ArticlePDF Available

Dependency Network-based Trust Management for Context-Aware Web Services

Authors:

Abstract

A major challenge for Web services trust management is the continuous changing running environments. In such context environments, Web services have difficulty guaranteeing the quality of service. In this paper, we propose a trust management model for context-aware Web services based on the Dependency Network. The novelty of our trust model lies in leveraging the cyclic dependency relations among the quality of service metrics and context environments. Experiments conducted on a real-life dataset demonstrate the capability of our trust model compared to Bayesian Network-based trust model.
ScienceDirect
Available online at www.sciencedirect.com
Procedia Computer Science 151 (2019) 583–590
1877-0509 © 2019 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the Conference Program Chairs.
10.1016/j.procs.2019.04.078
10.1016/j.procs.2019.04.078 1877-0509
Available online at www.sciencedirect.com
Procedia Computer Science 00 (2018) 000–000
www.elsevier.com/locate/procedia
The 10th International Conference on Ambient Systems, Networks and Technologies (ANT)
April 29 - May 2, 2019, Leuven, Belgium
Dependency Network-based Trust Management for Context-Aware
Web Services
Afaf Mousaa, Jamal Bentahara, Omar Alamb,
aConcordia Institute for Information Systems Engineering, Concordia University, Montreal, Canada
bDepartment of Computing and Information Systems, Trent University, Peterborough, Canada
Abstract
A major challenge for Web services trust management is the continuous changing running environments. In such context envi-
ronments, Web services have diculty guaranteeing the quality of service. In this paper, we propose a trust management model
for context-aware Web services based on the Dependency Network. The novelty of our trust model lies in leveraging the cyclic
dependency relations among the quality of service metrics and context environments. Experiments conducted on a real-life dataset
demonstrate the capability of our trust model compared to Bayesian Network-based trust model.
c
2018 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the Conference Program Chairs.
Keywords: Dependency Network; Trust; Context-aware services
1. Introduction
Nowadays, online services provision has been implemented through self-contained applications known as Web
services. In addition, Web services facilitate loosely-coupled distributed business integration which promotes the
research to attract increasing attention.
Web services run in context environments where service-level agreement (SLA), which is a commitment between
a service provider and a service user to define the level of service expected from the provider [17], could be violated.
In other words, such Web services encounter significant challenges to meet SLA and guarantee the quality of service
(QoS) due to the constantly changing environment. To face this dynamism and volatility, certain types of services,
namely context-aware services, are designed to continue oering their functionalists without compromising their
operational and financial eciencies.
Corresponding author.
E-mail address: afafmousa11@gmail.com
1877-0509 c
2018 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the Conference Program Chairs.
Available online at www.sciencedirect.com
Procedia Computer Science 00 (2018) 000–000
www.elsevier.com/locate/procedia
The 10th International Conference on Ambient Systems, Networks and Technologies (ANT)
April 29 - May 2, 2019, Leuven, Belgium
Dependency Network-based Trust Management for Context-Aware
Web Services
Afaf Mousaa, Jamal Bentahara, Omar Alamb,
aConcordia Institute for Information Systems Engineering, Concordia University, Montreal, Canada
bDepartment of Computing and Information Systems, Trent University, Peterborough, Canada
Abstract
A major challenge for Web services trust management is the continuous changing running environments. In such context envi-
ronments, Web services have diculty guaranteeing the quality of service. In this paper, we propose a trust management model
for context-aware Web services based on the Dependency Network. The novelty of our trust model lies in leveraging the cyclic
dependency relations among the quality of service metrics and context environments. Experiments conducted on a real-life dataset
demonstrate the capability of our trust model compared to Bayesian Network-based trust model.
c
2018 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the Conference Program Chairs.
Keywords: Dependency Network; Trust; Context-aware services
1. Introduction
Nowadays, online services provision has been implemented through self-contained applications known as Web
services. In addition, Web services facilitate loosely-coupled distributed business integration which promotes the
research to attract increasing attention.
Web services run in context environments where service-level agreement (SLA), which is a commitment between
a service provider and a service user to define the level of service expected from the provider [17], could be violated.
In other words, such Web services encounter significant challenges to meet SLA and guarantee the quality of service
(QoS) due to the constantly changing environment. To face this dynamism and volatility, certain types of services,
namely context-aware services, are designed to continue oering their functionalists without compromising their
operational and financial eciencies.
Corresponding author.
E-mail address: afafmousa11@gmail.com
1877-0509 c
2018 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the Conference Program Chairs.
© 2019 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the Conference Program Chairs.
584 Afaf Mousa et al. / Procedia Computer Science 151 (2019) 583–590
2Afaf M. et al. /Procedia Computer Science 00 (2018) 000–000
Fig. 1: Bayesian Network of Web Service Trust
Fig. 2: Dependency Network of Web Service Trust
In this paper, we are concerned with trust management for context-aware Web services. Trust is the confidence in
others behaviors [15]. In [1], the authors classify trust research into four areas: (1) policy-based trust, (2) reputation-
based trust, (3) general models of trust, and trust in information resources. Moreover, the authors in [13] classify the
dierent computational trust models into four categories (1) feedback-based models, (2) statistics-based models, (3)
fuzzy-logic-based models, and (4) datamining-based.
This work extends reputation-based trust learning models using machine learning techniques to predict the be-
havior of a web service based on QoS data history of direct interactions between web services and users given the
environmental context information capitalizing on Dependency Network. Our approach uses trust to predict the prob-
ability of delivering a satisfactory service level to the service user under context variables.We view context as the
information about the present environment [10]. Accordingly, the context variable is the environmental condition,
such as price-awareness, that aects the behavior of the service [14].
Most current statistical trust models, particularly Bayesian Network-based models [9,11,16,7], describe the
relations among QoS metrics to estimate the probability of delivering a satisfactory QoS values. However, they ignore
the context environments of Web services, thus these model suer from prediction inaccuracy. On the other hand,
the approaches of [14,2,12], which consider the context environments, argue that the QoS degradation is a result
of context environments while they neglect the fact that this degradation could force some context variables to be
activated and the dependency relations among context variables. For example, price-awareness context variable, user-
side, is activated by delivering low QoS values to the user which on its role will activate profit-awareness context
variable, provider-side, leading to more QoS degradation. In other words, there are cyclic dependency relations among
QoS values and context variables.
Dependency Network is a graphical model that approximates the full joint probability distribution over the cor-
responding domain by means of Gibbs sampling [6]. Dependency Network is similar to Bayesian Network, but its
graphical structure is not required to be acyclic. Therefore, the Dependency Network can represent the mutual de-
pendency or cycle among domain variables. The graphical structure is a directed graph where each node represents a
variable in the problem domain and contains a conditional probability given its parents in the network. Edges represent
the global constraints among nodes and their absence mean the independence of the nodes.
Furthermore, the fact that the size of a conditional probability table grows exponentially with the number of parents
of a node limits the scalability of Bayesian Network. Unlike Bayesian Network, Dependency Network is able to learn
from complete data due to the approximate method described in [6]. Namely, the local distribution for each node in the
network can be estimated using any classification or regression algorithms. Since these algorithms have been scaled
up for large data sets, this improves the scalability of Dependency Network in large scale environments.
1.1. Motivation
Context-aware Web services run in context environments where there are many context variables that could aect
their trust with regard to delivering satisfying QoS values. Therefore, trust of context-aware Web service is viewed as
the probability of providing a satisfactory QoS value under context variables. This paper investigates the problem of
trust management for context-aware Web service. Existing approaches did not consider the cyclic dependency rela-
tions that exist among QoS values and context variables. As argued in [8], ignoring dependency relations overestimates
Afaf Mousa et al. / Procedia Computer Science 151 (2019) 583–590 585
Afaf M. et al. /Procedia Computer Science 00 (2018) 000–000 3
the computed trust values. This trust overshoot is detrimental to the stability of a trust model [4] which motivates this
work.
For simplicity, we consider price-awareness and profit-awareness as context variables. Price-awareness is a user-
side context variable whereas profit-awareness is a provider-side context variable. In other words, a price-aware user
pays for the received QoS values of the service [18]. Hence, QoS degradation may change the behavior of the user
to pay less. On the other hand, profit-aware provider delivers high QoS values for high price [18], and degrade the
delivered QoS for low prices. Accordingly, QoS degradation, which is caused by the activation of profit-awareness
variable, activates price-awareness variable. The latter activates again profit-awareness variable leading to more QoS
degradation. This is what we mean by cyclic relations.
We explore WS-DREAM dataset [19] that includes dierent QoS measurements (response time and throughput)
from 142 service users on 4,500 Web services over 64 consecutive time slices at a 15-minute interval. For each user,
truster, we create a Dependency Network and a Bayesian Network for each Web service, trustee, used by the user.
For example, Figure 1and Figure 2illustrate the Bayesian Network and the Dependency Network using WinMine
Toolkit [3] for the web service ws0used by the user s0, respectively. The value of price-awareness context variable
upon service completion is calculated as the ratio of the actual delivered QoS and the asked price to the promised QoS.
While the value of profit-awareness context value is the dierence between the asked price and the actual price. As it
is illustrated by the figures, the Dependency Network shows cyclic relations among QoS metrics and context variables
which are ignored by the Bayesian Network. The existence of cyclic relations proves that QoS degradation could lead
to the activation of some context variables and the dependence of context variables. Furthermore, the dependencies
among QoS metrics add to the complexity of the task. For example, the throughput of service is strongly dependent on
its response time and vice versa, i.e. decreasing the response time of service increases the throughput and increasing
throughput of service decreases the response time.
Those cyclic dependency relations aect the service trust estimation. Ignoring some relations could lead to over-
estimated confidence in the computed trust values. Unlike current approaches, this work considers the dierent de-
pendency relations among context variables and QoS metric. To this end, this work adopts Dependency Network for
its ability to capture cyclic relations than Bayesian Network, the commonly used machine learning approach.
In this paper, we propose Dependency Network-based service trust model which improves trust perdition accuracy.
To the best of our knowledge, no existing service trust model considers these cyclic dependency relations. We compare
the performance of the our model with Bayesian Network-based model and we find that out model performs better as
data density increases.
The remaining of this paper is structured as follows. Section 2 discusses the theoretical background relevant to the
concept developed in the paper. Section 3 describes our proposed service trust model in detail. Experimental settings
and results are discussed in Sections 4. Finally, Section 5 provides the conclusion.
2. Related Work
In [16], a Bayesian network approach is proposed to combine the trust of P2P-style interactions between agents.
The authors demonstrate that the exchange of information about trust increases the performance of the network. In
[11], a trust model based on Bayesian Network is proposed that integrates both subjective and objective trust sources.
Based on these sources, the final trust value is calculated. In [9], Bayesian network query is proposed to select the
service candidates resulting highest global trust of a composite web service. In [7], the authors use a multinomial
generalized Dirichlet distribution in learning Bayesian Networks to model QoS and compute QoS-based trust values.
The authors learn and model the composition structure of composite services.
To deal with the context environments, [12] presents a framework to estimate QoS. Their framework involves
three steps. It begins building a Bayesian Network model to represent QoS capabilities of the service. Then, the
model is trained with feedback from dierent sources to learn the unknown parameters for the service. Finally, QoS
is estimated by making probabilistic inference on the basis of certain context variables. In [2], the authors propose
a Bayesian network-based trust approach to evaluate trust from user satisfaction experiences. The authors represent
user satisfaction experience as a binary value, with 1 indicating satisfied and 0 not satisfied, which is the outcome of a
Bernoulli trial. However, this work is based on the available social networks as prior knowledge, which is not granted.
The focus of their work is SOA-Based IoT. In [14], a context-aware approach for trust management of IoT service
586 Afaf Mousa et al. / Procedia Computer Science 151 (2019) 583–590
4Afaf M. et al. /Procedia Computer Science 00 (2018) 000–000
network is proposed. The proposed model in this work predicts trustworthiness of a service provider based on its
behavior in proving QoS in response to context variables. However, the authors neglect the fact that QoS degradation
could force some context variables to be activated. We extend these approaches by considering the relation among
QoS metrics and context variables to enhance the estimation of trust values of Web services.
3. Trust Management for Web Services
In the rest of this paper, we use the following syntactic conventions. A variable is denoted by an upper-case token
(e.g. Xi) and its value is denoted by a lower case token (e.g. xi). Bold-face capitalized token (e.g. Pai) is used to
denote a set and its instantiation is denoted by bold-face lower-case token (e.g. pai). For models and graphs, we use
calligraphic tokens (e.g. G).
To estimate the service trust value for each constituent, the QoS values and context variables’ values are stored
either 1,2 or 3 to represent a high, medium or low outcomes respectively. For example, the record of a web service
after delivering medium response time and high availability under high price context and high-profit context could be
stored as {2, 1, 1, 1}respectively.
3.1. Service Direct Trust
We define Web service as a set of finite variables X=(X1, ..., Xn) corresponding to QoS metrics and context variables.
Particularly, we consider response time (XRT ) and availability XAV as QoS metrics, and price XPC and profit XPF as
context variables. Therefore, the corresponding web service is X=(XRT ,XAV,XPC ,XPF ). For clarity of notation, we
use XRT ,XAV ,XPC and XPF as illustrative examples.
The corresponding Dependency Network Dfor X is (G,P) where Gis a cyclic directed graph, each node represents
Xis.t. XiX, and P=(p(x1|x\x1), ..., p(xn|x\xn)) is the set of local conditional probability distributions.
Dis constructed by learning the local conditional distribution of each XiX. For simplicity, we adopt probabilistic
decision trees to estimate local conditional distribution of each Xi, i.e. p(xi|x\xi)=p(xi|paii) where Paiare the parents
of Xiwhere Pai(X1, ..., Xi1,Xi+1, ...Xn).
3.1.1. Learning a Dependency Network from data
We learn a probabilistic decision tree for each variable XiX, such that Xiis the target variable and X\Xiare the
input variables. For this purpose, we adopt Bayesian score-based structure learning approach [5] and Greedy search
algorithm. In other words, learning a decision-tree structure for Xiis an optimization problem for finding a structure
Twith the maximum score given observed data D. The used score function, namely Bayesian information criterion
(BIC), is defined as follows:
BIC(T|D)=
n
i
L(Xi|Pai)log|D|
2Dim[T] (1)
where the first term is the log-likelihood function of Tand the Dim[T] is the number of independent variables in the
structure.
Greedy search algorithm is adopted to search the structure space and find the one with the highest score. It starts
with T, and for each modified structure T, by edge deletions, edge additions, and/or reversals, a new score is
calculated. Tis accepted if score(T)>score(T). This is repeated until no increase in the score of the tree.
3.1.2. Dependency Network of web service
Dis constructed by learning the local conditional distribution of each XiX. For each Xi, the local distribution
conditioned on its parents Paiis estimated by using probabilistic decision tree, as it is illustrated in the previous
subsection. Afterwards, the overall Dof the web service is constructed as G=(V,E) where edge ei j connects Xjwith
its parent Xi.
Afaf Mousa et al. / Procedia Computer Science 151 (2019) 583–590 587
Afaf M. et al. /Procedia Computer Science 00 (2018) 000–000 5
3.1.3. Probabilistic inference
For probabilistic trust inference of the service, Gibbs sampling is adopted to recover the joint distribution for X,
i.e. p(X).
We begin with some initial value X(k)
The next sample X(k+1) =(x(k+1)
1,x(k+1)
2,˙..., x(k+1)
n) is obtained by sampling each x(k+1)
iX(k+1) by updating it
according to the distribution specified by p(x(k+1)
i|x(k+1)
1,...,x(k+1)
i1,x(k)
i+1,...,x(k)
n) as follows:
X(k+1) =
x(k+1)
1p(x1|x(k)
2,x(k)
3, ..., x(k)
n)
x(k+1)
2p(x2|x(k+1)
1,x(k)
3, ..., x(k)
n)
x(k+1)
3p(x3|x(k+1)
1,x(k+1)
2, ..., x(k)
n)
x(k+1)
np(xn|x(k+1)
1,x(k+1)
2, ..., x(k+1)
n1)
Repeat N times.
For particular instances, e.g. x2, we can compute p(x1|x2) by fixing the value of X2to x2during the iterations. The
advantage of this is the ability to answer probabilistic queries of the form p(x1|x2), where X1,X2X. For example,
we can infer about the probability of getting high response time given low price as p(XRT =1|XPC =3).
Accordingly, the trust of truster utoward the trustee Xis computed as:
trustu(X)=1
N
N
k=1
Wk×(X(k)) (2)
where (X(k))=n
i=1pi×xk
i, s.t. piis the user’s preferences to the dierent used metrics. In addition, according to
the law of large numbers, we define an ascending weight for the samples as follows:
Wk=k
Nk(3)
4. Experiments
The evaluation of our approach was carried out on the dataset described in Subsection 1.2.For the implementation
of Dependency Network and Bayesian Network, we use WinMine Toolkit which allows us to build the required
statistical models from the dataset in a 64-bit Windows 7 environment on a machine equipped with an Intel Core
i7-4790 CPU 3.60 GHz Processor and 16 GB RAM.
4.1. Direct trust prediction accuracy
In this section, we explore the eciency of the service direct trust prediction. In this experiment, we vary the
training data density from 100 to 500 in order to estimate the accuracy of the learned Dependency Network-based
service trust model. To this end, the log score [6] is used to measure the prediction accuracy of the model by Equation
4on the corresponding test set (x1, ..., xS). This score reports the average of log posterior probability values across the
test set. This means that on average, the log probability that each output variable assigns to the given value in the test
case, given the values of all other input variables, i.e. variables that are used only to predict output variables.
S core(x1, ..., xS|model)=S
i=1log2p(xi|model)
nS (4)
where for each Gibbs sampler invoked to determine p(xi|model), we average 5000 iteration.
Figure 3depicts the evolution of Dependency Network prediction accuracy where the xaxis represents the density
of the training data and the yaxis indicates the accuracy of the learned model on the test data. Furthermore, we vary the
588 Afaf Mousa et al. / Procedia Computer Science 151 (2019) 583–590
6Afaf M. et al. /Procedia Computer Science 00 (2018) 000–000
Fig. 3: Dependency Network performance
(a)
(b)
Fig. 4: DN vs BN . (a) prediction accuracy of the models(b) computational eciency for learning the models.
number of input variables that are used to predict the output variables from 10 to 50. As it is illustrated by the figure,
the prediction accuracy is related to the density of the training data. This result indicates that the prediction accuracy
of Dependency Network increases as more data accumulated to the model. In addition, the prediction accuracy of
Dependency Network increases as more information is accumulated for the evidence, .i.e input variables.
4.2. Dependency Network vs Bayesian Network
This experiment compares the performance between Dependency Network-based trust model (DN) and Bayesian
Network-based trust model (BN). First, we compare the prediction accuracy of the models. Then, we compare the
computational eciency for learning the models.
For learning Bayesian Network-based trust model, we use the same parameter and structure priors as used in the
learning of Dependency Network. The joint probability from a Bayesian Network is determined using the law of total
probability.
Figure 4a depicts the dierence of prediction accuracy of the models. As shown in the figure, Bayesian Network-
based trust model shows higher prediction accuracy for small training data sets than Dependency Network-based trust
model. This is attributed to the fact that Bayesian Network utilizes exact inference algorithm which results in better
quality than Dependency Network. However, as more data accumulated to the models, contrary to Bayesian Network-
based model, the prediction accuracy of the Dependency Network-based model increases. The decrease in prediction
Afaf Mousa et al. / Procedia Computer Science 151 (2019) 583–590 589
Afaf M. et al. /Procedia Computer Science 00 (2018) 000–000 7
accuracy of the Bayesian Network-based model proves the limitation of Bayesian Network in capturing dependency
relations that deteriorate its prediction accuracy. This result implies that the proposed Dependency Network-based
trust model is more suitable for large and complex environment. Moreover, it proves its ability to capture dependency
relations which improves its prediction accuracy.
Afterward, we measure the computational eciency of the models on a larger scale. We measure computation time
for learning the models. Figure plots the relation between the learning time, yaxis, of both models and the training
data density, xaxis where we fix the number of input variable to 50 . As shown by Figure 4b , the proposed Depen-
dency Network-based model is superior to Bayesian Network-based model in learning time. This is due to the fact that
Bayesian Network-based model uses an exact algorithm which is known as NP-complete, whereas the proposed De-
pendency Network-based model uses an approximate algorithm. This proves the robustness of the proposed approach
in the large-scale environment.
5. Conclusion
We proposed a trust model for managing context-aware Web services based on the Dependency Network. The
proposed model is capable of estimating the service trust, i.e. the probability of delivering a satisfactory service
level, under context environments. To enhance the prediction accuracy, we leveraged dependency relation among QoS
metrics and context variables. Using a real dataset, we explored new relations among them than the used by the state-
of-the-art. Our results demonstrate the out-performance of the proposed approach compared to Bayesian Network-
based trust model. DN significantly outperforms BN by reducing the learning time and improving the prediction
accuracy. It is worth mentioning that this out-performance takes eect in a large-scale setting. In future work, we
will extend our model by considering more trust sources such as objective trust. We also intend to scale the model to
handle distributed composite Web services in open dynamic environments where they face malicious users in addition
to context variables.
References
[1] Artz, D., Gil, Y., 2007. A survey of trust in computer science and the semantic web. Journal of Web Semantics 5, 58 –
71. URL: http://www.sciencedirect.com/science/article/pii/S1570826807000133, doi:https://doi.org/10.1016/j.
websem.2007.03.002. software Engineering and the Semantic Web.
[2] Chen, I., Guo, J., Bao, F., 2016. Trust management for soa-based iot and its application to service composition. IEEE Transactions on Services
Computing 9, 482–495. doi:10.1109/TSC.2014.2365797.
[3] Chickering, M., 2002. The winmine toolkit. URL: https://www.microsoft.com/en-us/research/publication/
the-winmine- toolkit-2/.
[4] Cho, J.H., Swami, A., Chen, I.R., 2012. Modeling and analysis of trust management with trust chain optimization in mobile ad hoc net-
works. Journal of Network and Computer Applications 35, 1001 – 1012. URL: http://www.sciencedirect.com/science/article/
pii/S1084804511000701, doi:https://doi.org/10.1016/j.jnca.2011.03.016. special Issue on Trusted Computing and Communi-
cations.
[5] Friedman, N., Goldszmidt, M., 1998. Learning Bayesian Networks with Local Structure. Springer Netherlands, Dordrecht. pp. 421–459. URL:
https://doi.org/10.1007/978-94- 011-5014- 9_15, doi:10.1007/978- 94-011- 5014- 9_15.
[6] Heckerman, D., Chickering, D.M., Meek, C., Rounthwaite, R., Kadie, C., 2001. Dependency networks for inference, collaborative filter-
ing, and data visualization. J. Mach. Learn. Res. 1, 49–75. URL: https://doi.org/10.1162/153244301753344614, doi:10.1162/
153244301753344614.
[7] Mehdi, M., Bouguila, N., Bentahar, J., 2013. A qos-based trust approach for service selection and composition via bayesian networks, in: 2013
IEEE 20th International Conference on Web Services, pp. 211–218. doi:10.1109/ICWS.2013.37.
[8] Mehdi, M., Bouguila, N., Bentahar, J., 2016. Trust and reputation of web services through qos correlation lens. IEEE Transactions on Services
Computing 9, 968–981. doi:10.1109/TSC.2015.2426185.
[9] Motallebi, M., Ishikawa, F., Honiden, S., 2012. Trust computation in web service compositions using bayesian networks, in: 2012 IEEE 19th
International Conference on Web Services, pp. 623–625. doi:10.1109/ICWS.2012.100.
[10] Neisse, R., 2012. Trust and privacy management support for context-aware service platforms. Ph.D. thesis. University of Twente. Netherlands.
doi:10.3990/1.9789036533362. sIKS Dissertation Series No. 2012-09.
[11] Nguyen, H.T., Zhao, W., Yang, J., 2010. A trust and reputation model based on bayesian network for web services, in: 2010 IEEE International
Conference on Web Services, pp. 251–258. doi:10.1109/ICWS.2010.36.
[12] Vu, L., Aberer, K., 2009. Towards probabilistic estimation of quality of online services, in: 2009 IEEE International Conference on Web
Services, pp. 99–106. doi:10.1109/ICWS.2009.11.
590 Afaf Mousa et al. / Procedia Computer Science 151 (2019) 583–590
8Afaf M. et al. /Procedia Computer Science 00 (2018) 000–000
[13] Wahab, O.A., Bentahar, J., Otrok, H., Mourad, A., 2015. A survey on trust and reputation models for web services: Single, compos-
ite, and communities. Decision Support Systems 74, 121 – 134. URL: http://www.sciencedirect.com/science/article/pii/
S0167923615000809, doi:https://doi.org/10.1016/j.dss.2015.04.009.
[14] Wang, Y., Chen, I., Cho, J., Swami, A., Lu, Y., Lu, C., Tsai, J.J.P., 2018. Catrust: Context-aware trust management for service-oriented ad hoc
networks. IEEE Transactions on Services Computing 11, 908–921. doi:10.1109/TSC.2016.2587259.
[15] Wang, Y., Singh, M.P., 2007. Formal trust model for multiagent systems, in: Proceedings of the 20th International Joint Conference on Artifical
Intelligence, Morgan Kaufmann Publishers Inc., San Francisco, CA, USA. pp. 1551–1556. URL: http://dl.acm.org/citation.cfm?id=
1625275.1625526.
[16] Wang, Y., Vassileva, J., 2003. Bayesian network-based trust model, in: Proceedings IEEE/WIC International Conference on Web Intelligence
(WI 2003), pp. 372–378. doi:10.1109/WI.2003.1241218.
[17] Wieder, P., Butler, J., Theilmann, W., Yahyapour, R., 2011. Service Level Agreements for Cloud Computing. SpringerLink : B¨
ucher, Springer
New York. URL: https://books.google.ca/books?id=z306GUfFL5gC.
[18] Zeithaml, V.A., 1988. Consumer perceptions of price, quality, and value: A means-end model and synthesis of evidence. Journal of Marketing
52, 2–22. URL: http://www.jstor.org/stable/1251446.
[19] Zheng, Z., Lyu, M.R., 2010. Collaborative reliability prediction for service-oriented systems, in: Proc. IEEE/ACM 32nd Int’l Conf. Software
Engineering (ICSE’10), pp. 35–44.
... Thus, our previous work (Mousa et al., 2019b) estimates the services' subjective trust using the framework of dependency networks which enabled us to consider the 95 cyclic dependency relationships linking QoS metrics to context variables. We find that considering such relations improved the prediction accuracy since ignoring them overestimates the computed values. ...
... The second objective is to detect concept drift and adapt the trust model. Subsequently, this article extends our previous work (Mousa et al., 2019b) to incorporates the 105 dynamic context environment besides the cyclic dependency relations among context variables and QoS metrics. We modify the dependency network formalism to adapt to the changing running environment. ...
... Wang et al. (2018a) observe the behaviour in presence of context variables to estimate the value and level of trust. Mousa et al. (2019b) introduce a dependency network model of trust to capture the cyclic dependency relations that link QoS metrics and context variables. These 190 relations are explored using a real-life dataset. ...
Article
Full-text available
The paper addresses the problem of trust management of cloud, fog and IoT services in dynamically changing environments. The continuous dynamic environment is one of the challenges that trustworthy services management in the cloud, fog and IoT settings faces. Services in such an environmental context have difficulty securing an acceptable quality of service (QoS). This article proposes a trust management framework that establishes service trust by considering the direct trust from the truster (subjective trust), aggregating referrals about the service in a collusion-resistant manner (objective trust), and bootstrapping new services. We introduce a subjective trust model based on the formalism of dependency networks to dynamically predict the provided QoS in response to context environment changes. The proposed approach leverages the dependency relations that exist among the QoS metrics and environmental context variables. The novelty at the subjective trust level lies in considering the dynamic cyclic dependency relations that enhances the prediction accuracy. However, subjective trust based on direct interactions could be insufficient to make the trust estimate credible. Hence, on top of the subjective layer, we propose an objective trust management model resilient to collusion attacks by leveraging the power of mass collaboration among referees. Finally, we propose a bootstrapping mechanism that is resilient to the white-washing attacks by observing the behaviours of newcomer services with no trust resources using the concept of social adoption to estimate their initial trust values. Experiments conducted on real-life and synthetic datasets demonstrate the effectiveness of our approach compared with state-of-the-art approaches. We used the statistical log score to assess the model’s prediction accuracy and employed the estimation error of the objective trust as indicated by referees relative to the one calculated by the multi-round simulation. The ROC (Receiver Operating Characteristic) curves are finally used to measure the accuracy of the classifier used in the proposed trust bootstrapping mechanism in providing accurate initial trust values. The main findings of the paper are around the new trust model of cloud, fog and IoT services considering their dynamically changing environments. The first finding is that the prediction of the provided QoS shows better results when it is dynamic and responds to context environment changes by leveraging the dynamic dependency network linking the QoS metrics and context variables of the environment. The second finding is that the objective trust performs better when it is resilient to collusion attacks by leveraging the power of mass collaboration among referees. The third finding is that the bootstrapping mechanism that observes the behaviours of new comer services with no trust resources using the concept of social adoption to estimate their initial trust values excels by being resilient to white watching attacks.
Article
In the next few years, the number of devices connected to the web will increase dramatically. This has encouraged the development of complexes and intelligent service-based applications using new paradigms and technologies such as edge computing and cloud computing. These applications are mainly built by the interaction between heterogonous web services. Answering end-users’ queries demands that these services collect, integrate and communicate sensitive data, which makes the vulnerability of data a serious issue. Thereby, this issue is a critical problem in the automatic services composition while users are frequently unquieted about the security and control of their sensitive data. To address this problem, our main idea is that avoiding disclosure of sensitive information can be ensured by selecting services where their mutual trust is high while preserving a good quality of service (QoS) as possible. A new Heuristic Trust-aware Web Services Composition Approach (CWS_SMA) is proposed. CWS_SMA aims to: (i) compute an optimal trust and QoS-aware web services composition based on the mathematical coordinate system, (ii) improve the composition response time based on a set of cooperative intelligent agents. Eventually, the simulations results show the effectiveness of the proposed solution.
Book
Service Level Agreements for Cloud Computing provides a unique combination of business-driven application scenarios and advanced research in the area of service-level agreements for Clouds and service-oriented infrastructures. Current state-of-the-art research findings are presented in this book, as well as business-ready solutions applicable to Cloud infrastructures or ERP (Enterprise Resource Planning) environments. Service Level Agreements for Cloud Computing contributes to the various levels of service-level management from the infrastructure over the software to the business layer, including horizontal aspects like service monitoring. This book provides readers with essential information on how to deploy and manage Cloud infrastructures. Case studies are presented at the end of most chapters. Service Level Agreements for Cloud Computing is designed as a reference book for high-end practitioners working in cloud computing, distributed systems and IT services. Advanced-level students focused on computer science will also find this book valuable as a secondary text book or reference.
Article
We propose a context-aware trust management model called CATrust for service-oriented ad hoc networks such as peer-to-peer and Internet of Things networks wherein a node can be a service requester or a service provider. The novelty of our design lies in the use of logistic regression to dynamically estimate trustworthiness of a service provider based on its service behavior patterns in response to context environment changes. We develop a recommendation filtering mechanism to effectively screen out dishonest recommendations even in extremely hostile environments in which the majority recommenders are dishonest. We demonstrate desirable convergence, accuracy, and resiliency properties of CATrust. We also demonstrate that CATrust outperforms contemporary peer-to-peer and Internet of Things trust models in terms of service trust prediction accuracy against collusion recommendation attacks.
Article
A future Internet of Things (IoT) system will connect the physical world into cyberspace everywhere and everything via billions of smart objects. On the one hand, IoT devices are physically connected via communication networks. The service oriented architecture (SOA) can provide interoperability among heterogeneous IoT devices in physical networks. On the other hand, IoT devices are virtually connected via social networks. In this paper we propose adaptive and scalable trust management to support service composition applications in SOA-based IoT systems. We develop a technique based on distributed collaborative filtering to select feedback using similarity rating of friendship, social contact, and community of interest relationships as the filter. Further we develop a novel adaptive filtering technique to determine the best way to combine direct trust and indirect trust dynamically to minimize convergence time and trust estimation bias in the presence of malicious nodes performing opportunistic service and collusion attacks. For scalability, we consider a design by which a capacity-limited node only keeps trust information of a subset of nodes of interest and performs minimum computation to update trust. We demonstrate the effectiveness of our proposed trust management through service composition application scenarios with a comparative performance analysis against EigenTrust and PeerTrust.
Article
In modern distributed systems, service consumers are faced with pools of service providers that offer similar functionalities. This reality renders the selection of web services a challenging task. One popular solution is to base the selection decisions on the web services' nonfunctional requirements depicted by a variety of QoS metrics. In this paper, we present a new approach for solving the web service selection problem; a QoS-aware trust model that leverages the correlation information among various QoS metrics. This model, based on the probability theory, estimates the trustworthiness of web services by exploiting two statistical distributions, namely, Dirichlet and generalized Dirichlet. These distributions represent the outcomes of multiple correlated QoS metrics. The former distribution is employed when the QoS metrics are positively correlated while the latter handles negatively correlated metrics. We also propose an algorithm to aggregate reputation feedback that propagate among the interacting web services. This algorithm deals with malicious feedback and various strategic behavior commonly performed by web services. Experimental results endorse the advantageous capability of our trust model and reputation algorithm compared to the state-of-the-art.
Conference Paper
The concept of trust in web services mainly deals with the degree of belief that a client or a group of clients have over services functioning satisfactorily and providing the expected results. With services being invoked in composition with each other, computing the trust of the composition and selecting services that deliver the highest trust for it becomes a desired goal. In this paper we demonstrate how using Bayesian networks and its supporting queries, we can select the set of services among all candidates that would provide highest global trust.
Conference Paper
Service oriented computing is being increasingly exploited in the architecture of current web applications. The interactions among the deployed web services are becoming vital to accomplish heterogeneous and compound business goals. Service selection and composition are two common tasks which are highly influenced by the quality of these interactions. Thus, assigning web services QoS-based trust scores that are incrementally updated, provides a means to assist both tasks. Then, services with higher trust scores are more likely to be selected than those with smaller ones. They are, additionally, more prone to be incorporated as part of composite services. In this paper, we propose a probabilistic approach based on Bayesian networks (BN) to learn the composition structure of composite services and compute QoS-based trust scores in an online setting. The learning of the BN structure and parameters is based on modeling the QoS, which is represented by the BN's variables, using a multinomial generalized Dirichlet distribution (MGDD). The effectiveness of our approaches is empirically assessed using real and synthetic data. Our experimental results show that MGDD provides a flexible and accurate representation of the QoS. They also prove the capability of the BN approach to learn the composition structure, and further the responsibility of the constituent services in the quality of the composite service even when their QoS is partially observed.
Article
We develop and analyze a trust management protocol for mission-driven group communication systems in mobile ad hoc networks using hierarchical modeling techniques based on stochastic Petri nets. Trust among mobile nodes is crucial for team collaborations with new coalition partners without prior interactions for mission-driven group communication systems in battlefield situations. In addition, ensuring a certain level of trust is also critical for successful mission completion. Our work seeks to identify the optimal length of a trust chain among peers in a trust web that generates the most accurate trust levels without revealing risk based on a tradeoff between trust availability and path reliability over trust space. We define a trust metric for mission-driven group communication systems in mobile ad hoc networks to properly reflect unique characteristics of trust concepts and demonstrate that an optimal trust chain length exists for generating the most accurate trust levels for trust-based collaboration among peers in mobile ad hoc networks while meeting trust availability and path reliability requirements.