A COMPUTATIONAL DYNAMIC TRUST MODEL FOR USER AUTHORIZATION

A COMPUTATIONAL DYNAMIC TRUST MODEL FOR USER AUTHORIZATION
Abstract—Development of authorization mechanisms for secure information
access by a large community of users in an open environment is an important
problem in the ever-growing Internet world. In this paper we propose a
computational dynamic trust model for user authorization, rooted in findings from
social science. Unlike most existing computational trust models, this model
distinguishes trusting belief in integrity from that in competence in different
contexts and accounts for subjectivity in the evaluation of a particular trustee by
different trusters. Simulation studies were conducted to compare the performance
of the proposed integrity belief model with other trust models from the literature
for different user behavior patterns. Experiments show that the proposed model
achieves higher performance than other models especially in predicting the
behavior of unstable users.
EXISTING SYSTEM:
The social trust model, which guides the design of the computational model in this
paper, was proposed by McKnight and Chervany after surveying more than 60
papers across a wide range of disciplines. It has been validated via empirical study.
This model defines five conceptual trust types: trusting behavior, trusting intention,

trusting belief, institution-based trust, and disposition to trust. Trusting behavior is
an action that increases a truster’s risk or makes the truster vulnerable to the
trustee. Trusting intention indicates that a truster is willing to engage in trusting
behaviors with the trustee. A trusting intention implies a trust decision and leads to
a trusting behavior. Two subtypes of trusting intention are: 1) Willingness to
depend: the volitional preparedness to make oneself vulnerable to the trustee. 2)
Subjective probability of depending: the likelihood that a truster will depend on a
trustee. Trusting belief is a truster’s subjective belief in the fact that a trustee has
attributes beneficial to the truster. The following are the four attributes used most
often: 1) Competence: a trustee has the ability or expertise to perform certain tasks.
2) Benevolence: a trustee cares about a truster’s interests. 3) Integrity: a trustee is
honest and keeps commitments. 4) Predictability: a trustee’s actions are
sufficiently consistent. Institution-based trust is the belief that proper structural
conditions are in place to enhance the probability of achieving a successful
outcome. Two subtypes of institution-based trust are: 1) Structural assurance: the
belief that structures deployed promote positive outcomes. Structures include
guarantees, regulations, promises etc. 2) Situational normality: the belief that the
properly ordered environments facilitate success outcomes.
PROPOSED SYSTEM:

In this work, we propose a computational dynamic trust model for user
authorization. Mechanisms for building trusting belief using the first-hand (direct
experience) as well as second-hand information (recommendation and reputation)
are integrated into the model. The contributions of the model to computational trust
literature are:
_ The model is rooted in findings from social science, i.e., it provides automated
trust management that mimics trusting behaviors in the society, bringing trust
computation for the digital world closer to the evaluation of trust in the real world.
_ Unlike other trust models in the literature, the proposed model accounts for
different types of trust. Specifically, it distinguishes trusting belief in integrity from
that in competence.
_ The model takes into account the subjectivity of trust ratings by different entities,
and introduces a mechanism to eliminate the impact of subjectivity in reputation
aggregation.
Module1
Trust model
The trust model we propose in this paper distinguishes integrity trust from
competence trust. Competence trust is the trusting belief in a trustee’s ability or

expertise to perform certain tasks in a specific situation. Integrity trust is the belief
that a trustee is honest and acts in favor of the truster. Integrity and benevolence in
social trust models are combined together. Predictability is attached to a
competence or integrity belief as a secondary measure. The elements of the model
environment, as seen in Fig. 1, include two main types of actors, namely trusters
and trustees, a database of trust information, and different contexts, which depend
on the concerns of a truster and the competence of a trustee. For the online auction
site example in Section 1, let us assume that buyer B needs to decide whether to
authorize seller S to charge his credit card for an item I (authorize access to his
credit card/contactinformation). The elements of the model in this case are:
_ Trusters are the buyers registered to the auction site.
_ Trustees are the sellers registered to the auction site.
_ The context states how important for B the shipping, packaging and item quality
competences of S for item I are. It also states how important for B the integrity of
S is for this transaction.
_ B can gather trust information about S from a database maintained by the site or
a trusted third party. This information includes the ratings that S received from
buyers (including B’s previous ratings, if any) for competence in shipping,
packaging and quality of I as well as S’s integrity. It also includes the ratings of
buyers (including B) for sellers other than S in different contexts and ratings of S

for different items. Trust evaluation is recorded in the database when a buyer rates
a transaction with a seller on the site.
Module 2
Trusting belief
Beliefs in two attributes, competence and integrity, are separated. Context
identifier is included for competence belief. Values of both beliefs are real
numbers ranging from 0 to 1. The higher the value, the more a truster believes in a
trustee. Predictability is a positive real number. It characterizes the goodness of
belief formed. The smaller the predictability or uncertainty, the more confident a
truster is about the associated belief value. Both the variability of a trustee’s
behaviors and lack of observations negatively impact the goodness of belief
formed. iNumber in competence belief records the number of observations
accumulated. Trusting beliefs can be classified into initial and continuous trust.
Initial trust is the belief established before a truster t1 interacts with a trustee u1.
Continuous trust is the belief after t1 has had appropriate direct experience with u1.
Module 3
Global and Local Profiles
Each truster t1 has one global profile. The profile contains: (1) t1’s priori integrity
and competence trusting belief; (2) method preference policies; (3) imprecision

handling policies; (4) uncertainty handling policies; (5) parameters needed by
trust-building methods. t1 can have one local profile for each context. Local
profiles have a similar structure as global profiles. The content in a local profile
overrides that in the global one. Fig. 4 shows the definition of global and local
profiles. As aforementioned, method preference policies, defined as Preference
Policy, are to extend the partial order _ to a total order. Therefore, no two methods
have the same priority. iCompetence and cCompetence are used when building
initial and continuous competence trust respectively. iCompetence consists of four
parts corresponding to the four scenarios to build initial competence trust.
iIntegrity and cIntegrity are for establishing integrity trusting belief. Relationships
are separately defined on each ambiguous priority set.
Module 4
Integrity belief
Integrity may change fast with time. Furthermore, it possesses a meaningful trend.
Evaluation of integrity belief is based on two assumptions:
_ We assume integrity of a trustee is consistent in all contexts.
_ Integrity belief may vary largely with time. An example is a user behaving well
until he reaches a high trust value and then starts committing fraud. We used mean
as an estimator for competence belief as it is relatively steady with time. For
integrity belief, this assumption is excluded. When behavior patterns are present,

the mean is no more a good estimator. The similarity between a rating sequence
and a time series inspires us to adopt the method of double exponential moothing
[7] to predict the next rating based on a previous rating sequence. Let ri denote the
ith rating and fiþ1 denote the forecast value of riþ1 after observing the rating
sequence r1; . . . ; ri.
CONCLUSION
In this paper we presented a dynamic computational trust model for user
authorization. This model is rooted in findings from social science, and is not
limited to trusting belief as most computational methods are. We presented a
representation of context and functions that relate different contexts, enabling
building of trusting belief using crosscontext information. The proposed dynamic
trust model enables automated trust management that mimics trusting behaviors in
society, such as selecting a corporate partner, forming a coalition, or choosing
negotiation protocols or strategies in e-commerce. The formalization of trust helps
in designing algorithms to choose reliable resources in peer-to-peer systems,
developing secure protocols for ad hoc networks and detecting deceptive agents in
a virtual community. Experiments in a simulated trust environment show that the
proposed integrity trust model performs better than other major trust models in
predicting the behavior of users whose actions change based on certain patterns
over time.

REFERENCES
[1] G.R. Barnes and P.B. Cerrito, “A Mathematical Model for Interpersonal
Relationships in Social Networks,” Social Networks, vol. 20, no. 2, pp. 179-196,
1998.
[2] R. Brent, Algorithms for Minimization without Derivatives. Prentice- Hall,
1973.
[3] A. Das and M.M. Islam, “SecuredTrust: A Dynamic Trust Computation Model
for Secured Communication in Multiagent Systems,” IEEE Trans. Dependable and
Secure Computing, vol. 9, no. 2, pp. 261- 274, Mar./Apr. 2012.
[4] C. Dellarocas, “Immunizing Online Reputation Reporting Systems against
Unfair Ratings and Discriminatory Behavior,” Proc. Second ACM Conf.
Electronic Commerce, pp. 150-157, 2000.
[5] L. Fan, “A Grid Authorization Mechanism with Dynamic Role Based on Trust
Model,” J. Computational Information Systems, vol. 8, no. 12, pp. 5077-5084,
2012.
[6] T. Grandison and M. Sloman, “A Survey of Trust in Internet Applications,”
IEEE Comm. Surveys, vol. 3, no. 4, pp. 2-16, Fourth Quarter 2000.
[7] J.D.Hamilton, TimeSeriesAnalysis. PrincetonUniversity Press, 1994.

[8] J. Hu, Q. Wu, and B. Zhou, “FCTrust: A Robust and Efficient Feedback
Credibility-Based Distributed P2P Trust Model,” Proc. IEEE Ninth Int’l Conf.
Young Computer Scientists (ICYCS ‘08), pp. 1963- 1968, 2008.
[9] B. Lang, “A Computational Trust Model for Access Control in P2P,” Science
China Information Sciences, vol. 53, no. 5, pp. 896-910, May 2010.
[10] C. Liu and L. Liu, “A Trust Evaluation Model for Dynamic Authorization,”
Proc. Int’l Conf. Computational Intelligence and Software Eng. (CiSE), pp. 1-4,
2010.

A COMPUTATIONAL DYNAMIC TRUST MODEL FOR USER AUTHORIZATION

More Related Content

Viewers also liked (14)

Similar to A COMPUTATIONAL DYNAMIC TRUST MODEL FOR USER AUTHORIZATION (20)

Recently uploaded (20)

A COMPUTATIONAL DYNAMIC TRUST MODEL FOR USER AUTHORIZATION