This document provides an overview of trust in recommender systems and discusses their historical development and recent advances. It begins with an introduction to information overload and the differences between information retrieval and information filtering. It then describes content-based filtering techniques and their weaknesses, particularly for recommending non-text items like movies and songs that require time-consuming human tagging. The document also discusses collaborative filtering recommender systems and the need for trust-aware filtering that considers users' trust in other users and in the recommendations.
99 scholars - increase your online research visibility99Scholars
This is the powerpoint of the talk given in the Sydney University Inspire Postgraduate event. In this talk I have briefly describe how to use social media (Google Scholar, Academia, Reseasrchgate, LinkedIn, Twitter, Quora etc.) to increase researchers' online presence. I have also explain a few tips on how to do search engine optimization for researchers' academic website so that they can be ranked higher for their research keywords.
Startupfest 2012 - Coefficients of frictionStartupfest
It must have been amazing to live when the steam engine was invented. For millennia, human enterprise has tried to do one thing: overcome the friction of the physical world. From the first wheel and the earliest lever, to the structure of representative government and the design of broadcast TV, we’ve been fighting friction since we crawled out of the primordial ooze. That steam engine promised spare muscle, a beast of burden than never complained. Machinery would set us free. As it turned out, we were wrong. The answer wasn’t a better way to overcome friction—it was a move to the near-frictionless world of electrons. Today, every edifice we’ve erected to fight friction is crumbling in the face of a frictionless future. Join Alistair Croll for a wild romp through the economics of abundance, augmented humanity, home manufacturing, firing before aiming, coal supplies, education, and more, and see why there is simply no better time in human history to be a disruptor.
Building Trust in a Tense Information Society, Daniel Weitzner, Director, MIT...MIT Startup Exchange
Building Trust in a Tense Information Society, Daniel Weitzner, Director, MIT CSAIL Decentralized Information Group. Keynote held at MIT Startup Exchange (STEX) Cybersecurity Innovation workshop (5/28) at MIT on Thursday May 28, 2015, 8:30 AM to 11:30 AM, at One Main Street, Cambridge, MA, USA.
SmartPLS is a software application for (graphical) path modeling with latent variables (LVP). The partial least squares (PLS)-method is used for the LVP-analysis in this software.
Research with Partial Least Square (PLS) based Structural Equation Modelling ...Tuhin AI Advisory
A STRUCTURAL MODELING APPROACH TO COMPREHEND PURCHASE INTENTION INFLUENCED BY SOCIAL MEDIA : THE MEDIATING ROLE OF CONSUMER ATTITUDE AND THE MODERATING ROLE OF MARKET MAVENS
Introduction to Structural Equation Modeling Partial Least Sqaures (SEM-PLS)Ali Asgari
Partial least squares structural equation modelling (PLS-SEM) has recently received considerable attention in a variety of disciplines.The goal of PLS-SEM is the explanation of variances (prediction-oriented approach of the methodology) rather than explaining covariances (theory testing via covariance-based SEM).
99 scholars - increase your online research visibility99Scholars
This is the powerpoint of the talk given in the Sydney University Inspire Postgraduate event. In this talk I have briefly describe how to use social media (Google Scholar, Academia, Reseasrchgate, LinkedIn, Twitter, Quora etc.) to increase researchers' online presence. I have also explain a few tips on how to do search engine optimization for researchers' academic website so that they can be ranked higher for their research keywords.
Startupfest 2012 - Coefficients of frictionStartupfest
It must have been amazing to live when the steam engine was invented. For millennia, human enterprise has tried to do one thing: overcome the friction of the physical world. From the first wheel and the earliest lever, to the structure of representative government and the design of broadcast TV, we’ve been fighting friction since we crawled out of the primordial ooze. That steam engine promised spare muscle, a beast of burden than never complained. Machinery would set us free. As it turned out, we were wrong. The answer wasn’t a better way to overcome friction—it was a move to the near-frictionless world of electrons. Today, every edifice we’ve erected to fight friction is crumbling in the face of a frictionless future. Join Alistair Croll for a wild romp through the economics of abundance, augmented humanity, home manufacturing, firing before aiming, coal supplies, education, and more, and see why there is simply no better time in human history to be a disruptor.
Building Trust in a Tense Information Society, Daniel Weitzner, Director, MIT...MIT Startup Exchange
Building Trust in a Tense Information Society, Daniel Weitzner, Director, MIT CSAIL Decentralized Information Group. Keynote held at MIT Startup Exchange (STEX) Cybersecurity Innovation workshop (5/28) at MIT on Thursday May 28, 2015, 8:30 AM to 11:30 AM, at One Main Street, Cambridge, MA, USA.
SmartPLS is a software application for (graphical) path modeling with latent variables (LVP). The partial least squares (PLS)-method is used for the LVP-analysis in this software.
Research with Partial Least Square (PLS) based Structural Equation Modelling ...Tuhin AI Advisory
A STRUCTURAL MODELING APPROACH TO COMPREHEND PURCHASE INTENTION INFLUENCED BY SOCIAL MEDIA : THE MEDIATING ROLE OF CONSUMER ATTITUDE AND THE MODERATING ROLE OF MARKET MAVENS
Introduction to Structural Equation Modeling Partial Least Sqaures (SEM-PLS)Ali Asgari
Partial least squares structural equation modelling (PLS-SEM) has recently received considerable attention in a variety of disciplines.The goal of PLS-SEM is the explanation of variances (prediction-oriented approach of the methodology) rather than explaining covariances (theory testing via covariance-based SEM).
Santa Fe Complex
March 13, 2009
Martin Klein, Frank McCown,
Joan Smith, Michael L. Nelson
Department of Computer Science
Old Dominion University
Norfolk VA
API's, Freebase, and the Collaborative Semantic webDan Delany
A presentation about the state of the collaborative semantic web, including:
- What?
- Why?
- Where do we stand?
- A case study on Metaweb's Freebase project
Interest in Neural networks is growing with many areas from image recognition to speech processing reporting impressive results. Applications in Natural language processing with Neural networks have found multiple applications. With advances in software and hardware technologies, and interest in AI based applications growing, it is time to understand neural networks applied to natural language processing better!
In this workshop, we will discuss the basics of neural networks and natural language processing and discuss how neural approaches differ from traditional natural language modeling techniques with practical applications.
Crisis or Opportunity? Cataloging, Catalogers, RDA, and ChangeDiane Hillmann
What we need to change, what's changing us, and what we can do about it. Presented to members of the Five Colleges consortium in Western Massachusetts on May 1, 2009.
A Family That Hacks Together, Interacts Together!Daniel Davis
Everyone's heard the old saying "A family thats plays together, stays together", right? Well this session is going to kick that up a notch and talk about getting the family to "hack" together. It starts out discussing the different forms of hacking a family might embrace. Along the way it highlights what supplies & equipment you might need, as well as suggesting the overall costs that one might expect to incur. It then takes a brief focus on safety and proper use of that equipment. The session finishes off with some tips and tricks that may help with both the hacks themselves, as well as getting the family to actually interact by joining in on the fun.
Moved to https://slidr.io/azzazzel/web-application-performance-tuning-beyond-xmxMilen Dyankov
This slide deck will be removed from here in the future. It has been moved to : https://slidr.io/azzazzel/web-application-performance-tuning-beyond-xmx
Manypedia: Comparing Language Points of View of Wikipedia CommunitiesPaolo Massa
Manypedia is at http://www.manypedia.com
These slides have been presented by Paolo Massa at WikiSym, 8th International Symposium on Wikis and Open Collaboration, 29 August 2013, Linz, Austria.
Manypedia is joint work of Paolo Massa and Federico Scrinzi (and it is open source too!)
The paper is at http://www.gnuband.org/papers/manypedia-comparing-language-points-of-view-of-wikipedia-communities/
If you like Manypedia and you have a chance, don't forget to cite our paper, thanks!
More Related Content
Similar to Trust in Recommender Systems
a historical overview and recent developments
Santa Fe Complex
March 13, 2009
Martin Klein, Frank McCown,
Joan Smith, Michael L. Nelson
Department of Computer Science
Old Dominion University
Norfolk VA
API's, Freebase, and the Collaborative Semantic webDan Delany
A presentation about the state of the collaborative semantic web, including:
- What?
- Why?
- Where do we stand?
- A case study on Metaweb's Freebase project
Interest in Neural networks is growing with many areas from image recognition to speech processing reporting impressive results. Applications in Natural language processing with Neural networks have found multiple applications. With advances in software and hardware technologies, and interest in AI based applications growing, it is time to understand neural networks applied to natural language processing better!
In this workshop, we will discuss the basics of neural networks and natural language processing and discuss how neural approaches differ from traditional natural language modeling techniques with practical applications.
Crisis or Opportunity? Cataloging, Catalogers, RDA, and ChangeDiane Hillmann
What we need to change, what's changing us, and what we can do about it. Presented to members of the Five Colleges consortium in Western Massachusetts on May 1, 2009.
A Family That Hacks Together, Interacts Together!Daniel Davis
Everyone's heard the old saying "A family thats plays together, stays together", right? Well this session is going to kick that up a notch and talk about getting the family to "hack" together. It starts out discussing the different forms of hacking a family might embrace. Along the way it highlights what supplies & equipment you might need, as well as suggesting the overall costs that one might expect to incur. It then takes a brief focus on safety and proper use of that equipment. The session finishes off with some tips and tricks that may help with both the hacks themselves, as well as getting the family to actually interact by joining in on the fun.
Moved to https://slidr.io/azzazzel/web-application-performance-tuning-beyond-xmxMilen Dyankov
This slide deck will be removed from here in the future. It has been moved to : https://slidr.io/azzazzel/web-application-performance-tuning-beyond-xmx
Similar to Trust in Recommender Systems
a historical overview and recent developments (20)
Manypedia: Comparing Language Points of View of Wikipedia CommunitiesPaolo Massa
Manypedia is at http://www.manypedia.com
These slides have been presented by Paolo Massa at WikiSym, 8th International Symposium on Wikis and Open Collaboration, 29 August 2013, Linz, Austria.
Manypedia is joint work of Paolo Massa and Federico Scrinzi (and it is open source too!)
The paper is at http://www.gnuband.org/papers/manypedia-comparing-language-points-of-view-of-wikipedia-communities/
If you like Manypedia and you have a chance, don't forget to cite our paper, thanks!
Presentazione di Paolo Massa nell'ambito del Seminario residenziale “L’approccio territoriale tra aiuto e crescita” - 22-23 giugno 2012 - Villa Flangini - Asolo - Organizzato dal SerAT (Servizio Alcologia e Tabagismo Ulss 8)
Con il contributo di ACAT-ULSS 8 onlus e Cooperativa Sonda. Con il patrocinio di Alcologia Ecologica
DESIGN PRINCIPLES OF WIKIS AND THEIR IMPACT ON KNOWLEDGE EXCHANGE PROCESSES Paolo Massa
DESIGN PRINCIPLES OF WIKIS AND THEIR IMPACT ON KNOWLEDGE EXCHANGE PROCESSES
From Analyzing Wiki-based Networks to Improve Knowledge Processes in Organizations by Claudia Müller, Benedikt Meuthrath, Anne Baumgraß Slides by Paolo Massa
Collective Memory building in Wikipedia: the case of North African uprisingsPaolo Massa
Paper presented at Wikisym 2011, 7th International Symposium on Wikis and Open Collaboration
Read the paper at http://www.gnuband.org/papers/collective_memory_building_in_wikipedia_the_case_of_north_african_uprisings/
Authors: Michela Ferron, Paolo Massa
Abstract:
Since December 2010, a series of protests and uprisings have shocked North African countries such as Tunisia, Egypt, Libya, Syria, Yemen and more. In this paper, focusing mainly on the Egyptian revolution, we provide evidence of the intense edit activity occurred during these uprisings on the related Wikipedia
pages. Thousands of people provided their contribution on the content pages and discussed improvements and disagreements on the associated talk pages as the traumatic events unfolded. We
propose to interpret this phenomenon as a process of collective memory building and argue how on Wikipedia this can be studied empirically and quantitatively in real time. We explore and suggest possible directions for future research on collective memory formation of traumatic and controversial events in Wikipedia.
Social networks of Wikipedia - Paolo Massa - Presentation at (2011). ACM Hype...Paolo Massa
The paper is at http://www.gnuband.org/papers/social_networks_of_wikipedia/
Wikipedia, the free online encyclopedia anyone can edit, is a live social experiment: millions of individuals volunteer their knowledge and time to collective create it. It is hence interesting trying to understand how they do it. While most of the attention concentrated on article pages, a less known share of activities happen on user talk pages, Wikipedia pages where a message can be left for the specific user. This public conversations can be studied from a Social Network Analysis perspective in order to highlight the structure of the “talk” network. In this paper we focus on this preliminary extraction step by proposing different algorithms. We then empirically validate the differences in the networks they generate on the Venetian Wikipedia with the real network of conversations extracted manually by coding every message left on all user talk pages. The comparisons show that both the algorithms and the manual process contain inaccuracies that are intrinsic in the freedom and unpredictability of Wikipedia growth. Nevertheless, a precise description of the involved issues allows to make informed decisions and to base empirical findings on reproducible evidence. Our goal is to lay the foundation for a solid computational sociology of wikis. For this reason we release the scripts encoding our algorithms as open source and also some datasets extracted out of Wikipedia conversations, in order to let other researchers replicate and improve our initial effort.
Scripts (Python) has been released as open source and networks datasets (in GraphML format) too. See http://sonetlab.fbk.eu/data/social_networks_of_wikipedia/
An Empirical Analysis on Social Capital and Enterprise 2.0 Participation in a...Paolo Massa
An Empirical Analysis on Social Capital and Enterprise 2.0 Participation in a Research Institute
by
Ferron Michela, Frassoni Marco, Massa Paolo, Napolitano Maurizio, Setti Davide
SoNet project - Fondazione Bruno Kessler (FBK) - Trento, Italy
http://sonet.fbk.eu
2010 International Conference on Advances in Social Networks Analysis and Mining
Odense, Denmark
August 09-August 11
ISBN: 978-0-7695-4138-9
The paper is at http://www.gnuband.org/papers/an_empirical_analysis_on_social_capital_and_enterprise_20_participation_in_a_research_institute
Supporting Collaborative Networks in Organizational Settings using an Enterpr...Paolo Massa
Presentation of the paper "Supporting Collaborative Networks in Organizational Settings using an Enterprise 2.0 platform" at NETSCI 09 International Workshop and Conference on Complex Networks and their Applications, Venezia, Italy. July 2009
The paper is at http://www.gnuband.org/papers/supporting_collaborative_networks_in_organizational_settings_using_an_enterprise_20_platform/
The Future of Work, Fun, and Being Social: an introduction to the nascent adv...Paolo Massa
How Internet Reputation Systems and
The Online Coordination of Offline Life are
Changing the Fundamental Structure of Society
v1.0 28 Feb 2007 Joe Edelman <joe>
on
CouchSurfing Int’l & Emergency Communities
CC-SA-BY
Feedback Effects Between Similarity And Social Influence In Online CommunitiesPaolo Massa
SoNet Research Meeting presentation
Feedback Effects Between Similarity And Social Influence In Online Communities.
Authors: David Crandall, Dan Cosley, Daniel Huttenlocher, Jon Kleinberg, Siddharth Suri
Cornell University Ithaca, NY
2008 KDD: Proceeding of the 14th ACM KDD international conference on Knowledge discovery and data mining
#citations at 2010/04/09 from Google Scholar:44
Presenter: Paolo Massa, SoNet group, http://sonet.fbk.eu
Bowling Alone and Trust Decline in Social Network SitesPaolo Massa
In this paper we analyze the community of a social network site, Advogato. The peculiar characteristics of Advogato is that users can explicitly express weighted trust relationships among themselves. We conduct a longitudinal analysis of the trust network over a time period of 4 years, exploring the community as it grew from a knit circle of 300 users to an society of almost 6500 individuals. We report the changes over time of standard indexes in social network analysis such as clustering and degrees of separation. We then focus on specific measures about trust such as reciprocity and changes over time of average trust. A decline in trust is observed as the community grows. Following what we believe to be the first empirical analysis of trust evolution over time in a real community, we conclude suggesting how the availability of data about human relationships in social network sites is opening up the possibility of monitoring changes in trust in real time. In order to foster this research line, we released the datasets and the code we used in our analysis.
Welocme to ViralQR, your best QR code generator.ViralQR
Welcome to ViralQR, your best QR code generator available on the market!
At ViralQR, we design static and dynamic QR codes. Our mission is to make business operations easier and customer engagement more powerful through the use of QR technology. Be it a small-scale business or a huge enterprise, our easy-to-use platform provides multiple choices that can be tailored according to your company's branding and marketing strategies.
Our Vision
We are here to make the process of creating QR codes easy and smooth, thus enhancing customer interaction and making business more fluid. We very strongly believe in the ability of QR codes to change the world for businesses in their interaction with customers and are set on making that technology accessible and usable far and wide.
Our Achievements
Ever since its inception, we have successfully served many clients by offering QR codes in their marketing, service delivery, and collection of feedback across various industries. Our platform has been recognized for its ease of use and amazing features, which helped a business to make QR codes.
Our Services
At ViralQR, here is a comprehensive suite of services that caters to your very needs:
Static QR Codes: Create free static QR codes. These QR codes are able to store significant information such as URLs, vCards, plain text, emails and SMS, Wi-Fi credentials, and Bitcoin addresses.
Dynamic QR codes: These also have all the advanced features but are subscription-based. They can directly link to PDF files, images, micro-landing pages, social accounts, review forms, business pages, and applications. In addition, they can be branded with CTAs, frames, patterns, colors, and logos to enhance your branding.
Pricing and Packages
Additionally, there is a 14-day free offer to ViralQR, which is an exceptional opportunity for new users to take a feel of this platform. One can easily subscribe from there and experience the full dynamic of using QR codes. The subscription plans are not only meant for business; they are priced very flexibly so that literally every business could afford to benefit from our service.
Why choose us?
ViralQR will provide services for marketing, advertising, catering, retail, and the like. The QR codes can be posted on fliers, packaging, merchandise, and banners, as well as to substitute for cash and cards in a restaurant or coffee shop. With QR codes integrated into your business, improve customer engagement and streamline operations.
Comprehensive Analytics
Subscribers of ViralQR receive detailed analytics and tracking tools in light of having a view of the core values of QR code performance. Our analytics dashboard shows aggregate views and unique views, as well as detailed information about each impression, including time, device, browser, and estimated location by city and country.
So, thank you for choosing ViralQR; we have an offer of nothing but the best in terms of QR code services to meet business diversity!
UiPath Test Automation using UiPath Test Suite series, part 4DianaGray10
Welcome to UiPath Test Automation using UiPath Test Suite series part 4. In this session, we will cover Test Manager overview along with SAP heatmap.
The UiPath Test Manager overview with SAP heatmap webinar offers a concise yet comprehensive exploration of the role of a Test Manager within SAP environments, coupled with the utilization of heatmaps for effective testing strategies.
Participants will gain insights into the responsibilities, challenges, and best practices associated with test management in SAP projects. Additionally, the webinar delves into the significance of heatmaps as a visual aid for identifying testing priorities, areas of risk, and resource allocation within SAP landscapes. Through this session, attendees can expect to enhance their understanding of test management principles while learning practical approaches to optimize testing processes in SAP environments using heatmap visualization techniques
What will you get from this session?
1. Insights into SAP testing best practices
2. Heatmap utilization for testing
3. Optimization of testing processes
4. Demo
Topics covered:
Execution from the test manager
Orchestrator execution result
Defect reporting
SAP heatmap example with demo
Speaker:
Deepak Rai, Automation Practice Lead, Boundaryless Group and UiPath MVP
Essentials of Automations: Optimizing FME Workflows with ParametersSafe Software
Are you looking to streamline your workflows and boost your projects’ efficiency? Do you find yourself searching for ways to add flexibility and control over your FME workflows? If so, you’re in the right place.
Join us for an insightful dive into the world of FME parameters, a critical element in optimizing workflow efficiency. This webinar marks the beginning of our three-part “Essentials of Automation” series. This first webinar is designed to equip you with the knowledge and skills to utilize parameters effectively: enhancing the flexibility, maintainability, and user control of your FME projects.
Here’s what you’ll gain:
- Essentials of FME Parameters: Understand the pivotal role of parameters, including Reader/Writer, Transformer, User, and FME Flow categories. Discover how they are the key to unlocking automation and optimization within your workflows.
- Practical Applications in FME Form: Delve into key user parameter types including choice, connections, and file URLs. Allow users to control how a workflow runs, making your workflows more reusable. Learn to import values and deliver the best user experience for your workflows while enhancing accuracy.
- Optimization Strategies in FME Flow: Explore the creation and strategic deployment of parameters in FME Flow, including the use of deployment and geometry parameters, to maximize workflow efficiency.
- Pro Tips for Success: Gain insights on parameterizing connections and leveraging new features like Conditional Visibility for clarity and simplicity.
We’ll wrap up with a glimpse into future webinars, followed by a Q&A session to address your specific questions surrounding this topic.
Don’t miss this opportunity to elevate your FME expertise and drive your projects to new heights of efficiency.
Generative AI Deep Dive: Advancing from Proof of Concept to ProductionAggregage
Join Maher Hanafi, VP of Engineering at Betterworks, in this new session where he'll share a practical framework to transform Gen AI prototypes into impactful products! He'll delve into the complexities of data collection and management, model selection and optimization, and ensuring security, scalability, and responsible use.
Transcript: Selling digital books in 2024: Insights from industry leaders - T...BookNet Canada
The publishing industry has been selling digital audiobooks and ebooks for over a decade and has found its groove. What’s changed? What has stayed the same? Where do we go from here? Join a group of leading sales peers from across the industry for a conversation about the lessons learned since the popularization of digital books, best practices, digital book supply chain management, and more.
Link to video recording: https://bnctechforum.ca/sessions/selling-digital-books-in-2024-insights-from-industry-leaders/
Presented by BookNet Canada on May 28, 2024, with support from the Department of Canadian Heritage.
Securing your Kubernetes cluster_ a step-by-step guide to success !KatiaHIMEUR1
Today, after several years of existence, an extremely active community and an ultra-dynamic ecosystem, Kubernetes has established itself as the de facto standard in container orchestration. Thanks to a wide range of managed services, it has never been so easy to set up a ready-to-use Kubernetes cluster.
However, this ease of use means that the subject of security in Kubernetes is often left for later, or even neglected. This exposes companies to significant risks.
In this talk, I'll show you step-by-step how to secure your Kubernetes cluster for greater peace of mind and reliability.
Key Trends Shaping the Future of Infrastructure.pdfCheryl Hung
Keynote at DIGIT West Expo, Glasgow on 29 May 2024.
Cheryl Hung, ochery.com
Sr Director, Infrastructure Ecosystem, Arm.
The key trends across hardware, cloud and open-source; exploring how these areas are likely to mature and develop over the short and long-term, and then considering how organisations can position themselves to adapt and thrive.
State of ICS and IoT Cyber Threat Landscape Report 2024 previewPrayukth K V
The IoT and OT threat landscape report has been prepared by the Threat Research Team at Sectrio using data from Sectrio, cyber threat intelligence farming facilities spread across over 85 cities around the world. In addition, Sectrio also runs AI-based advanced threat and payload engagement facilities that serve as sinks to attract and engage sophisticated threat actors, and newer malware including new variants and latent threats that are at an earlier stage of development.
The latest edition of the OT/ICS and IoT security Threat Landscape Report 2024 also covers:
State of global ICS asset and network exposure
Sectoral targets and attacks as well as the cost of ransom
Global APT activity, AI usage, actor and tactic profiles, and implications
Rise in volumes of AI-powered cyberattacks
Major cyber events in 2024
Malware and malicious payload trends
Cyberattack types and targets
Vulnerability exploit attempts on CVEs
Attacks on counties – USA
Expansion of bot farms – how, where, and why
In-depth analysis of the cyber threat landscape across North America, South America, Europe, APAC, and the Middle East
Why are attacks on smart factories rising?
Cyber risk predictions
Axis of attacks – Europe
Systemic attacks in the Middle East
Download the full report from here:
https://sectrio.com/resources/ot-threat-landscape-reports/sectrio-releases-ot-ics-and-iot-security-threat-landscape-report-2024/
State of ICS and IoT Cyber Threat Landscape Report 2024 preview
Trust in Recommender Systems
a historical overview and recent developments
1. Trust in Recommender Systems
a historical overview and recent
developments
Paolo Massa
Universita' di Trento e ITC/iRST
http://moloko.itc.it/paoloblog/
(adapted by Hassan Masum)
Slides licenced under CreativeCommons AttributionShareAlike (see last slide for more info) 1
5. Info Overload
■ 5 seconds:
Scientific information written in this 5 seconds can keep you
busy reading for 40 minutes (based on 1985 data!)
400KB of new text published on paper (24TB printed each
year, 2000, “How Much Information” project at Berkeley)
You have received an email (probably spam) ;)
■ Is this true? Who can tell .... Take facts with
a grain of salt.
■ quot;Technology reduces the amount of time it takes to do any one
task but also leads to the expansion of tasks that people are
expected to do.quot; Juliet Schor 5
6. Info Overload (IO)
■ IO refers to the state of having too much
information to make a decision or remain
informed about a topic.
■ The term was coined in 1970 by Alvin Toffler
in his book “Future Shock.”
■ http://en.wikipedia.org/wiki/Information_overload
■ Too much information can be worse than too little –
illusion of being informed
6
7. Info Overload Stats
■ (NO NO – I'm not reading it, it is just a practical example of information overload!)
■ The daily New York Times now contains more information that the 17th century man or woman would have encountered in a
lifetime. (Wurman, S.A. (1987) Information Anxiety. New York: Doubleday, 32.)
■ quot;As we go from grade school to high school we learn only a billionth of what there is to learn. There is enough scientific information written every
day to fill seven complete sets of Encyclopedia Britannica; there is enough scientific information written every year to keep a person busy
reading day and night for 460 years!quot; (Siegel, B.L. (1984, April 15). Knowledge with commitment: Teaching is the central task of the
university. Vital Speeches of the Day, 50, 394.)
■ quot;In the last 30 years mankind has produced more information than in the previous 5,000.quot; (Information Overload Causes Stress. (1997,
March/April). Reuters Magazine. Available: Lexis Nexis Universe [4/28/98].)
■ Gordon Moore, cofounder of Intel, coined Moore's Law which states that the processing power of computer chips doubles about every 18 months.
■ quot;About 1,000 books are published internationally every day, and the total of all printed knowledge doubles every five years.
(Information Overload Causes Stress. (1997, March/April). Reuters Magazine. Available: Lexis Nexis Universe [4/28/98].)
■ quot;The average Fortune 1000 worker already is sending and receiving approximately 178 messages and documents each day, according to a recent
study, quot;Managing Corporate Communications in the Information Age.quot; (Boles, M. (1997) Help! Information overload. Workforce, 76, 20.)
■ quot;Dr Dharma Singh Khalsa, in his book Brain Longevity,...says the average American sees 16,000 advertisements, logos, and labels in a day.quot; (Gore,
A. (1998, January 18) . Stressed? Maybe it's information overload. Sun Herald, 27.)
■ University of California Berkely has a quot;How Much Informationquot; project which studies the amount of information produced each year. quot;The world's
total yearly production of print, film, optical, and magnetic content would require roughly 1.5 billion gigabytes of storage. This is the equivalent of
250 megabytes per person for each man, woman, and child on earth.quot; Berkeley: How Much Information?
(http://www.sims.berkeley.edu/research/projects/howmuchinfo/)
■ http://library.humboldt.edu/~ccm/fingertips/ioverloadstats.html
■ http://www.sims.berkeley.edu/research/projects/howmuchinfo2003/execsum.htm#summary
■ Data Smog: Surviving the Information Glut, by David Shenk 7
9. Info Retrieval vs Info Filtering
■ Info Retrieval: deals with static information
(Reuters, a database, a book): you want to find
information that is “lying there”
■ Info Filtering: deals with dynamic information
(flows such as the Web or the media): you want to
prioritize important incoming information, and
block the rest
■ Relevance and Quality of items
On a paper repository like Citeseer: no papers about “spam”
but good papers about “spam”
Which “spam” papers are worth your while? 9
12. ContentBased RSs
●
RSs find items similar to ones you liked in past. How?
Analyse the “syntactic content” of all the items.
●
Example: If you like papers containing word “Info
Retrieval”, RS recommends to you another paper with the
word “Info Retrieval” in it.
●
If you read news containing word “Darfur”, it recommends
to you other news with the word “Darfur”.
●
If you like movies of Kubrik, you get one more movie of
Kubrik.
●
Techniques of Info Retrieval ...
.... What are the weaknesses? STOP!
12
13. ContentBased RSs weaknesses
●
Good for text: If you like papers containing word
“Info Retrieval”, RS recommends you another
paper with the word “Info Retrieval” in it. (And
partially effective ways to find “similar” papers
vector space, LSI.)
●
For movies or songs, humans must tag the
content (genre, actors, year, ...) but this is time
consuming, costly, errorsprone and subjective.
– Can your employees “correctly” tag all the
podcasts? All the videos? All the photos? 13
14. ContentBased RSs weaknesses
Contentbased RSs weaknesses summary:
●
Text Items (papers, news): Doable but RSs tend
to propose always the same soup (boring).
Difficult to recognize synonyms, concepts, or
new emerging words (such as “folksonomy”).
●
Movies or Songs: not parsable at the moment by
machines, so humans must tag them.
●
Jokes (or subjective items such as political
ideas): What are the “right” features? Tagging
“objectively” is not possible! 14
15. Collaborative Filtering
●
Users give ratings to items (implicit or
explicit)
●
I like “Titanic” as 4/5
●
RS finds users similar to you (User
similarity)
●
Suggests to you items liked by similar user
Idea: out there, there is someone that is similar to
you and you will like what they liked. 15
16. 1
Item2
4
Item3
Item
Item
Ratings from 1 (min) to 5 (max)
ME 2 5 ? 5
2 5 5 5
Sim(ME,User2) = 0,2 User2 5 1 3
Sim(ME,User3) = 0,4 User3 5 5 1
Sim(ME,User4) = +0,9 User4 2
2 5
5 5
5 4
4
It does not consider the content of the items, only
the ratings given by users.
It works independently of the domain (also jokes)
BUT
Overlapping of rated items required!
16
20. CF WEAKNESSES!!!
■ User Similarity often not computable
– Ratings Matrix sparseness (9599%) > Low or No
overlapping
■ Cold start
– New users have 0 ratings (> not comparable)
– At the beginning, your RS is not Amazon!
■ Easy Attacks by Malicious Users
– Copy profile and become the most similar
– Even easier on the Semantic Web
■ Hard to understand and control
– Black box (bad recs > user gives up)
Solution? Trust! 20
21. Trustawareness
Trust: explicit rating of a user on another user
●
about the perceived quality of the user's
characteristics
●
in RSs, you “trust” someone if you like her
tastes
We will now speak about trust and trust metrics
and then we will come back to “trust and RSs” 21
23. Trust networks
■ Aggregate all the trust statements to produce a
trust network. A node is a user.
A direct edge is a trust statement
0
Mena Ben
0.2 Properties of Trust:
0.9 weighted (0=distrust, 1=max trust)
0.6
subjective
1
ME Doc asymmetric contextdependent?
Trust Metric (TM):
? ? Uses existing edges for predicting values
of trust for nonexisting edges.
1 Thanks to trust propagation, if you trust
Cory Mary
someone, then you have some degree of
trust in anyone that person trusts. 23
24. PageRank: a trust metric?
Imagine the web as a ■ Nodes are web pages, Edges
trust network are links (not weighted).
Web
■ PageRank
Web
page (Google)computes the
page
“importance” of every single
page based on number and
Web Web
page quality of incoming edges...
page
■ So, YES: PageRank is a
trust metric.
Web Web ■ HITS as well.
page page
24
26. TM perspective: Local or Global
1 1
Mary Mena Bill
How much can Bill be trusted?
0 On average (by the community)?
ME 1 By Mary?
Doc
And by ME?
■ Global Trust Metrics:
“Reputation” of user is based on number and quality of incoming edges. Bill has
just one predicted trust value (0.5).
PageRank (eBay, Slashdot, ). Work bad for controversial people (bush)
■ Local Trust Metrics
Trust is subjective > consider personal views (trust “Bill”?)
AppleSeed, Golbeck TM, Advogato, ...
Local can be more effective if people are not standardized. 26
27. Local vs Global
■ Search engine: abortion, jew, scientology, ...
■ Who can define what is spam? Google? A site that opposes
Chinese Comm. Party should be removed?
■ Local vs global:
Is gwbush.com a good page? Is johnkerry.com a good page? Is
sex.com a good page?
■ Maybe these questions are meaningless?
It depends on YOUR LOCAL point of view!
republican/democrats, child/parent, federal/newglobal,
catholic/atheist, pro/against abortion, ...
■ Tyranny of the majority / Daily Me (Sunstein)
27
28. Sociology and Trust
■ Is this Sociology?
■ Yes ...
■ You have seen many graphs, but the first to model groups
in this way was Moreno, a sociologist (1934, sociogram).
■ Social network analysis
(faculty.ucr.edu/~hanneman/nettext/)
■ Degree, betweeness, centrality, ...
■ Is this Politics? Yes ...
■ Read “Republic.com” and “Why Societies Need Dissent” by
Cass Sunstein 28
29. Economy and Trust
■ Is this Economy? Yes...
■ Reputation is an asset, for companies (marketing) but
also for people
■ Centrality in Network is money as well.
■ Open source movement: your peers knows you and will
hire you when they need someone they trust and value.
But also researchers (who gets the next Nobel in
Physics? The most “trusted” by physicists!)
■ Read “Down and out in the Magic Kingdom” SciFi,
reputation (whuffie) is the only currency
29
30. Trust and Search Engines
■ 3 generation search engines:
rd
■ personalization of results based on trust networks
(LOCAL!), based on what your friends like/dislike.
■ Google and Yahoo! are moving in this direction (I'm
speculating). [TrustRank]
■ Problem: Scalability! You cannot recompute
PageRank of every site for every user!
■ But you can do it on your laptop/mobile for
yourself, aggregating only the information “close”
to you ...
30
31. Which Trust Metric works better?
■ And under which conditions?
■ Still an open question. [you can work on it ;) ]
■ Few papers until now evaluate trust metrics:
●
Input data not easily available
●
(advogato.org (8K), FOAF, epinions.com(150K), ... but
not weighted)
■ No papers compare different TMs
Leaveoneout technique
■ Is local better than global? Only for the few users who are
atypical? Computational Expensive? Attackresistant
(googlebomb)? 31
32. Trust propagation
1
Mena Bill
0.6
0.2
ME 0.8
Doc
■ Trust chains (propagation)
■ Combining different trust chains
■ 0.6 * 1 = 0.6, and 0.8 * 0.2 = 0.16
■ Then average? Not that simple ...
■ And how far does trust propagate? 32
33. Trust metrics open issues
(There are no comparative evaluations of TMs)
■ Cycles are a problem > Order peers based on distance from source user
Trust of users at level k is based only on trust of users at level k1 (and k)
Trust propagation horizon (computation)
Find all trust paths from source to target
Propagate trust along trust paths
Trust decay: every hop reduces trust (or certainty of trust).
●
A user can't propagate more than received trust.
●
Distrust (trust=0) blocks the propagation.
Trust about quality vs Trust as judger
Tquality(A,C)=f(Tjudger(A,B),Tquality(B,C))
Combine different trust paths
– Unpredictable Trust = minimum trust value.
– There are no globally “bad” users.
1 1
– Warn about Paradoxes or inconsistencies.
1 0 33
34. How to use distrust
■ Distrust? Opinions of distrusted peers should
simply be discarded, otherwise they could
manipulate them to influence our recs
■ Example: suppose we distrust some that is
distrusted by our enemy, then our enemy could
say “I trust A” and we come to distrust A (that
could be anyone ... from pope to bush)
■ But, it is worth knowing about someone who is
trusted by many, even if distrusted by you...
34
36. RS evaluation: let us count the ways...
■ Many ways to evaluate Recommender Systems.
■ Leaveoneout: hide one rating and try to predict it
Accuracy: are predictions correct?
Coverage: how often are we able to predict?
■ Accuracy: differences between real value and predicted
value.
MAE, MSE, Weighted MAE, MAUE, ...
■ Ability to identify some new items user will like
(unwatched movie), or bad items (spam, products).
■ Evaluation is still problematic 36
38. Trust Propagation
ME
6 degrees of separation “theorem” (Stanley Milgram, 1967)
With few trust steps it is possible to reach every person in
the world! (but more steps needed for highertrust actions)
> Ideally, using trust metrics, no more unknown users. 38
39. Trust solves RS problems
■ User Similarity often not computable
➔
trust propagation and “6 degrees” > we are
now able to predict trust for many users
■ Cold start
➔
“just add 1
friend”
■ Easy copyprofile Attacks
➔
“you can be similar but if no trust path to
you ...”
■ Hard to understand and control
➔
Showing Trust Networks supports
Explanation 39
40. Epinions.com Experiments
■ Some experiments to show that trust solves RSs problems...
■ Epinions.com users can
Review and rate items (from 1 to 5)
Keep web of trust (trust=1) and block list (trust=0). [Epinions
FAQ says to put in Web of Trust “Reviewers whose reviews and
ratings you have consistently found to be valuable”]
■ Dataset (collected by crawling site):
~50K users, ~140K items, ~660K ratings.
~500K trust statements.
➔
No block list (not shown on site, kept hidden)
40
42. UserSimilarity and Trust
computability
Mean number of Comparable users for Mean number of Comparable users for
All users Cold Start users
Propagating Trust Using Propagating Trust Using
Pearson Pearson
Dist 1 Dist 2 Dist 3 Dist 4 Dist 1 Dist 2 Dist 3 Dist 4
9.88 400 4386 16334 161 2.14 94.54 1675 9121 2.74
42
46. Experimental Results
Rows:
#Expressed Ratings ALL 2 3 4 UserSim = CollaborativeFiltering
User population size 40169 3937 2917 2317
Trustx = Trust propagation up to distance x
Mean Web of Trust Size 9.88 2.54 3.15 3.64
Ratings UserSim 51% N/A 4% 8%
Coverage Trust-1 28% 10% 11% 12% RatingsCoverage = how many hidden
60% 23% 26% 31%
Trust-2
ratings are predictable.
Trust-3 74% 39% 45% 51%
Trust-4 77% 45% 53% 59%
UsersCoverage = how many users get at
Users UserSim 41% N/A 6% 14% least a prediction
Coverage Trust-1 45% 17% 25% 32% MAE = |real_ratingpred_rating| averaged
56% 32% 43% 53%
Trust-2
over all the ratings.
Trust-3 61% 46% 57% 64%
Trust-4 62% 56% 59% 66% MAUE = |real_ratingpred_rating| averaged
Mean UserSim 0.843 N/A 1.244 1.027 over the ratings of one user, then averaged
Absolute Trust-1 0.837 0.929 0.903 0.840 over all users.
Error Trust-2 0.829 1.050 0.940 0.927
(MAE) Trust-3 0.811 1.046 0.940 0.918
Trust-4 0.805 1.033 0.926 0.903 Columns:
Views over users.
ALL = all the users (with at least 1 rating)
2 = only the subset of users that gave 2
ratings (there are 3937)
(similarly for 3 and 4)...
46
48. Experimental Results
On average, Trustx
#Expressed Ratings ALL 2 3 4
User population size 40169 3937 2917 2317 achieves better coverage
without loss of accuracy.
Mean Web of Trust Size 9.88 2.54 3.15 3.64
Ratings UserSim 51% N/A 4% 8%
Coverage Trust-1 28% 10% 11% 12%
Trust-2 60% 23% 26% 31% UserSim performs well
Trust-3
Trust-4
74%
77%
39%
45%
45%
53%
51%
59%
with heavy raters and
Users UserSim 41% N/A 6% 14% poorly with cold start users.
Coverage Trust-1 45% 17% 25% 32%
Trust-2 56% 32% 43% 53%
Trust-3 61% 46% 57% 64%
Trust-4 62% 56% 59% 66%
Mean UserSim 0.843 N/A 1.244 1.027
Absolute Trust-1 0.837 0.929 0.903 0.840
Error Trust-2 0.829 1.050 0.940 0.927
(MAE) Trust-3 0.811 1.046 0.940 0.918
Trust-4 0.805 1.033 0.926 0.903
48
49. Experimental Results
On average, Trustx
#Expressed Ratings ALL 2 3 4
User population size 40169 3937 2917 2317 achieves better coverage
without loss of accuracy.
Mean Web of Trust Size 9.88 2.54 3.15 3.64
Ratings UserSim 51% N/A 4% 8%
Coverage Trust-1 28% 10% 11% 12%
Trust-2 60% 23% 26% 31% UserSim performs well
Trust-3
Trust-4
74%
77%
39%
45%
45%
53%
51%
59%
with heavy raters and
Users UserSim 41% N/A 6% 14% poorly with cold start users.
Coverage Trust-1 45% 17% 25% 32%
Trust-2 56% 32% 43% 53%
Trust-3
Trust-4
61%
62%
46%
56%
57%
59%
64%
66%
For cold start users (50% of
Mean UserSim 0.843 N/A 1.244 1.027 the total!), Trustx achieves
Absolute Trust-1 0.837 0.929 0.903 0.840
Error Trust-2 0.829 1.050 0.940 0.927 also better accuracy.
(MAE) Trust-3
Trust-4
0.811
0.805
1.046
1.033
0.940
0.926
0.918
0.903
For bootstrapping RSs,
asking one trust statement
is better than asking one
rating.
(experiments on 660.000 ratings) 49
50. Centralized vs decentralized
■ Another problem with current RS: centralization
■ Information is Centralized in one server
Your “profile” scattered in many RS (Amazon, B&B, ...)
Profile not reusable (your profile in Amazon is NOT yours)
Recommendation computation out of your control
■ Decentralized
The Web is decentralized: anyone can write whatever she
wants, in whatever “language” she wants (spam is good)
No censorship, innovation can happen on the edges, not a
single mind but many minds... you are in control of what you
produce
50
51. Semantic Web
■ A Web of content designed for and understandable by machines.
(matrix?)
■ Promising Semantic Web formats
FOAF (FriendOfAFriend): trust info <
XFN (Xhtml Friend Network): social info
VoteLink: votefor, voteagainst, voteabstain links
Blogroll: not semantic!
hReview: review/rating info <
RSS of OutFoxed (http://getoutfoxed.com/rss)
RVW, OpenReview, ...: review/rating info
■ Allow decentralized publishing of information that RSs
aggregate and exploit. 51
52. Note: Adoption of a Language
■ Suppose you can define the language we have to use
for communicating
Which language is better? Chinese? Italiano? The
one you invent?
■ Interesting question, but as long as “good enough”,
matters little (or not at all) for adoption of language.
■ Do you know why keys in your keyboard are placed in
that way? > how standards get adopted ...
■ Who has the power to “propose” changes in the
language of the Web?
52
53. Format adoption
■ Adoption does not depend on quality of the format but
(mainly) on the authority of the proponent.
■ Google can push changes in HTML (example:
rel=nofollow)
■ Certainly Microsoft could (even without you noticing it
or telling it you)
■ I can't.
■ ...But maybe if I create a format that's immediately
useful, it will be taken up by a user community and
spread “virally” (how HTML started)
53
56. hReview
■ hReview is a simple, open, distributed reviews
format suitable for embedding in (X)HTML, Atom,
RSS, and arbitrary XML.
■ In order to enable and encourage the sharing,
distribution, syndication, and aggregation of
reviews
■ Proposed by Technorati.com on a wiki page
■ http://developers.technorati.com/wiki/hReview
■ You are free to participate and give suggestions
and feedback!
56
58. Conclusions
■ Info Filtering:
■ From Contentbased ... to Collaborative Filtering ... to
Trustaware (?)
■ Trust is a simple and complicated concept
■ It is a rapidly evolving and increasingly important topic:
there is room for your contributions!
■ Forecast: In 3 years, anyone will publish her opinions
(about stuff, organizations...people, ideas?) in some
semantic format and trustaware aggregators
(googletrust?) will help in coping with an increased
information overload.
58