scispace - formally typeset

Journal ArticleDOI

Report on the Models of Trust for the Web workshop (MTW'06)

01 Dec 2006-Vol. 35, Iss: 4, pp 54-56

AbstractWe live in a time when millions of people are adding information to the Web through a growing collection of tools and platforms. Ordinary citizens publish all kinds of content on Web pages, blogs, wikis, podcasts, vlogs, message boards, shared spreadsheets, and new publishing forums that seem to appear almost monthly. As it becomes easier for people to add information to the Web, it is more difficult, and also more important, to distinguish reliable information and sources from those that are not.

Topics: Web development (63%), Web 2.0 (61%), Web standards (61%), Web page (60%), Web design (59%)

Summary (3 min read)

1 Introduction

  • Ordinary citizens publish all kinds of content on Web pages, blogs, wikis, podcasts, vlogs, message boards, shared spreadsheets, and new publishing forums that seem to appear almost monthly.
  • As it becomes easier for people to add information to the Web, it is more difficult, and also more important, to distinguish reliable information and sources from those that are not.
  • The MTW’06 workshop was attended by over thirty researchers for a full day of presentations, panels and spirited discussions.

2 Presented Papers

  • The keynote speaker, Ricardo Baeza-Yates (Yahoo! Research), discussed how social networks can be exploited to provide social and economic deterrents for spamming.
  • There are several kinds of spam that need to be monitored: scraper scam that copies good data from other sites and adds monetization, synthetic text that provides boilerplate text built around key phrases, query-targeted spam in which each page targets a single tail query, DNS spam where many domains use the same servers, and blog spam.
  • Using Flickr as an example, Ricardo showed how the “wisdom of crowds can be used to search” as Flickr users collaboratively search and tag each other’s photos and the anchor text is collective knowledge used to create a search.
  • At Yahoo!, spam is detected and characterized using a combination of algorithmic and editorial techniques in order to prevent it from distorting the rank- ing of web pages.

2.1 Session I : Trust Networks

  • David Brondsema and Andrew Schamp described their work in using social trust networks to filter spam in “Konfidi: Trust Networks Using PGP and RDF”.
  • They proposed that spam can be filtered by reasoning over trust relationships in RDF.
  • These relationships include who (both identity and public key) is trusted, value of trust, and with respect to what topic.
  • They discussed the FilmTrust project, which is a movie recommendation system developed using their approach.
  • Patricia Victor et al. defined their billatice trust model that takes trust, distrust, lack of data, and contradictory data into consideration while calculating trust in “Towards a Provenance-Preserving Trust Model in Agent Networks”.

2.2 Session II : Inferring Trust

  • The paper “Propagating Trust and Distrust to Demote Web Spam” by Baoning Wu, Vinay Goel, and Brian Davison addressed the problem of web spam also known as search engine spam in which a target page gets undeserved ranking.
  • They described different methods that a parent page can use to divide its trust or distrust among its child pages.
  • L. Jean Camp, Cathleen McGrath, and Alla Genkina approached human trust behavior from a social science perspective.
  • They described their results in “Security and Morality: A Tale of User Deceit“ in which they present how users “consider failures in benevolence more serious than failures in competence”.
  • Deborah McGuinness et al. reported in “Investigations into Trust for Collaborative Information Repositories: A Wikipedia Case Study“ that both provenance of information and revision details are required to improve the trustworthiness of collaborative information systems such as Wikipedia.

2.3 Session III : Trust Models

  • Santtu Toivonen, Gabriele Lenzini, and Ilkka Uusitalo explored the role of context in trust determination in their paper “Context-aware Trust Evaluation Functions for Dynamic Reconfigurable Systems”.
  • They distinguished between quality attributes, which are static attributes of the trustee and context attributes, which are optional attributes that can change dynamically such as location, and device type.
  • They discussed how context-aware trust is calculated from quality attributes, reputation, and recommendations within a certain trust scope at a certain time.
  • In “How Certain is Recommended Trust-Information”, Uthe authors Roth and Volker Fusenig suggested that trust information given by a recommender may not be reliable and could negatively affect the trust decision.
  • They proposed a strategy for making trust decisions based on a converting a network of relations of direct and recommended trust information into a decision tree by choosing the path as well as whether to trust the recommended information along the path at random.

2.4 Session IV : Trust in Applications

  • The paper titled “Quality Labeling of Web Content: The Quatro approach” by Vangelis Karkaletsis et al. reports on a common machine-readable vocabulary for labelling web content that will be represented by user friendly icons for ease of understanding.
  • The vocabulary includes several kinds of labels: page-specific such as whether the page uses clear language, whether it includes a privacy statement, content provider specific such as whether his credentials have been verified, business-specific such as whether it complies with rules and regulations of ebusiness, and label-specific such as when the label was issued, and when it was last viewed.
  • Ing-Xiang Chen and Cheng-Zen Yang examined biased search engine results and their deviations in “A Study of Web Search Engine Bias and its Assessment”.
  • The authors proposed a twodimensional scheme by adopting both indexical bias (differences in the sets of URLs retrieved) and content bias (deviations of content).
  • Tsow suggested that users are prone to attacks from malicious software embedded on their routers in “Phishing with Consumer Electronics - Malicious Home Routers” and hinted that trust is not a software only matter but there is an implicit trust in hardware vendors.

3 Future Directions / Open Research Issues

  • TheModels of Trust for the Webworkshop helped to understand current state-of-the-art research and to provide a discussion forum for researchers working on trust issues.
  • Some open questions and problems include: Trust modeling: Trust awareness:Users’ trust on computers highly depends on previous experiences.
  • Furthermore, SWAM makes this task even more difficult.
  • Trust in data and systems over time (compliance storage) is currently a major issue (specially after latest laws in which companies need to store data for longer time), also known as Database Security.

4 Conclusion

  • The Web and it’s evolving infrastructure have made it easy to access virtually all of the world’s knowledge and is the first source to which most of us turn when the authors need to know something.
  • Search engines and other tools have focused on finding informationrelevantto users’ queries with results ranked at best by their popularity.
  • The papers from the WWW’06Models of Trust for the Webworkshop addressed these issues, identified important issues, and offered some partial solutions.

Did you find this useful? Give us your feedback

...read more

Content maybe subject to copyright    Report

Report on the Models of Trust for the Web Workshop (MTW’06)
Tim Finin
University of Maryland, Baltimore County
Maryland MD 21250 USA
finin@umbc.edu
Lalana Kagal
Massachusetts Institute of Technology
Cambridge MA 02139 USA
lkagal@csail.mit.edu
Daniel Olmedilla
L3S Research Center and Hannover University
30539 Hannover Germany
olmedilla@l3s.de
1 Introduction
We live in a time when millions of people are adding
information to the Web through a growing collection of
tools and platforms. Ordinary citizens publish all kinds of
content on Web pages, blogs, wikis, podcasts, vlogs, mes-
sage boards, shared spreadsheets, and new publishing fo-
rums that seem to appear almost monthly. As it becomes
easier for people to add information to the Web, it is more
difficult, and also more important, to distinguish reliable
information and sources from those that are not.
Search engines excel at finding results that are relevant
to a user’s query, but many are outdated, biased, inac-
curate, and/or from unreliable sources. Popularity based
metrics such as Google’s PageRank help, but users are
still forced to filter the results to select the most reliable
information based on their particular trust requirements.
With the introduction of web services, the problem is fur-
ther exacerbated as users have to come up with a new set
of requirements for trusting web services and web ser-
vices themselves require a more automated way of trust-
ing each other. Apart from inaccurate or outdated infor-
mation, we also need to anticipate Semantic Web spam
(SWAM) where malicious sources publish false facts
and scams to deliberately mislead software agents and
programs.
The Models of Trust for the Web workshop was held in
conjunction with the 15th International World Wide Web
Conference (WWW2006) on 22nd May, 2006 in Edin-
burgh, Scotland. The goal of the workshop was to bring
together researchers and experts from different communi-
ties (e.g., Information Systems, Database, Semantic Web,
Web Services) who have been working on topics like trust,
provenance, privacy, security, reputation, and spam, in or-
der to understand the challenges associated with facilitat-
ing trust on the Web, to deliver a state-of-the-art overview
in the area, and to identify guidelines for future research.
The workshop built on several related workshops, includ-
ing the Workshop on Policy Management for the Web [1]
held at the 2005 World Wide Web conference and Seman-
tic Web and Policy Workshop [2] held in conjunction with
2005 International Semantic Web Conference.
The MTW’06 workshop was attended by over thirty re-
searchers for a full day of presentations, panels and spir-
ited discussions. The eleven papers that were presented
covered a wide spectrum of topics from inferring trust,
to using trust to prevent spam, and the role of social net-
works in calculating trust [3].
2 Presented Papers
The keynote speaker, Ricardo Baeza-Yates (Yahoo! Re-
search), discussed how social networks can be exploited
to provide social and economic deterrents for spamming.
There are several kinds of spam that need to be mon-
itored: scraper scam that copies good data from other
sites and adds monetization, synthetic text that provides
boilerplate text built around key phrases, query-targeted
spam in which each page targets a single tail query, DNS
spam where many domains use the same servers, and blog
spam. Using Flickr as an example, Ricardo showed how
the “wisdom of crowds can be used to search” as Flickr
users collaboratively search and tag each other’s photos
and the anchor text is collective knowledge used to cre-
ate a search. At Yahoo!, spam is detected and charac-
terized using a combination of algorithmic and editorial
techniques in order to prevent it from distorting the rank-
1

ing of web pages.
2.1 Session I : Trust Networks
David Brondsema and Andrew Schamp described their
work in using social trust networks to filter spam in “Kon-
fidi: Trust Networks Using PGP and RDF”. They pro-
posed that spam can be filtered by reasoning over trust re-
lationships in RDF. These relationships include who (both
identity and public key) is trusted, value of trust, and with
respect to what topic.
In “Using Trust and Provenance for Content Filter-
ing on the Semantic Web”, Jennifer Golbeck and Aaron
Mannes showed that annotating relationships with binary
trust values in web-based social networks allowed trust
values to be inferred between unrelated entities. They
discussed the FilmTrust project, which is a movie recom-
mendation system developed using their approach.
Patricia Victor et al. defined their billatice trust model
that takes trust, distrust, lack of data, and contradictory
data into consideration while calculating trust in “To-
wards a Provenance-Preserving Trust Model in Agent
Networks”.
2.2 Session II : Inferring Trust
The paper “Propagating Trust and Distrust to Demote
Web Spam” by Baoning Wu, Vinay Goel, and Brian Davi-
son addressed the problem of web spam also known as
search engine spam in which a target page gets unde-
served ranking. They described different methods that a
parent page can use to divide its trust or distrust among its
child pages. They also defined mechanisms for calculat-
ing trust (using outgoing links) and distrust (using incom-
ing links) - accumulation, maximum share, and maximum
parent.
L. Jean Camp, Cathleen McGrath, and Alla Genkina
approached human trust behavior from a social science
perspective. They described their results in “Security and
Morality: A Tale of User Deceit“ in which they present
how users “consider failures in benevolence more serious
than failures in competence”.
Deborah McGuinness et al. reported in “Investigations
into Trust for Collaborative Information Repositories: A
Wikipedia Case Study“ that both provenance of informa-
tion and revision details are required to improve the trust-
worthiness of collaborative information systems such as
Wikipedia. They discussed citation-based trust, which
is derived from citation relationships among articles, and
revision-based trust, which is derived from the original
article, revision operators, and revision authors, as mech-
anisms for inferring trustworthiness of a Wikipedia arti-
cle. They also presented a mockup of a Wikipedia version
marked up with trust.
2.3 Session III : Trust Models
Santtu Toivonen, Gabriele Lenzini, and Ilkka Uusitalo ex-
plored the role of context in trust determination in their
paper “Context-aware Trust Evaluation Functions for Dy-
namic Reconfigurable Systems”. They distinguished be-
tween quality attributes, which are static attributes of
the trustee and context attributes, which are optional at-
tributes that can change dynamically such as location, and
device type. They discussed how context-aware trust is
calculated from quality attributes, reputation, and recom-
mendations within a certain trust scope at a certain time.
In “How Certain is Recommended Trust-Information”,
Uwe Roth and Volker Fusenig suggested that trust infor-
mation given by a recommender may not be reliable and
could negatively affect the trust decision. They proposed
a strategy for making trust decisions based on a converting
a network of relations of direct and recommended trust in-
formation into a decision tree by choosing the path as well
as whether to trust the recommended information along
the path at random.
2.4 Session IV : Trust in Applications
The paper titled “Quality Labeling of Web Content: The
Quatro approach” by Vangelis Karkaletsis et al. reports
on a common machine-readable vocabulary for labelling
web content that will be represented by user friendly
icons for ease of understanding. The vocabulary includes
several kinds of labels: page-specific such as whether
the page uses clear language, whether it includes a pri-
vacy statement, content provider specific such as whether
his credentials have been verified, business-specific such
as whether it complies with rules and regulations of e-
business, and label-specific such as when the label was
issued, and when it was last viewed.
Ing-Xiang Chen and Cheng-Zen Yang examined biased
search engine results and their deviations in A Study of
Web Search Engine Bias and its Assessment”. An ex-
ample of search engine bias is results in China for the
keyword “Falun Gong”. The authors proposed a two-
dimensional scheme by adopting both indexical bias (dif-
ferences in the sets of URLs retrieved) and content bias
(deviations of content).
Alex Tsow suggested that users are prone to attacks
from malicious software embedded on their routers in
2

“Phishing with Consumer Electronics - Malicious Home
Routers” and hinted that trust is not a software only matter
but there is an implicit trust in hardware vendors.
3 Future Directions / Open Re-
search Issues
The Models of Trust for the Web workshop helped to un-
derstand current state-of-the-art research and to provide a
discussion forum for researchers working on trust issues.
It highlighted the importance of existing lines of research
and brought up some salient emerging problems. Some
open questions and problems include:
Trust modeling: Is trust just boolean or does it have
some certainty associated? Is it transitive? May we
infer that the enemy of my enemy is my friend? Is it
context or time dependent?
Trust awareness: Users’ trust on computers highly
depends on previous experiences. They are initially
too trustful and typically unaware of the risks asso-
ciated with their computer usage. Improving user
interfaces and increasing user awareness on privacy
and security issues remains an open issue.
Trustworthy information: Whether information re-
trieved from unknown sources is correct or not is a
major question in current Web. Furthermore, SWAM
makes this task even more difficult.
Database Security: Trust in data and systems over
time (compliance storage) is currently a major is-
sue (specially after latest laws in which compa-
nies need to store data for longer time). The cur-
rent winner of the best paper award at the Very
Large Database Conference “Trustworthy Keyword
Search for Regulatory-Compliant Record Retention”
[4] demonstrates the importance of this line of re-
search.
Access control & Trust Management: Specification
of conditions under which service access is granted
or private information is released in an open dis-
tributed world has been and is still one of the most
important lines of research. Work on policies is cur-
rently a hot topic in areas such as the Semantic Web
[2, 5] for example.
4 Conclusion
The Web and it’s evolving infrastructure have made it easy
to access virtually all of the world’s knowledge and is
the first source to which most of us turn when we need
to know something. Search engines and other tools have
focused on finding information relevant to users’ queries
with results ranked at best by their popularity. As it be-
comes easier to publish information on the Web, it is
increasingly important to develop good frameworks for
evaluating the trustworthiness of the information found.
The papers from the WWW’06 Models of Trust for the
Web workshop addressed these issues, identified impor-
tant issues, and offered some partial solutions.
References
[1] Lalana Kagal, Tim Finin, and James Hendler, edi-
tors. Proceedings of the WWW’05 Workshop on Pol-
icy Management for the Web, Chiba JP, May 2005.
http://ebiquity.umbc.edu/paper/html/id/221/.
[2] Lalana Kagal, Tim Finin, and James Hendler, edi-
tors. Proceedings of the ISWC 2005 Semantic Web
and Policy Workshop, Galway IE, November 2005.
http://ebiquity.umbc.edu/paper/html/id/268/.
[3] Tim Finin, Lalana Kagal, and Daniel Olmedilla, edi-
tors. Proceedings of the WWW’06 Workshop on Mod-
els of Trust for the Web, volume 190 of CEUR Work-
shop Proceedings, Edinburgh, Scotland, May 2006.
CEUR-WS.org. http://ceur-ws.org/Vol-190.
[4] Soumyadeb Mitra, Windsor W. Hsu, and Marianne
Winslett. Trustworthy keyword search for regulatory-
compliant record retention. In Proceedings of the
32nd International Conference on Very Large Data
Bases, pages 1001–1012, Seoul, Korea, Sep 2006.
[5] Piero A. Bonatti, Li Ding, Tim Finin, and Daniel
Olmedilla, editors. Proceedings of the ISWC’06
2nd International Semantic Web Policy Workshop
(SWPW), Athens GA USA, November 2006.
3
Citations
More filters

Journal ArticleDOI
TL;DR: A trust-aware community approach to facilitate fake media detection by employing the concept of serious gaming and introducing a trust inference algorithm for yet unknown sources uploading and rating media is presented.
Abstract: Nowadays, it becomes increasingly difficult to find reliable multimedia content in the Web 2.0. Open decentralized networks (on the Web) are populated with lots of unauthenticated agents providing fake multimedia. Conventional automatic detection and authentication approaches lack scalability and the ability to capture media semantics by means of forgery. Using them in online scenarios is computationally expensive. Thus, our aim was to develop a trust-aware community approach to facilitate fake media detection. In this paper, we present our approach and highlight four important outcomes. First, a Media Quality Profile (MQP) is proposed for multimedia evaluation and semantic classification with one substantial part on estimating media authenticity based on trust-aware community ratings. Second, we employ the concept of serious gaming in our collaborative fake media detection approach overcoming the cold-start problem and providing sufficient data powering our Media Quality Profile. Third, we identify the notion of confidence, trust, distrust and their dynamics as necessary refinements of existing trust models. Finally, we improve the precision of trust-aware aggregated media authenticity ratings by introducing a trust inference algorithm for yet unknown sources uploading and rating media.

10 citations


Proceedings ArticleDOI
15 Jun 2009
TL;DR: A universal collaborative and competitive approach is introduced for deployment of digital collections in an ideal Digital Library (DL) for future's educational system.
Abstract: In this article, a universal collaborative and competitive approach is introduced for deployment of digital collections in an ideal Digital Library (DL) for future's educational system The collaborative and open-source aspects of the system guarantee its growth and the competitive aspects guarantee the accuracy

References
More filters

Proceedings ArticleDOI
01 Sep 2006
TL;DR: This paper proposes a novel scheme for effcient creation of a trustworthy inverted index and demonstrates, through extensive simulations and experiments with an enterprise keyword search engine, that the scheme can achieve online update speeds while maintaining good query performance.
Abstract: Recent litigation and intense regulatory focus on secure retention of electronic records have spurred a rush to introduce Write-Once-Read-Many (WORM) storage devices for retaining business records such as electronic mail. However, simply storing records in WORM storage is insuffcient to ensure that the records are trustworthy, i.e., able to provide irrefutable proof and accurate details of past events. Specifically, some form of index is needed for timely access to the records, but unless the index is maintained securely, the records can in effect be hidden or altered, even if stored in WORM storage. In this paper, we systematically analyze the requirements for establishing a trustworthy inverted index to enable keyword-based search queries. We propose a novel scheme for effcient creation of such an index and demonstrate, through extensive simulations and experiments with an enterprise keyword search engine, that the scheme can achieve online update speeds while maintaining good query performance. In addition, we present a secure index structure for multi-keyword queries that supports insert, lookup and range queries in time logarithmic in the number of documents.

35 citations



Frequently Asked Questions (1)
Q1. What are the contributions mentioned in the paper "Report on the models of trust for the web workshop (mtw’06)" ?

The Models of Trust for the Web workshop was held in conjunction with the 15th International World Wide Web Conference ( WWW2006 ) on 22nd May, 2006 in Edinburgh, Scotland this paper.