1 Whats Up: P2P news recommender Anne-Marie Kermarrec Joint work with Antoine Boutet, Davide Frey (INRIA) and Rachid Guerraoui (EPFL) Gossple workshop.

Slides:



Advertisements
Similar presentations
10/22/20101 Westar Formula Rate and your bill. 10/22/20102 What we will be going over Transferring the results from the Westar Formula Rate to the SPP.
Advertisements

PDAs/Smart Phones and Medical Records in Health Care Mary Z. Mays, PhD Associate Dean and Professor Innovation Institute for Health Professions.
Castellana Grotte 12 aprile Alternating training in Italy Dott. Ruggiero Francavilla Dirigente Ufficio III – USR Puglia.
Information Security of Embedded Systems : Design of Secure Systems Prof. Dr. Holger Schlingloff Institut für Informatik und Fraunhofer FIRST.
Summary Overview of Vireo Student Submission of ETDs
COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED. INFORM: a dynamic INterest FORwarding Mechanism for Information Centric Networking Raffaele Chiocchetti,
Defending against large-scale crawls in online social networks Mainack Mondal Bimal Viswanath Allen Clement Peter Druschel Krishna Gummadi Alan Mislove.
Presentation at IEEE AWSITC, June 4, Energy-Efficient Communications via Network Coding Jos Weber Delft University of Technology The Netherlands.
Enterprise Social Networking Tool Comparison October 2010.
25 Novembre 2010 Safe Maintenance through design From machine engineering for safe maintenance to maintenance engineering for safe machine Workshop Maintenance.
LIFE external communications team AEIDL 1 LIFE+ communication Presentation at LIFE+ Kick-off meeting.
Open Days 2010 D. Gubbels Professionalization within the range of volunteer work New challenges for volunteering organizations - Ehrenamt professionalisieren!
Guiding Principle Five Mentoring needs to be tailored to the needs of the individual teacher and, at the same time, verifying their skills as teachers.
January 12, 2010 Updated February 4, Starting in TEA will collect Teacher Class Assignments and Student Course Completion data at the.
January 12, 2010 Updated April 9, Starting in TEA will collect Teacher Class Assignments and Student Course Completion data at the classroom.
Shibboleth Development and Support Services SAML Protected Resources The theory and practice of granularity and management data Ed Dee EDINA.
NJJN JUNE Fulfilling the Promise of Juvenile Justice by Engaging Crime Victims & Survivors and Those Who Serve Them Presented by: ANNE SEYMOUR
Kevin M. Johnston MSU TAP Model Effective Discussion Techniques Engage one another concerning your most pressing questions concerning preparing.
® Microsoft Office 2010 Excel Tutorial 3: Working with Formulas and Functions.
Using A series of training presentations How to list your project September,
HOW MEDIA CONSUMPTION HAS CHANGED SINCE 2000 News is pervasive, portable, personalized, participatory – and a social experience Lee Rainie Director – Pew.
An Analysis of the P2P Traffic Characteristics on File Transfers Between Prefectures and Between Autonomous Systems in the Winny Network Nov. 1,
South Carolina School HACCP
® Microsoft Office 2010 Managing Your Files. XP Objectives Develop file management strategies Explore files, folders, and libraries Create, name, copy,
Operations Management
4/6/20100Office/Department || Understanding Academic Probation & Academic Rules and Regulations Presented by the Academic Advisement Center UNVH
August 4, The following PEIMS reporting changes have been made to the PEIMS Collection in order to collect the Classroom Link information.
Chapter 14 – Resource Planning
Chapter 9– Capacity Planning & Facility Location
Linked Lists in C and C++ CS-2303, C-Term Linked Lists in C and C++ CS-2303 System Programming Concepts (Slides include materials from The C Programming.
A useful testing technique and more…
Hash Tables and Constant Access Time CS-2303, C-Term Hash Tables and Constant Access Time CS-2303 System Programming Concepts (Slides include materials.
Multichannel publishing of statistics (electronic publications and database) - Finnish experience Seminar on dissemination of statistics and launching.
Tutorial 5: Working with Excel Tables, PivotTables, and PivotCharts
Preparing to Automate Data Management
Tutorial 1 Creating a Database
ACOT Intro/Copyright Succeeding in Business with Microsoft Excel 2010: Chapter1.
Anonymizing User Location and Profile Information for Privacy-aware Mobile Services Masanori Mano, Yoshiharu Ishikawa Nagoya University 11/2/
Tutorial 3: Communicating Project Information
Muntaha Gharaibeh RN PhD Associate Professor and Dean of Nursing Director of WHOCC for Human Resource Development in Nursing Faculty of Nursing Jordan.
® Microsoft Office 2010 Excel Tutorial 1: Getting Started with Excel.
Text Categorization.
LECTURE 18, NOVEMBER 2, 2010 ASTR 101, SECTION 2 INSTRUCTOR, JACK BRANDT 1ASTR 101-3, FALL 2010.
LECTURE 15, OCTOBER 21, 2010 ASTR 101, SECTION 3 INSTRUCTOR, JACK BRANDT 1ASTR 101-3, FALL 2010.
© Wiley Chapter 1 - Introduction to Operations Management Operations Management by R. Dan Reid & Nada R. Sanders 4th Edition © Wiley 2010.
Tutorial on KMIP and FCEAP/GPSK
The Digital Library Reference Model: Functionality Domain Carlo Meghini CNR-ISTI DL.org Autumn School, Athens, 3-8 October 2010.
Wyoming Department of Education Special Programs Division
Open Source – Open Knowledge: some stories from Cambodia by Norbert University of Phnom Penh, Department of Media and Communication 8 January.
® Microsoft Office 2010 Browser and Basics.
Student Learning Outcome Assessment Plan Backward Design with the ending in mind SLOAC Thinking on Paper 11/29/2010 1PRIE Draft.
Collaboration Works! 10/20/20101 Planning Research Institutional Effectiveness.
Quick Training Guide New SpringerLink, August 2010.
CHAPTER 7. Chapter 7Mugan-Akman Current assets assets that are expected to be converted into cash within one year or within the operating cycle.
Jan SedmidubskyOctober 28, 2011Scalability and Robustness in a Self-organizing Retrieval System Jan Sedmidubsky Vlastislav Dohnal Pavel Zezula On Investigating.
Chapter 13 – Aggregate Planning
[Networking Hardwares] [Maninder Kaur]
The Vocal Pedagogy Workshop 2011 Vocal Registers: Stephen F. Austin, M.M., Ph.D. Associate Professor of Voice University of North Texas 9/30/20101.
Importance of Modeling & Simulation Throughout In-service Lifecycle Phase Leigh Jarman Senior Reliability Engineer.
Tutorial 8 Sharing, Integrating, and Analyzing Data
Amit Goyal Laks V. S. Lakshmanan RecMax: Exploiting Recommender Systems for Fun and Profit University of British Columbia
Erik Amerikaner Oak Park High School Oak Park, California
August VLSI Memory Design Shmuel Wimer Bar Ilan University, School of Engineering.
SocioPlug Polystyrene: Survivable Shape for Self-Organising Data François Taïani Joint work with: Hoel Kervadec (INSA Rennes) Simon Bouget (ENS Rennes)
A P2P REcommender system based on Gossip Overlays (PREGO) ‏ R.Baraglia, P.Dazzi M.Mordacchini, L.Ricci A P2P REcommender system based on Gossip Overlays.
Democratizing personalization Anne-Marie Kermarrec Joint work with A. Boutet, D. Frey, R. Guerraoui, A. Jégou, H. Ribeiro.
Yusuf Simonson Title Suggesting Friends Using the Implicit Social Graph.
Personalized Ontologies for Web Search and Caching Susan Gauch Information and Telecommunications Technology Center Electrical Engineering and Computer.
Gil EinzigerRoy Friedman Computer Science Department Technion.
Recommender Systems Copyright: Dietmar Jannah, Markus Zanker and Gerhard Friedrich (slides based on their IJCAI talk „Tutorial: Recommender Systems”)
Presentation transcript:

1 Whats Up: P2P news recommender Anne-Marie Kermarrec Joint work with Antoine Boutet, Davide Frey (INRIA) and Rachid Guerraoui (EPFL) Gossple workshop 2010

2 The social Web Web content is generated by you, me, your friends and millions of others The Web has turned social

Content comes from everywhere Gossple workshop 20103

Is it equally relevant? Gossple workshop 20104

Is it equally relevant? Gossple workshop 20105

Is it equally relevant? Gossple workshop 20106

Whats wrong with news feed? Amazon recommends me a fryer Some of my Facebook write in Italian LeMonde.fr wants to inform me on the Champions ligue Gossple workshop 20107

Why is it so difficult? Even a space restricted to users explicit subscriptions is too large a database Dynamic Recommendations not always user-centric Explicit links not always that relevant Classical pub/sub do not filter enough Granularity of a user seems too coarse Gossple workshop 20108

Cascading over explicit links Gossple workshop 20109

Fine grain tuning calls for decentralisation Gossple workshop

Whats up Decentralised information dissemination channel Simple interface: I like it or I dont Exploit implicit social links Gossple workshop

Gossple workshop

Whats up in a nutshell Gossple workshop

Whats up challenges Gossple workshop Who are my social acquaintances How to discover them? How to disseminate news ? Similarity metric Through gossip Biased epidemic protocol

Whats up: Gossple net Gossple workshop

Whats up challenges Gossple workshop Who are my social acquaintances How to discover them? How to disseminate news ? Similarity metric

An implicit social network Gossple workshop

Which nodes should be considered as social acquaintances? Model U(sers) × I(tems) (news) Profile(u) = vector of liked news Minimal information Similarity metrics Overlap Cosine similarity Multi-interest similarity Gossple workshop

Item cosine similarity 19Gossple workshop 2010 Normalized overlap Profile(u)= Vector of news Items{u}

Individual rating might be too restrictive Rate the set of users instead of individuals 20 Items of interest for nodes in set(n) Items of interest for nodes in set(n) Distribution Normalized not to take into account non shared interests Normalized not to take into account non shared interests Gossple workshop 2010

Whats up challenges Gossple workshop Who are my social acquaintances How to discover them? How to disseminate news ? Through gossip

22 The Gossple network Gossple workshop 2010 Copyright: E. Rivière Gossip similarity protocol. Gossip-based peer sampling service

Gossple social network Gossple workshop port :2110 Bloom Filter Update :2020 Profile Update time5 Friends Uniform sample c entries k entries

Gossple workshop Building the social network Two gossip protocols Similarity-based Peer Sampling Random Peer Sampling When p encounters q Evaluate potential new view, based on set similarity metric Use of Bloom filters to limit the communication overhead RPS SPS RPS SPS

Whats up in a nutshell Gossple workshop

Whats up challenges Gossple workshop Who are my social acquaintances How to discover them? How to disseminate news ? Biased epidemic protocol

Dissemination Gossple workshop Heterogeneous Homogeneous HeterogeneousHomogeneous Involvement (fanout) Expectations Epidemic Dissemination F=log(N) Heterogeneous Gossip F log(N) on average

BEEP: orientation and amplification Orientation: to whom? Gossple workshop Forward to friends Forward to random Amplification: to how many? Increase fanout Decrease fanout

Beep: I like it Gossple workshop I like it!

Beep: I dont Gossple workshop I dislike it!

Tuning BEEP Orientation The news carries the list of visited users A profile: sum of interests of users who liked it Amplification F log(N) friends Amplification depends on the similarity between the news and the user F 1 or 2 random Gossple workshop

Evaluation User Metrics Spam Recall Precision System metric Number of messages Redundancy (useless messages) Traces Synthetic clustered traces Real dataset: 700 Digg users/2000 news/1 week Gossple workshop

Preliminary results Gossple workshop AlgorithmPrecisionRecallSpam Perfect110 Gossip fanout=log(n)= Cascading through explicit friends from Digg WhatsUp fanout=11/1 ; ttl= WhatsUp without no social users

To take away Automatic light news recommender Analysis through mean field theory Experimental evaluation Next: diversity of sources, trust, privacy Gossple workshop

Thank you 35 Gossple workshop 2010

36