CHARACTERISTICS OF TRUST IN ONLINE SOCIAL
NETWORKS AND COMMUNITY OF TRUST AS A SPECIAL
CASE OF ONLINE COMMUNITY
David Zejda
Faculty of Informatics and Management, University of Hradec Králové, Hradec Králové, Czech Republic
Keywords: Trust, Distrust, Community of trust, Social networks, Community.
Abstract: With boost of interest in Web 2.0 technologies, appropriate trust models are increasingly more important.
First section the paper contains state of the art about trust characteristics, in particular multidimensionality,
contextuality, scope of relevance, transitivity and asymmetry. Transitivity as a key aspect utilized in most
models is described in a slightly greater detail. Discussion on scope of relevance allowed us to introduce
taxonomy of trust from the scope point of view. Based on the general foundation, in the second section we
introduce community of trust as a niche type of online community where users trust each other as default
and where the trust loses most of its subjective flavour.
1 INTRODUCTION
Both individual social interactions and a whole
dynamics of personal social network are highly
influenced by trust. Trust may be defined as “the
willingness of a party to be vulnerable to the actions
of another party based on the expectation that the
other will perform a particular action important to
the trustor, irrespective of the ability to monitor or
control that other party.” (Mayer et al., 1995) For
our work we adopted rather the definition: “Trust in
a person is a commitment to an action based on a
belief that the future actions of that person will lead
to a good outcome.” (Golbeck & Hendler, 2006) The
level of trust which we feel toward someone helps us
to decide whether to rely on his promises or whether
to entrust him an information or a task.
Trust emerges primarily from our experiences
with others, their acts, words, their willingness to
help us in difficulties, promises which have been
kept. Another source of trust is recommendation or
guarantee from those, who we trust already. In
general, trust grows slowly, but falls sharply.
(Walter et al., 2008) It may take months or years
before we credit someone, whereas a single act of
betrayal destroys the trust from the roots.
We all belong to a global-world village. As
expressed in the small world phenomenon, everyone
is connected with anyone else through only several
steps of relations. (Pavlovic, 2009) Current
technology emphasizes the connectedness. Besides
milieu for implicit socialization (Wennerberg &
Oellinger, 2006), web provides variety of explicitly
social spaces, including dating sites, community
portals and social networking sites. If we add pace
of life nowadays, new social strategies are needed to
cope with the social and information overload.
(Walter et al., 2008) Reliable, efficient, and
appropriate trust solutions for social software should
reflect the needs. In the paper we present state of the
art about trust characteristics and define community
of trust as a niche kind of community where trust
among users is a default state.
2 TRUST CHARACTERISTICS
Online interactions may be viewed as a technical
extension of interactions in real world. (Dwyer et al.,
2007) So, trust in online networking systems keeps
most of its general characteristics. Meo et al. (Meo
et al., 2009) define three aspects of trust,
multidimensionality, contextuality and scope of
relevance. Goldbeck et al. identify transitivity,
asymmetry and personalization (Golbeck & Hendler,
2006). Personalization may be viewed as a special
case of scope of relevance. We decided to add
disproportion of impacts and dynamics.
531
Zejda D..
CHARACTERISTICS OF TRUST IN ONLINE SOCIAL NETWORKS AND COMMUNITY OF TRUST AS A SPECIAL CASE OF ONLINE COMMUNITY.
DOI: 10.5220/0003332905310534
In Proceedings of the 7th International Conference on Web Information Systems and Technologies (WEBIST-2011), pages 531-534
ISBN: 978-989-8425-51-5
Copyright
c
2011 SCITEPRESS (Science and Technology Publications, Lda.)
1
Eigenvector-type algorithm is PageRank by Google.
Multidimensionality. There is no single source of
trust, on the contrary various factors may be
considered to evaluate trust, such as honesty,
experience, precision, efficiency, or cooperativeness
of the party. We may mix the indices to get more
complex view. Dimensions grow with breadth of
a social network. In a virtual space on one hand we
miss non-verbal indices. We do not see others in
real, sometimes even not at all. It is also likely that
there are not many trustful people around who could
share their real world experiences. On the other hand
we may take the whole community into account and
use plenty of algorithms to overcome the drawback.
Contextuality. Social context and purpose of
trust evaluation affect our requirements on trust and
the process of trust formation - trust is contextually-
dependant. E.g. when we search for advices on
particular topic, we prefer experts on the domain.
Asymmetry. Trust of one to another does not
imply trust in reverse direction. Graph of trust is
directed, matrix of trust is not necessarily
symmetrical.
Transitivity. Admitting transitivity of trust, we
may follow trust relations to infer trust between
those who do not trust each other yet or who even do
not know each other. Multiplication along the path
performed by most algorithms effectively discounts
the resulting value (Huang & Fox 2006), thus those
whom the user trusts already are being taken more
seriously as a source of recommendations whom else
to trust. The algorithms differ in their focus. E.g.
some of them do not reduce cycles in a graph before
computation (Walter et al. 2009) or may be applied
in an environment with no central authority, e.g. to
find cooperative routes among selfish agents acting
as players in prisoner's dilemma. (Hales & Arteconi,
2005) Work on trust inference comprises e.g.:
(Ziegler & Lausen, 2004) (Kamvar et al., 2003)
(Guha et al., 2004) (Richardson et al. 2003).
Scope of relevance. It is necessary to distinguish
subjective trust to objective trust. Many models treat
trust as inherently subjective. (Golbeck & Hendler,
2006) Meo et al. classify subjective trust,
community-wide reputation, and general reliability.
(Meo et al., 2009) We further split reliability into
system-wide trustfulness and world-wide trust
identity exceeding borders of systems, as described
in Figure 1. Trustworthy user is usually being trusted
subjectively more quickly, reversely trustworthiness
may be inferred from a set of subjective trust
expressions. The inference my be performed with an
eigenvector
1
algorithm, weighing subjective trust
according to trustor's own trust. (Yan & Holtmanns,
2007). In result, trustworthiness of certain user stems
from trustworthiness of his neighbours in the graph
of trust (Walter et al., 2009). Explicit negative
experiences (signs of subjective distrust) may help to
reveal objectively malicious users. We lack
applicable solutions for world-wide trust identity.
Figure 1: Taxonomy of trust.
Disproportion of impacts. We may identify two
complementing types of errors in the process of trust
emergence. The first is 'excessive prudence' when an
user is excessively suspicious. The error inhibits
formation of vital trust and lead to certain losses. On
the contrary 'undue confidence' occurs if an user is
either intentionally careless or if he is prone to fraud
attempts. The second error may lead to more severe
impacts, which should be reflected in trust models.
Dynamics. Caverlee at al. recommend to fold
two main sources of information in a well-designed
trust metric – network topology and record of
behaviour. (Caverlee et al., 2008) Moghaddam et al.
provide model for rapidly evolving networks, with
puts emphasis on feedback as a source of trust.
(Moghaddam et al., 2009) Driven by the dynamics,
trust undergoes transitions between various states, it
may be gained, lowered, or even lost. Conceptual
representations of failures of trust, such as distrust,
mistrust, untrust and ignorance are available. Trust
may be recovered again, when regret followed by
forgiveness takes place. (Golbeck, 2008)
The characteristics mentioned are mutually
interrelated. E.g. contextuality brings further
dynamics to the model, severity of impacts is further
influenced by the context and scope of relevance,
etc. Yan et al. reflect most of the characteristics in
their conditional definition of trust: “Trustor A trusts
trustee B for purpose P under condition C based on
root trust R”. (Yan & Cofta, 2004) Trustor should be
informed about any distrustful behaviour of the
trustee according to the conditions and trust itself is
considered as dependant on the conditions.
3 COMMUNITY OF TRUST
What's the source of trust in social networking
systems? Trusted friendships may arise out of vital
interactions within a site, usually during a sufficient
WEBIST 2011 - 7th International Conference on Web Information Systems and Technologies
532
period of time and based on a sufficient level of
harmless activity. The model is meaningful for most
cases, however, perception of virtue of trust is not
unique among all communities. So, various models
of trust are needed to reflect the needs.
Besides the trust which evolves with online
interactions, also trust existing in a real social
background may be mapped into an online system
(Walter et al., 2008). For example, if you personally
invite someone to join a networking site, you
probably know him already and trust him, at least at
certain degree. The trust has been established in
advance already, based on your real world personal
experiences. You do not ask the system to show you
trustworthiness of the user. Rather reversely, you
may provide trust indices to the system. If we follow
the idea further, 'community of trust' is the scenario
where users of certain online social system trust
each other as default. Distrustful behaviour is rare
there and if occurs, it leads to immediate expulsion
from the community. Community of trust may exist
among relatives, among close friends who know
each other for a long time, among volunteers
working jointly on an issue, among members of a
church with strong influence on adherent's life or
within another group of people bonded with strong
shared principles. Table 1 outlines characteristics of
community of trust, discussed in more detail below.
Table 1: Community of trust vs. a common community.
common community community of trust
model of trust model of distrust
distrustful behaviour
relatively common
distrustful behaviour rare,
propagate distrust quickly
users are notably cautious users are careless
pre-validation of users
not necessary / possible
users have to prove their
membership first
users may express trust or
both trust and distrust
users may express distrust or
confirm trust
trust is important trust is pivotal
trust is to be gained trust is default state
trust is subjective trust is objective
trust is dynamic trust is not too dynamic
trust is transitive distrust is totally transitive
While in online social networking systems
supporting a common community we talk about
a model of trust, in community of trust more
appropriate name is model of distrust, because it
fulfils different purposes. Primarily it helps to reveal
intruders, impostors or those who turned bad.
Besides the main purpose, the model of distrust
indirectly fosters fair interactions within the
community, bringing deeper feeling of reliance and
connectedness. Healthful fear of possible
consequences motivates users to adhere to the
principles which keep the community together and
to avoid any bad behaviour.
According to (Golbeck & Hendler, 2006) trust is
a personal opinion, which means that each node has
different levels of trust for each other node (Meo et
al., 2009), but they admit, that systems based on
objective trust may exist. Community of trust is the
case. It is so tightly coupled that trust loses most of
its subjective flavour and turns objective. As long as
someone belongs to the community, others trust him.
If he behaves badly to one, nobody will trust him
more. Transitivity of distrust in a pure community of
trust is total. So, while in subjective models of trust
it gives sense to infer trust and distrust from a graph
of trust relations following paths of transitivity, in
community of trust it gives sense no more, because
trust is default and transitivity of distrust tends to
infinity. In most models, e.g. (Caverlee et al., 2008),
trust is dynamic, reflecting changes in both network
topology and activities of users. Model for
community of trust is not too dynamic, but distrust
has to be propagated as quickly as possible to the
whole community.
Generally, people are willing to make only the
effort, which brings obvious reward to them. Talking
about trust or distrust models, users should be
allowed to express their (dis)trust in situations and in
a way which reflects their pattern of thinking or their
habitual approach. The approach differs per context
or per community. In community of trust users do
not like to be annoyed with requests to evaluate trust
with every transaction or to express trust of each
other because trust is natural, implicit there. They
only wish to have something at hand to defend
themselves and the whole community if matters go
wrong. Eventually they would also like to confirm
the trust within the community to contribute to its
virtue.
Any model of trust itself should be trusted by
users, which implies that it should be also
understandable. Because trust is so vital within
a community of trust, it further underlines the
requirement to bring appropriate model, and to keep
it understandable. Users have to be authenticated
first before entering a community of trust. Details of
the validation process depend on a particular
community and are out of scope of this paper.
4 CONCLUSIONS
In the paper we outlined state of the art trust
CHARACTERISTICS OF TRUST IN ONLINE SOCIAL NETWORKS AND COMMUNITY OF TRUST AS A SPECIAL
CASE OF ONLINE COMMUNITY
533
solutions for online social networks. Besides
multidimensionality, contextuality, asymmetry,
transitivity, scope of relevance, two more
characteristics of trust have been identified -
disproportion of impacts and trust dynamics.
Subsequently we described basic ideas of trust
processing and inference in models with transitive
trust. Idea of scope of relevance has been extended
into simple taxonomy of trust. As a main
contribution, we introduced 'community of trust' to
describe niche tightly coupled communities where
trust among users is default state. Trust has objective
character there, so tracking paths of trust among
users has no sense. Model of distrust for community
of trust has to propagate every distrust quickly to the
whole community.
ACKNOWLEDGEMENTS
The research was supported by grant UHK FIM
specific research 2110/2010.
REFERENCES
Caverlee, J., Liu, L. & Webb, S., 2008. Towards robust
trust establishment in web-based social networks with
socialtrust. In Proceeding of the 17th conference on
WWW. Beijing, China: ACM, pp. 1163-1164.
Dwyer, C., Hiltz, S. & Passerini, K., 2007. Trust and
privacy concern within social networking sites: A
comparison of Facebook and MySpace. In
Proceedings of the Thirteenth Americas Conference
on Information Systems. Keystone, CO, USA.
Golbeck, J., 2008. Computing with Social Trust 1 ed.,
Springer.
Golbeck, J. & Hendler, J., 2006. Inferring binary trust
relationships in Web-based social networks. ACM
Trans. Internet Technol., 6(4), pp.497-529.
Guha, R. et al., 2004. Propagation of trust and distrust. In
Proceedings of the 13th international conference on
World Wide Web. New York, NY, USA: ACM, pp.
403-412.
Hales, D. & Arteconi, S., 2005. Friends for Free: Self-
Organizing Artificial Social Networks for Trust and
Cooperation.
Huang, J. & Fox, M. S., 2006. An ontology of trust:
formal semantics and transitivity. In Proceedings of
the 8th international conference on Electronic
commerce. Fredericton, New Brunswick, Canada:
ACM, pp. 259-270.
Kamvar, S. D., Schlosser, M. T. & Garcia-Molina, H.,
2003. The Eigentrust algorithm for reputation
management in P2P networks. In Proceedings of the
12th international conference on World Wide Web.
Budapest, Hungary: ACM, pp. 640-651.
Mayer, R., Davis, J. & Schoorman, D., 1995. An
Integrative Model of Organizational Trust. The
Academy of Management Review, 20(3), pp.734, 709.
Meo, P. D. et al., 2009. Finding reliable users and social
networks in a social internetworking system. In
Proceedings of the 2009 International Database
Engineering & Applications Symposium. Cetraro -
Calabria, Italy: ACM, pp. 173-181.
Moghaddam, S. et al., 2009. FeedbackTrust: using
feedback effects in trust-based recommendation
systems. In Proceedings of the third ACM conference
on Recommender systems. New York, New York,
USA: ACM, pp. 269-272.
Pavlovic, D., 2009. Dynamics, Robustness and Fragility of
Trust. In Formal Aspects in Security and Trust: 5th
International Workshop, FAST 2008 Malaga, Spain,
October 9-10, 2008 Revised Selected Papers.
Springer-Verlag, pp. 97-113.
Richardson, M., Agrawal, R. & Domingos, P., 2003. Trust
Management for the Semantic Web. In The
SemanticWeb - ISWC 2003. pp. 351-368.
Walter, F., Battiston, S. & Schweitzer, F., 2008. A model
of a trust-based recommendation system on a social
network. Autonomous Agents and Multi-Agent
Systems, 16(1), pp.57-74.
Walter, F. E., Battiston, S. & Schweitzer, F., 2009.
Personalised and dynamic trust in social networks. In
Proceedings of the third ACM conference on
Recommender systems. New York, New York, USA:
ACM, pp. 197-204.
Wennerberg, P. O. & Oellinger, T., 2006. Ontology Based
Modelling and Visualization of Social Networks for
the Web: Discovering Security Related Information
from Online News Sites.
Yan, Z. & Cofta, P., 2004. A Mechanism for Trust
Sustainability Among Trusted Computing Platforms.
In Trust and Privacy in Digital Business. pp. 11-19.
Yan, Z. & Holtmanns, S., 2007. Trust Modeling and
Management: from Social Trust to Digital Trust. book
chapter of Computer Security, Privacy and Politics:
Current Issues, Challenges and Solutions.
Ziegler, C. & Lausen, G., 2004. Spreading Activation
Models for Trust Propagation. In Proceedings of the
2004 IEEE International Conference on e-Technology,
e-Commerce and e-Service (EEE'04). IEEE Computer
Society, pp. 83-97.
WEBIST 2011 - 7th International Conference on Web Information Systems and Technologies
534