What Are The Drone Anti-jamming Systems Technology?What Are The Drone Anti-jamming Systems Technology?
What Are The Drone Anti-jamming Systems Technology?
 
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
 
Automating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps ScriptAutomating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps Script
 
AWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of Terraform
 
Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024
 

Social Recommendation

  • 1. Social Recommendation Yuan Quan (袁 泉) IBM Research - China
  • 2. About me • Yuan Quan – M.S. Computer Science and Engineering, Xi’an Jiaotong University, 2003-2006. – B.S. Computer Science and Engineering, Xi’an Jiaotong University, 1999-2003. • 2006 ~ now IBM China Research Lab • Research interest – Personalized recommendation – User modeling – Social network analysis
  • 3. Agenda • Social Recommendation – Categories & samples – Definition • Concept-level Overview • Effectiveness of Social Relationship • Technologies on Social Fusion – Pair-wise similarity fusion – Graph-based fusion • Graph-based data models • Algorithms
  • 4. Social Recommendation Categories • Collaborative Filtering is a kind of social recommender – compare with traditional content-based approach • Recommendation from friends – Offline: daily recommendation from friends – Online: news feeds from friends on Facebook, Re-tweet, 开心转帖 • Any recommendation using social data as input – Social relationship / social network • friendship, membership, trust/distrust, follow – Social tagging & bookmarking • Recommendation over Social Media (Blog, YouTube)
  • 7. Social Recommendation based on massive people’s wisdom
  • 8. Recommending Friends via Social Network
  • 9. Music Recommendation based on Taste & Friendship/Membership
  • 10. Agenda • Social Recommendation – Categories & samples – Definition • Concept-level Overview • Effectiveness of Social Relationship • Technologies on Social Fusion – Pair-wise similarity fusion – Graph-based fusion • 5 graph-based data models • Algorithms – Random walk – Class label propagation - adsorption
  • 11. Social Recommendation Overview Input: Output: Information item User-Item (Rating) Algorithms Merchandise/Ads User/Item KNN; Clustering-based Social Relations Graph-based Algorithms People Matrix Factorization Social Tagging Information Diffusion Community Probabilistic Model… Context: Time Location Query
  • 12. Effectiveness of Social Relationship • CF vs SF Familiarity vs Similarity • Social filtering approach outperforms the • Extensive user survey with 290 participants and a field study including 90 users, indicates superiority of the familiarity network as CF approach in all variants of the a basis for recommendations experiment • Trustworthy G. Groh et.al, Recommendations in Taste Related Domains: I.Guy, et.al Personalized Recommendation of Social Software Items Collaborative Filtering vs. Social Filtering, GROUP07 Based on Social Relations, ACM Recsys09
  • 13. Agenda • Social Recommendation – Categories & samples – Definition • Concept-level Overview • Effectiveness of Social Relationship • Technologies on Social Fusion – Pair-wise similarity fusion – Graph-based fusion • 5 graph-based data models • Algorithms – Random walk – Class label propagation - adsorption
  • 14. Fusing via weighted-similarity friendship only Item User Ia Ib Ic Ua Ub Uc Ua 1 0 1 Ua 1 0 1 User User Ub 0 1 0 Ub 0 1 0 Uc 1 1 0 Uc 1 0 1 User-Item Matrix Friendship Matrix Simui Simfri Neighborhood Similarity Formula: Simui+fri(ua,ub) = λ *Simui(ua,ub) + (1-λ)*Simfri (ua,ub) Optimal λ was learned by cross-validation Konstas, et, al. On social networks and collaborative recommendation, SIGIR09 Yuan, et, al. Augmenting Collaborative Recommender by Fusing Explicit Social Relationships. ACM RecSys09, workshop of Social Recommender
  • 15. Fusing via weighted-similarity membership only Item Group Ia Ib Ic Ga Gb Gc Ua 1 0 1 Ua 0 0 1 User User Ub 0 1 0 Ub 0 1 1 Uc 1 1 0 Uc 1 0 0 User-Item Matrix Membership Matrix Simui Simmem Neighborhood Similarity Formula: Simui+mem(ua,ub) = λ *Simui(ua,ub) + (1-λ)*Simmem(ua,ub)
  • 16. Fusing via weighted-similarity friendship + membership Item User Group Ia Ib Ic Ua Ub Uc Ga Gb Gc Ua 1 0 1 Ua 1 0 1 Ua 0 0 1 User User User Ub 0 1 0 Ub 0 1 0 Ub 0 1 1 Uc 1 1 0 Uc 1 0 1 Uc 1 0 0 User-Item Matrix Friendship Matrix Membership Matrix Simui Simfri Simmem Neighborhood Similarity Formula: Simui+fri+mem(ua,ub) = λSimui + (1-λ)[β Simmem + (1- β)Simfri ] Optimal λand β was learned by cross-validation
  • 17. Experimental results cont. • The baseline is user-based CF on user-item matrix only by cosine similarity
  • 18. Agenda • Social Recommendation – Categories & samples – Definition • Concept-level Overview • Effectiveness of Social Relationship • Technologies on Social Fusion – Pair-wise similarity fusion – Graph-based fusion • 5 graph-based data models • Algorithms – Random walk – Class label propagation - adsorption
  • 19. Model 1: Classic user-item bipartite graph with attributes attributes age gender loc item i1 i2 i3 user u1 u2 u3 attributes category color price
  • 20. Model 2: user-item bipartite graph with social relationships user item Ga Ia Ua i1 u1 Gb Ib Ub friendship u2 i2 Gc Ic Uc u3 i3 U user node membership friendship I item node user’s behavior on item G group node
  • 21. Model 3: Triple models & Temporal models tag group user item user item User-Item-Tag User-Item-Group
  • 22. Model 4: Temporal Models • Information flow – u and r have 40 items in common – u and v have 40 items in common Session: a combinational node of user & item session Fig.1 How adoption patterns affect the recommendations user item User-Item-Session Fig.2 illustration of Info. Flow X. Song et.al, Personalized Recommendation Driven by Information Flow, SIGIR 06
  • 23. Model 5 TrustWalker: RW on a trust network M Jamali, TrustWalker: a random walk model for combining A heterogeneous social network: trust-based and item-based recommendation, SIGKDD09 User-Resource-Tag-Category Zhang & Tang, Recommendation over a Heterogeneous Social Network, WAIM08
  • 24. Agenda • Social Recommendation – Categories & samples – Definition • Concept-level Overview • Effectiveness of social relationship • Technologies on fusing social relationships – Pair-wise similarity fusion – Graph-based fusion • 5 graph-based data models • Algorithms – Random walk – Class label propagation - adsorption
  • 25. Random Walk • Random walk is a mathematical formalization of a trajectory that consists of taking successive random steps. Often, random walks are assumed to have Markov properties: • E.g. the path traced by a molecule as it travels in a liquid or a gas, the search path of a foraging animal, the financial status of a gambler can all be modeled as random walks One dimension RW Two dimension RW
  • 26. Random Walk cont. • RW on graph: PageRank is a random walk on graph • RW’s usage in recommendation – For each user, rank & recommend top-N unknown items – Calculate similarities between nodes • E.g. user-user nodes similarity for neighborhood • Similarity measures: Average Commute-Time, Average FPT, L+, etc. • Notice: – Transition probability matrix – Personalized vector – Damping factor
  • 27. Class propagation - adsorption Shadow vertex 1 1 1 Baluja, et.al, Video Suggestion and Discovery for YouTube: Taking Random Walks Through the View Graph, WWW08
  • 28. Our work • Augmenting Collaborative Recommender by Fusing Explicit Social Relationships. – First work to discover membership as useful as friendship in recommendation. • ACM RecSys09, workshop of Social Recommender • Model Users’ Long-/short-term Preference on Graph for Recommendation. – First work to balance the influence of long-/short-term preference on graph • Submitted to SIGKDD10. • Temporal Dynamic of Social Trust for Recommendation – First work to study the temporal dynamics of social relations and its usage for recommendation • Draft for ACM Recsys10.