Research Article A Novel Trust-Aware Composite Semantic ...downloads.hindawi.com/journals/mpe/2015/928193.pdf · Research Article A Novel Trust-Aware Composite Semantic Web Service

Research ArticleA Novel Trust-Aware Composite Semantic Web ServiceSelection Approach

Denghui Wang1 Hao Huang2 and Changsheng Xie3

1School of Computer Science and Technology Huazhong University of Science amp Technology Wuhan 430074 China2School of Software Engineering Huazhong University of Science amp Technology Wuhan 430074 China3Wuhan National Laboratory for Optoelectronics Wuhan 430074 China

Correspondence should be addressed to Hao Huang thaohusteducn

Received 19 March 2015 Revised 12 June 2015 Accepted 14 June 2015

Academic Editor Bao Rong Chang

Copyright copy 2015 Denghui Wang et alThis is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

The functional characteristics and the nonfunctional properties of service profile always play very important roles in compositesemantic web service selection approach But the credibility of this information cannot be guaranteed This paper established anovel trust degree model of this information Based on this model the trust degrees can be calculated from execution log and userexperience evaluation of candidate web servicesThen the paper proposes a new composite semantic web service selection approachbased on this credible information Finally we present two experiments to prove that the new approach can avoid the influence ofexaggerated and unauthentic information effectively and accurately

1 Introduction

With the rapid development of sophisticated application asingle web service is usually too simple to meet the varioususer requirements Creating new services through web ser-vice composition to provide more complex and on-demandfunctions becomes essential Web service composition isintroduced to resolve the above problem However thereexist a number of available web services providing similar oridentical functional characteristics so users need a selectionapproach that can help them choose the best composite webservice

According to user request the service selection approachshould consider two aspects that include features and qual-ity of candidate service The feature of web service canbe described by inputs outputs precondition and effects(IOPE) in semantic web service model IOPE is calledfunctional property In many cases composition techniquesand the related tools exploit IOPE predicates that characterizestructural and semantic services descriptions to generatethe desired compositions [1] And the QoS represent thequality of web service QoS is called nonfunctional propertyRecently most of researchers use QoS as the main parameters

of composite web service selection [2ndash4] However inmost ofthe current web service composition selection methods it isassumed that the service profile offered by different organiza-tion is trusted and authentic But in fact some organizationsof web service are apt to publish unauthentic information forattracting end user To cope with this problem it is necessaryto avoid exaggerated information by dishonest providers inthe selection process In this paper we establish a new trustdegree model of composite semantic web service And basedon this model we propose a novel composite web serviceselection approach which can calculate the trust degree ofQoS and IOPE information from execution log and credibleuser experience evaluation

This paper is organized as follows in Section 2 weemphatically introduce the relatedwork about compositewebservice selection Section 3 introduces the trust degree modelfor getting reliable QoS value and credible IOPE informationSection 4 explains the approach to calculating trust degree ofQoS and IOPE for selecting the best composite web serviceSection 5 proposes the experiments that lead to proving thatthe proposed approach is accurate and effective Section 6gives the final conclusion

Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2015 Article ID 928193 7 pageshttpdxdoiorg1011552015928193

2 Mathematical Problems in Engineering

2 Related Works

The traditional selection and composition of web services relyon themanner of finding themost similar functionalities andthe best nonfunctionalities of web service The QoS is widelyemployed for describing nonfunctionalitiesThere are a lot ofservice selection methods expanded in this regard Reference[5] proposed an algorithm to combine global QoS constraintswith local selection Reference [6] proposed an approachfor web service dynamic composition based on global QoSconstraints decomposition Besides global QoS [7] proposeda distributed optimal scheme based on local QoS Reference[8] presented QoS-GRASP a metaheuristic algorithm forperforming QoS-aware web service composition at runtimeQoS-GRASP is a hybrid approach that combinesGRASPwithPath Relinking References [9 10] applied the neurofuzzydecision making approach in the process of selection andchoice of the most appropriate web service with respectto quality of service criteria The method deals with theimprecision of QoS constraints values And these QoS-basedservice selection methods always assume that the QoS datacoming from service providers and users are effective andtrustworthywhich is actually impossible in real environmentSo these web service selection methods mentioned aboveare not perfect In [11] the authors proposed a novel servicecomposition approach modeling the trust-based servicecomposition as the multidomain scheduling and assignmentproblem using the minimum service resources within acertain time constraint They considered that trust plays apivotal role in service composition approach In [12] theresearchers presented a trustworthy services selection basedon preference selection method that assists users in selectingthe right web service according to their own preferenceThismethod can effectively solve the weaknesses of recommenda-tion systems In [13] we proposed onemethod forweb servicerecommendation based on trust-aware QoS But the methoddid not consider the functionality property Thus this paperbased on the above study proposes a novel composite webservice selectionmethod according to credibleQoS and IOPEinformation

3 Trust Degree Model

31 Semantic Web Service Model

Definition 1 (semantic web service) A semantic web serviceis defined as a quadruple

WS = ⟨SNDes 119865QoS⟩ (1)

where SN is the identifier of the service Des is the generalinformation about the service including service contextssuch as text description ontology definition version con-tractor and the general information being independent of thespecific function 119865 is the functional attribute of a serviceQoS is a set of attribute parameters standing for the qualityof service including some attributes such as cost responsetime successful execution rate and availability The last twoparameters in the quadruple WS are actually related withservice composition

Moreover semantic web servicersquos function attribute canbe demonstrated as a quadruple

119865 = ⟨119868 119874 119875 119864⟩ (2)

where 119868 is the set of servicersquos semantic inputs 119874 is the setof servicersquos semantic outputs 119875 is the precondition 119864 is theeffects of the semantic web service Definitely a semantic webservice can be expressed as follows

WS = ⟨SNDes ⟨119868 119874 119875 119864⟩ QoS⟩ (3)

32 User Requirement Model

Definition 2 (user request) A user request of semantic webservice can be expressed as a quadruple

WS119903= ⟨SN

119903Des119903 119865119903QoS119903⟩ 119865

119903= ⟨119868119903 119874119903 119875119903 119864119903⟩ (4)

The service requesterrsquos information is in Des119903 A service

request can be expressed as follows

WS119903= ⟨SN

119903Des119903 ⟨119868119903 119874119903 119875119903 119864119903⟩ QoS

119903⟩ (5)

User requestmodel contains several requirement indexesand the consumer provides the corresponding constraintcondition for every index But the expression of every indexis different so we need to normalize these indexes Theexpression of user requirement indexes can be divided intotwo categories One is the value type And the other one is theinterval type In order to facilitate comparison with the realvalue we need to transform the interval type to the value typeAt present there is a variety of deterministic methods Inthis paper we use the expanded ordered weighted averagingoperator as determining the mathematical formula

Assume 119865 119877

119899

rarr 119877 if 119865(1198861 1198862 119886119899) = sum120596119895119887119895

wherein 119882 = (1205961 1205962 120596119899)119879 is weighted vector which is

associatedwith119865 and120596119895isin [0 1]Σ120596

119895= 1 and 119887

119895is the 119895th big

value in a group of data (1198861 1198862 119886119899) the119865 function is calledthe 119899-dimensional ordered weighted averaging operator

Assume that 119886 = [119886119871 119886119880] = 119909 | 119886119871 le 119909 le 119886119880 then 119886 isan interval value The 119865 function has the following formula

119891119896(119886

119871

119886

119880

) =

119886

119880

+ 119903119886

119871

119903 + 1

(6)

We transform the interval value to the value type throughthe following formula

UR119896(119886) = 119891

119896(119886

119871

119886

119880

) =

119886

119880

119903 997888rarr 0

119886

119871

+ 119886

119880

2 119903 997888rarr 1

119886

119871

119903 997888rarr infin

(7)

where 119896 is the number of user request indexes UR119896(119886) repre-

sents the normality request of the 119896th user After normalizingevery user requirement index we can get the user expectedvalue set UR

Mathematical Problems in Engineering 3


Definition 3 (trust degree) The trust degree of compositesemantic service can be expressed as follows

TD = ⟨TD119865TDQ⟩ (8)

where TD119865is the trust degree of functional attributes and

TDQ is the trust degree of QoSMoreover the genic QoS parameters can be classified into

two categories One is recordable typeThe execution value ofthis QoS type can be recorded in execution log at run timesuch as response time and successful execution rateTheotherone is unrecordable type The kind of these QoS parameterscannot be recorded in execution log It is only evaluated byuser experience such as cost and availability

TDQ = ⟨TDQRTDQU⟩

TDQR = ⟨TDtimeTDrate⟩ TDQU = ⟨TDcostTDavailable⟩ (9)

34 User Experience Evaluation Model User experiencerepresents the subjective feelings of past users It can beevaluated by different parameters According to [14] userexperience is evaluated by click rate for getting the webservice ranking with PageRank algorithm In [15] they useusage frequency to evaluate user experience Either click rateor usage frequency can only reflect overall impression of webservice It is ambiguity Therefore in the paper we establisheda new user experience evaluation model according to userrequirements The new model consists of the local userratings and the global user ratings The global user ratingspresent the overall impression of web service And the localuser ratings are good complement to the global user ratingswhich evaluate web service from several aspects However theevaluated index of web service is basically provided by serviceprovider or third parties according to their own professionalknowledge In fact the consumer cannot pay attention toall indicators or the professional degree of consumer is notenough to give an accurate evaluation So we use fuzzy logicto represent the user ratings Fuzzy logic is based on fuzzysets that represent vague data with the help of the so-calledmembership functions that represent the degree referred toas membership at which a certain datum belongs to a fuzzydata set

Definition 4 (user experience evaluation model) The fuzzyrepresentation is based on the assumption that the userratings can be expressed as a number in the range [0 1] Thatmeans a user experience evaluation of web service can bepresented by assigning values in the range [0 1] Thus userexperience evaluation of web service is represented as a fuzzyset UE

UE = ⟨UE119892UE119897⟩ UE

119897= 119909 120583

119896(119909) | 119909 isin QoS (10)

whereUE119892is the global user ratings evaluation ofweb service

UE119897is the local user ratings evaluation according to QoS

attributes 120583119896(119909) represents the grade of membership of 119909

evaluation index fromconsumer 119896Themembership function

is a function of ratings We define the membership functionfor 119909 in a fuzzy set defining WS as follows

120583119896(119909) =

0 0 le 119909 le 1205721

1 + [120573 (119909 minus 120572)] 120572 lt 119909 le 100

(11)

4 A Novel Composite Semantic Web ServiceSelection Approach

In this section a novel composite semantic web serviceselection approach is proposed which takes the candidatesrsquotrust degree into consideration Our selection approachmainly consists of the following five parts In Part 1 weanalyze execution log to get the real value of recordableQoS attributes Compared to QoS value which is providedby service provider calculate the trust degree of recordableQoS attributes In Part 2 we compare the QoS constraintof user request with the recordable QoS attribute value tocalculate the pass user satisfaction degree and compare passuser satisfaction degree with user experience evaluation forgetting the credibility of pass user evaluation In Part 3we use credible local user evaluation to calculate the trustdegree of the unrecordable QoS attributes In Part 4 we usecredible global user evaluation to calculate the trust degree offunctional property In Part 5 according to user requirementmodel we consider the credible similar degree of IOPEand the credible QoS attribute parameters to select the bestcomposite web service

41 Analyze Execution Log Traditionally web services areindividually deployed on proprietary infrastructure owed bythe organization which operates and utilizes these servicesWith the increasing adoptions of ldquoPlatform as a servicerdquoparadigm which provides a centralized runtime executionenvironment more and more web services are publishedon centralized runtime execution environments such asIBM Web Sphere Process Server Microsoft Azure ServicesPlatform and Google App Engine Adoptions of such amodefacilitate the monitoring of the services execution to obtainthe execution logs

Once a composite web service is deployed in a runtimeexecution environment the composite web service can beexecuted inmany execution instances of service compositionEach execution instance of service composition is uniquelyidentified with an identifier (id) In each execution instanceevents can be triggered We record the triggered events inthe log using the logging facility provided by the executionenvironment An execution log contains different types ofevents For example service error events are triggered whenservice error occurs Service invocation events indicate thetimeline of a web service execution

We use service invocation events and service error eventsto evaluate the real value of the successful execution rateQRrate is defined as the recorded successful execution rateattribute that can be calculated as follows

QRrate = 1minus119873error

119873invocation (12)


where 119873error is the number of the service error events119873invocation is the total number of the service invocation events

In particular an ENTRY event is triggered when a serviceis invoked An EXIT event occurs when a service completesthe computation and returns results Each event is recordedwith the time of triggering the name of the service whichtriggers the event and the id of the execution instance andthe underlying application So we can get the response timefrom the ENTRY event to the EXIT event

QRtime =1119870

119870

sum

119896=1(119879exit minus119879entry) (13)

where 119870 represents the number of execution results andQRtime represents the recorded response time attribute 119879exitrepresents the triggering time of the EXIT event 119879entry is thetriggering time of the ENTRY event

For recordable QoS attributes the distance betweenexecution results with QoS information describes its trustdegree So the greater distance means the worse credibilityWe calculate the distance to use the following formula

TDQR (119894) =

1 minus1003816100381610038161003816

QR119894minusQ119894

1003816100381610038161003816

Q119894

QR119894lt 2Q119894

0 QR119894gt 2Q119894

(14)

where 119894 represent the number of recordable QoS dimensionsQR119894represent the value of Q

119894in execution log and if QR

119894gt

2Q119894 that means the distance between execution results with

QoS information is so big that the credibility of the serviceprovider is 0

42 User Satisfaction Degree Then we use the gray correla-tion analysis method to get the user satisfaction degree Thegray correlation analysis method can obtain the relationshipof two groups of sequences through calculating their distance[16] So we can get the following formula

119889119896(119894) =

1003816100381610038161003816

UR119894minusQ119894

1003816100381610038161003816

US119896(119894) = 119903 (UR

119894Q119894) =

120588119889max119889119896(119894) + 120588119889max

120588 isin [0 1] (15)

where US119896(119894) represents the user satisfaction degree of the 119894th

recordable user requirement index 119903(UR119894Q119894) represents the

correlation value between the user expected value and realvalue of operation 120588 represents the resolution value 119889maxrepresents the max value of the distance of the user expectedvalue and real value of operation

Finally we compare the user evaluation with user sat-isfaction degree The distance of two values is closer thetrust degree of the user evaluation is higher Assumed TD

119896

represents the trust degree of the 119896th user evaluation We canget the following formula

TD119896(119894) = 1minus

1003816100381610038161003816

US119896(119894) minus 120583

119896(119894)

1003816100381610038161003816

120583119896(119894)

TD119896=

sum

119868

119894=1 TD119896 (119894)119868

(16)

where 119868 is the total number of the recordable QoS attributes

43 Trust Degree of Unrecordable QoS Attribute In this sec-tionwewill use the user experience to calculate the credibilityof unrecordable QoS dimensions We use the user require-ment to evaluate the bygone score of unrecordable QoSdimensions If this user gives the superior limit the bygonescore should be computed using the following formula

BS119896(119895) =

UR119896(119895) minusQ

119895

UR119896(119895) minusQ

119871

+ 06 UR119896(119895) gt Q

119895

0 UR119896(119895) le Q

119895

(1 le 119895 le 119869)

(17)

119876119871represent the minimum value of the QoS dimension in

formula (17) If users give the lower limit the bygone score ofunrecordable QoS dimensions should be computed using thefollowing formula

BS119896(119895) =

Q119895minus UR119896(119895)

Q119898minus UR119896(119895)

+ 06 UR119896(119895) lt Q

119895

0 UR119896(119895) ge Q

119895

(1 le 119895 le 119869)

(18)

Q119898represent the maximum value of the QoS dimension in

formula (18) 119869 represents the number of unrecordable QoSdimensions and 119896 represents the number of users We cancalculate the distance of the bygone score and the credibilityuser comment to get the credibility of the unrecordable QoSdimension as the following formula

TDQU (119895)

=

1119870

119870

sum

119896=1(1minus

1003816100381610038161003816

(TD119896(119895) times 120583

119896(119895)) minus BS

119896(119895)

1003816100381610038161003816

BS119896(119895)

)

(1 le 119895 le 119869)

(19)

44 Trust Degree of IOPE IOPE is the functional property ofweb service So the trust degree of IOPE is due to the globaluser evaluation The following formula helps us to get TD

119865

TD119865=

1119870

119870

sum

119896=1(TD119896timesUE119892)

=

1119870

119870

sum

119896=1(

sum

119869

119895=1 TD119896 (119895)119869

timesUE119892)

(20)

where 119870 is the number of consumers 119869 is the number of theunrecordable QoS attributes UE

119892represents the global user

experience evaluation

45TheNovelWeb Service Selection Approach Asmentionedabove we compute the trust degree of the recordable QoSattributes and the unrecordable QoS attributes respectivelyFinally we can use formula (21) to get the evaluation result ofthe composite semantic web service Among all the candidate


Table 1 User requirement indexes

Indexes Cost Response time Successful execution rate AvailabilityType Unrecordable Recordable Recordable UnrecordableConstraint (weight) 8 (02) 03 s (02) gt80 (03) gt95 (03)

services the service with the highest score of evaluation isselected

WS =WS119865+WSQ

= 119882119865timesTD119865times Sim

119865+

119868

sum

119894=1(119882Q119894 timesTDQR (119894) timesQ119894)

+

119869

sum

119895=1(119882Q119895 timesTDQU (119895) timesQ119895)

(21)

It is supposed that 119868 recordable QoS dimensions and119869 unrecordable QoS dimensions are considered and eachcandidate service is executed119870 timesThat means we have119870pieces of execution results and user experience evaluationsThe credibility of recordable QoS dimensions is computedseparately and each piece of execution log is used once sothe time cost is 119874(119868 times 119870) During computing the credibilityof unrecordable QoS dimensions the time cost is119874(119873times119872)119872 represents the number of web service composition nodesand 119873 represents the number of candidate web services inevery nodeThe complexity of the proposed algorithmwhichcalculates the evaluation of all candidate services is 119874(119873 times

119872 times (119868 times 119870 + 119869))

5 Case Study

In this section we design two experiments to evaluate theperformance of the proposed composite web service selectionmethod The experiments have been performed on a PCpowered by an AMD Quad Core A4 15 GHZ processorequipped with 4GB RAM and a 500GB hard disk and thesoftware environment of the experiments is Win 8 SP1 Java16 Our objective is to prove the availability of our proposedcomposite service selection method For this purpose weadopt the traditional web service selection based on QoS andIOPE evaluation to compare with our approach It does notconsider trust degree of QoS and IOPE information in tradi-tional composite web service selection method According toQoS value and IOPE similar degree it uses formula (22) tosort the candidate web service Consider

WS =WS119865+WSQ = 119882119865 times Sim119865 +

119868

sum

119894=1(119882Q119894 timesQ119894) (22)

The test case is a web service composition that imple-ments a travel planning process It looks for tourist destina-tion books flight ticket and hotel reservation in parallel andfinally invokes a car rental operation Per each of the tasksin the process there are 10 candidate services distributedamong the servers that fulfill the required functionality andoffer different QoS Firstly we give their requirement indexeswhich are presented in Table 1

Table 2 The percentage of unauthentic candidate web service

Tasks Case 1 Case 2 Case 3 Case 4Looking destination 20 40 60 80Booking ticket 30 50 70 80Hotel reservation 20 40 60 80Car rental operation 10 30 50 80

Table 3 The first experiment result

Case 1 Case 2 Case 3 Case 4The selected probabilityTraditional 014 035 052 076Proposed 010 018 024 048

The fitness valueTraditional 091 078 065 040Proposed 1 1 1 08

Two experiments are designed to illustrate the availabilityof the novel proposed approach Every method will execute50 times For the first experiment the traditional algorithmand the proposed algorithm run under four different casesto monitor the influence of two methods as the numberof the unauthentic considered services increases Table 2shows the different proportion of unauthentic candidate webservices The first experiment results are given in Table 3For the second experiment the traditional algorithm andthe proposed algorithm run with four different groups ofexaggerated degree of QoS and IOPE under Case 2 Thesecond experiment results are given in Table 5

Table 3 shows the selected probability of unauthenticcandidate web services and the fitness value The selectedprobability of unauthentic candidate web services can becalculated by the following formula

119875 =

119873119878

4 times 50

(23)

where119873119878is the number of selected unauthentic web services

It can be seen fromTable 3 that the selected probability of pro-posed approach is less than the traditional approach under allthe four cases which illustrates that the proposed approachcan filter unauthentic web service effectively It can also beseen that the selected probability of proposed approach didnot rise as the number of the unauthentic considered servicesincreases which illustrates that the proposed approach isnot influenced by the number of unauthentic web servicesFurthermore Table 3 also shows that the fitness values ofproposed approach under all the four cases are equal toone which means that the selected services set of proposedapproach can satisfy userrsquos constraint condition under each


Table 4 The exaggerated degree of QoS and IOPE

Indexes Cost Response time Successful execution rate Availability IOPEGroup 1 01 02 015 02 02Group 2 03 04 05 05 05Group 3 06 07 07 06 06Group 4 08 08 08 08 08

Table 5 The second experiment result

Group 1 Group 2 Group 3 Group 4The selected probability

Traditional 055 068 080 085Proposed 018 014 006 003


case But the traditional approach cannot fully satisfy theuserrsquos needs under the influence of unauthentic candidateweb services

In the next step we assume that the probability ofunauthentic web service is fixed under Case 2 We increasethe exaggerated degree of QoS and IOPE information tomonitor the influence of two methods As mentioned abovein QoS model the QoS values are four-dimensional costresponse time successful execution rate and availabilityThere services have been registered into service databaseThey executed several times The database establishes theexecution logs to record historical data and collect the userratings to evaluate the user experience Table 4 shows theexaggerated degree

Table 5 shows the second experiment results It can beseen from Table 5 that the exaggerated degree of QoS andIOPE is higher the selected probability of proposed approachis lower but the selected probability of traditional approachis higher which illustrates that the proposed approach cannotinfluence by the exaggerated degree of QoS and IOPE It canalso be seen that the fitness values of proposed approachunder different exaggerated degree are equal to one whichmeans that the selected services set of proposed approach cansatisfy userrsquos constraint condition under different exaggerateddegree But the fitness value of traditional approach is lowerwhen the exaggerated degrees increase which illustrates thatthe traditional approach seems to opt for more unauthenticcandidate web services

6 Conclusion

In this paper the content of the research is to propose anovel trust-aware composite semantic web service selectionapproach In order to filter exaggerated QoS and IOPEinformation this paper established a trust degree modelAccording to the execution log and user experience wecalculate the credibility of QoS information and IOPE similardegree Then we get the best candidate web service based on

trustworthyQoS and IOPE Finally through two experimentswe proved that the new method can effectively avoid theinfluence of web services which include exaggerated andunauthentic service profile

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

Denghui Wang would like to extend sincere gratitude tocorresponding author Hao Huang for his instructive adviceand useful suggestions on this research And the authorsthank the anonymous reviewers for their valuable feedbackand suggestions

References

[1] A Furno and E Zimeo ldquoContext-aware composition of seman-tic web servicesrdquoMobile Networks amp Applications vol 19 no 2pp 235ndash248 2014

[2] H Zheng W Zhao J Yang and A Bouguettaya ldquoQoS analysisfor web service compositions with complex structuresrdquo IEEETransactions on Services Computing vol 6 no 3 pp 373ndash3862013

[3] S Wang X Zhu and F Yang ldquoEfficient QoS management forQoS-aware web service compositionrdquo International Journal ofWeb and Grid Services vol 10 no 1 pp 1ndash23 2014

[4] C-F Lin R-K Sheu Y-S Chang and S-M Yuan ldquoA relaxableservice selection algorithm for QoS-based web service compo-sitionrdquo Information and Software Technology vol 53 no 12 pp1370ndash1381 2011

[5] Z Yanwei N Hong D Haojiang and L Lei ldquoA dynamicweb services selection based on decomposition of global QoSconstraintsrdquo in Proceedings of the IEEE Youth Conference onInformation Computing and Telecommunications (YC-ICT rsquo10)pp 77ndash80 November 2010

[6] Z-Z Liu X Xue J-Q Shen and W-R Li ldquoWeb servicedynamic composition based on decomposition of global QoSconstraintsrdquo International Journal of Advanced ManufacturingTechnology vol 69 no 9ndash12 pp 2247ndash2260 2013

[7] D Huijun Q Hua Z Jihong D Wenhan and X Wujie ldquoAdistributed optimal scheme based on local QoS for web servicecompositionrdquo China Communications vol 11 no 13 pp 142ndash147 2014

[8] J A Parejo S Segura P Fernandez and A Ruiz-CortesldquoQoS-aware web services composition using GRASP with PathRelinkingrdquo Expert Systems with Applications vol 41 no 9 pp4211ndash4223 2014


[9] A Missaoui ldquoA QoS-based neuro-fuzzy model for ranking webservicesrdquo in Proceedings of the 3rd International Conferenceon Information Technology and e-Services (ICITeS rsquo13) pp 1ndash5March 2013

[10] B Pernici and S H Siadat ldquoEvaluating web service QoS aneural fuzzy approachrdquo in Proceedings of the IEEE InternationalConference on Service-Oriented Computing and Applications(SOCA rsquo11) December 2011

[11] T Zhang J Ma Q Li N Xi and C Sun ldquoTrust-basedservice composition inmulti-domain environments under timeconstraintrdquo Science China Information Sciences vol 57 no 9 pp1ndash16 2014

[12] R Zhu H-M Wang and D-W Feng ldquoTrustworthy servicesselection based on preference recommendationrdquo Journal ofSoftware vol 22 no 5 pp 852ndash864 2011

[13] W Denghui H Hao and X Changsheng ldquoA novel web servicecomposition recommendation approach based on reliableQoSrdquoin Proceedings of the IEEE 8th International Conference onNetworking Architecture and Storage (NAS rsquo13) pp 321ndash325IEEE Xirsquoan China July 2013

[14] C Li B Cheng J Chen P Gu N Deng and D Li ldquoA webservice performance evaluation approach based on users expe-riencerdquo in Proceedings of the IEEE 9th International Conferenceon Web Services (ICWS rsquo11) pp 734ndash735 July 2011

[15] X Huang ldquoUsageQoS Estimating the QoS of web servicesthrough online user communitiesrdquo ACM Transactions on theWeb vol 8 no 1 article 1 2013

[16] J DengGrayControl System Huazhong Institute of TechnologyPress Wuhan China 1985

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MathematicsJournal of


Mathematical Problems in Engineering

Hindawi Publishing Corporationhttpwwwhindawicom

Differential EquationsInternational Journal of

Volume 2014

Applied MathematicsJournal of


Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of


Mathematical PhysicsAdvances in

Complex AnalysisJournal of


OptimizationJournal of


CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of


Operations ResearchAdvances in

Journal of


Function Spaces

Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of Mathematics and Mathematical Sciences


The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Algebra

Discrete Dynamics in Nature and Society



Decision SciencesAdvances in

Discrete MathematicsJournal of


Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Stochastic AnalysisInternational Journal of


2 Related Works

The traditional selection and composition of web services relyon themanner of finding themost similar functionalities andthe best nonfunctionalities of web service The QoS is widelyemployed for describing nonfunctionalitiesThere are a lot ofservice selection methods expanded in this regard Reference[5] proposed an algorithm to combine global QoS constraintswith local selection Reference [6] proposed an approachfor web service dynamic composition based on global QoSconstraints decomposition Besides global QoS [7] proposeda distributed optimal scheme based on local QoS Reference[8] presented QoS-GRASP a metaheuristic algorithm forperforming QoS-aware web service composition at runtimeQoS-GRASP is a hybrid approach that combinesGRASPwithPath Relinking References [9 10] applied the neurofuzzydecision making approach in the process of selection andchoice of the most appropriate web service with respectto quality of service criteria The method deals with theimprecision of QoS constraints values And these QoS-basedservice selection methods always assume that the QoS datacoming from service providers and users are effective andtrustworthywhich is actually impossible in real environmentSo these web service selection methods mentioned aboveare not perfect In [11] the authors proposed a novel servicecomposition approach modeling the trust-based servicecomposition as the multidomain scheduling and assignmentproblem using the minimum service resources within acertain time constraint They considered that trust plays apivotal role in service composition approach In [12] theresearchers presented a trustworthy services selection basedon preference selection method that assists users in selectingthe right web service according to their own preferenceThismethod can effectively solve the weaknesses of recommenda-tion systems In [13] we proposed onemethod forweb servicerecommendation based on trust-aware QoS But the methoddid not consider the functionality property Thus this paperbased on the above study proposes a novel composite webservice selectionmethod according to credibleQoS and IOPEinformation


31 Semantic Web Service Model

Definition 1 (semantic web service) A semantic web serviceis defined as a quadruple

WS = ⟨SNDes 119865QoS⟩ (1)

where SN is the identifier of the service Des is the generalinformation about the service including service contextssuch as text description ontology definition version con-tractor and the general information being independent of thespecific function 119865 is the functional attribute of a serviceQoS is a set of attribute parameters standing for the qualityof service including some attributes such as cost responsetime successful execution rate and availability The last twoparameters in the quadruple WS are actually related withservice composition

Moreover semantic web servicersquos function attribute canbe demonstrated as a quadruple

119865 = ⟨119868 119874 119875 119864⟩ (2)

where 119868 is the set of servicersquos semantic inputs 119874 is the setof servicersquos semantic outputs 119875 is the precondition 119864 is theeffects of the semantic web service Definitely a semantic webservice can be expressed as follows

WS = ⟨SNDes ⟨119868 119874 119875 119864⟩ QoS⟩ (3)

32 User Requirement Model

Definition 2 (user request) A user request of semantic webservice can be expressed as a quadruple

WS119903= ⟨SN

119903Des119903 119865119903QoS119903⟩ 119865

119903= ⟨119868119903 119874119903 119875119903 119864119903⟩ (4)

The service requesterrsquos information is in Des119903 A service

request can be expressed as follows

WS119903= ⟨SN

119903Des119903 ⟨119868119903 119874119903 119875119903 119864119903⟩ QoS

119903⟩ (5)

User requestmodel contains several requirement indexesand the consumer provides the corresponding constraintcondition for every index But the expression of every indexis different so we need to normalize these indexes Theexpression of user requirement indexes can be divided intotwo categories One is the value type And the other one is theinterval type In order to facilitate comparison with the realvalue we need to transform the interval type to the value typeAt present there is a variety of deterministic methods Inthis paper we use the expanded ordered weighted averagingoperator as determining the mathematical formula

Assume 119865 119877

119899

rarr 119877 if 119865(1198861 1198862 119886119899) = sum120596119895119887119895

wherein 119882 = (1205961 1205962 120596119899)119879 is weighted vector which is

associatedwith119865 and120596119895isin [0 1]Σ120596

119895= 1 and 119887

119895is the 119895th big

value in a group of data (1198861 1198862 119886119899) the119865 function is calledthe 119899-dimensional ordered weighted averaging operator

Assume that 119886 = [119886119871 119886119880] = 119909 | 119886119871 le 119909 le 119886119880 then 119886 isan interval value The 119865 function has the following formula

119891119896(119886

119871

119886

119880

) =

119886

119880

+ 119903119886

119871

119903 + 1

(6)

We transform the interval value to the value type throughthe following formula

UR119896(119886) = 119891

119896(119886

119871

119886

119880

) =

119886

119880

119903 997888rarr 0

119886

119871

+ 119886

119880

2 119903 997888rarr 1

119886

119871

119903 997888rarr infin

(7)

where 119896 is the number of user request indexes UR119896(119886) repre-

sents the normality request of the 119896th user After normalizingevery user requirement index we can get the user expectedvalue set UR




TD = ⟨TD119865TDQ⟩ (8)








UE = ⟨UE119892UE119897⟩ UE

119897= 119909 120583

119896(119909) | 119909 isin QoS (10)






120583119896(119909) =

0 0 le 119909 le 1205721

1 + [120573 (119909 minus 120572)] 120572 lt 119909 le 100

(11)











QRtime =1119870

119870

sum

119896=1(119879exit minus119879entry) (13)



TDQR (119894) =

1 minus1003816100381610038161003816


1003816100381610038161003816

Q119894

QR119894lt 2Q119894

0 QR119894gt 2Q119894

(14)



119894gt




119889119896(119894) =

1003816100381610038161003816


1003816100381610038161003816

US119896(119894) = 119903 (UR

119894Q119894) =

120588119889max119889119896(119894) + 120588119889max

120588 isin [0 1] (15)





119896


TD119896(119894) = 1minus

1003816100381610038161003816

US119896(119894) minus 120583

119896(119894)

1003816100381610038161003816

120583119896(119894)

TD119896=

sum

119868

119894=1 TD119896 (119894)119868

(16)



BS119896(119895) =

UR119896(119895) minusQ

119895

UR119896(119895) minusQ

119871

+ 06 UR119896(119895) gt Q

119895

0 UR119896(119895) le Q

119895

(1 le 119895 le 119869)

(17)



BS119896(119895) =

Q119895minus UR119896(119895)

Q119898minus UR119896(119895)

+ 06 UR119896(119895) lt Q

119895

0 UR119896(119895) ge Q

119895

(1 le 119895 le 119869)

(18)



TDQU (119895)

=

1119870

119870

sum

119896=1(1minus

1003816100381610038161003816

(TD119896(119895) times 120583

119896(119895)) minus BS

119896(119895)

1003816100381610038161003816

BS119896(119895)

)

(1 le 119895 le 119869)

(19)


119865

TD119865=

1119870

119870

sum

119896=1(TD119896timesUE119892)

=

1119870

119870

sum

119896=1(

sum

119869

119895=1 TD119896 (119895)119869

timesUE119892)

(20)









WS =WS119865+WSQ


119865+

119868

sum


+

119869

sum


(21)


119872 times (119868 times 119870 + 119869))

5 Case Study



119868

sum

119894=1(119882Q119894 timesQ119894) (22)









119875 =

119873119878

4 times 50

(23)













6 Conclusion





Acknowledgments


References

























Volume 2014




Journal of











Journal of


Function Spaces






Algebra












TD = ⟨TD119865TDQ⟩ (8)








UE = ⟨UE119892UE119897⟩ UE

119897= 119909 120583

119896(119909) | 119909 isin QoS (10)






120583119896(119909) =

0 0 le 119909 le 1205721

1 + [120573 (119909 minus 120572)] 120572 lt 119909 le 100

(11)











QRtime =1119870

119870

sum

119896=1(119879exit minus119879entry) (13)



TDQR (119894) =

1 minus1003816100381610038161003816


1003816100381610038161003816

Q119894

QR119894lt 2Q119894

0 QR119894gt 2Q119894

(14)



119894gt




119889119896(119894) =

1003816100381610038161003816


1003816100381610038161003816

US119896(119894) = 119903 (UR

119894Q119894) =

120588119889max119889119896(119894) + 120588119889max

120588 isin [0 1] (15)





119896


TD119896(119894) = 1minus

1003816100381610038161003816

US119896(119894) minus 120583

119896(119894)

1003816100381610038161003816

120583119896(119894)

TD119896=

sum

119868

119894=1 TD119896 (119894)119868

(16)



BS119896(119895) =

UR119896(119895) minusQ

119895

UR119896(119895) minusQ

119871

+ 06 UR119896(119895) gt Q

119895

0 UR119896(119895) le Q

119895

(1 le 119895 le 119869)

(17)



BS119896(119895) =

Q119895minus UR119896(119895)

Q119898minus UR119896(119895)

+ 06 UR119896(119895) lt Q

119895

0 UR119896(119895) ge Q

119895

(1 le 119895 le 119869)

(18)



TDQU (119895)

=

1119870

119870

sum

119896=1(1minus

1003816100381610038161003816

(TD119896(119895) times 120583

119896(119895)) minus BS

119896(119895)

1003816100381610038161003816

BS119896(119895)

)

(1 le 119895 le 119869)

(19)


119865

TD119865=

1119870

119870

sum

119896=1(TD119896timesUE119892)

=

1119870

119870

sum

119896=1(

sum

119869

119895=1 TD119896 (119895)119869

timesUE119892)

(20)









WS =WS119865+WSQ


119865+

119868

sum


+

119869

sum


(21)


119872 times (119868 times 119870 + 119869))

5 Case Study



119868

sum

119894=1(119882Q119894 timesQ119894) (22)









119875 =

119873119878

4 times 50

(23)













6 Conclusion





Acknowledgments


References

























Volume 2014




Journal of











Journal of


Function Spaces






Algebra












QRtime =1119870

119870

sum

119896=1(119879exit minus119879entry) (13)



TDQR (119894) =

1 minus1003816100381610038161003816


1003816100381610038161003816

Q119894

QR119894lt 2Q119894

0 QR119894gt 2Q119894

(14)



119894gt




119889119896(119894) =

1003816100381610038161003816


1003816100381610038161003816

US119896(119894) = 119903 (UR

119894Q119894) =

120588119889max119889119896(119894) + 120588119889max

120588 isin [0 1] (15)





119896


TD119896(119894) = 1minus

1003816100381610038161003816

US119896(119894) minus 120583

119896(119894)

1003816100381610038161003816

120583119896(119894)

TD119896=

sum

119868

119894=1 TD119896 (119894)119868

(16)



BS119896(119895) =

UR119896(119895) minusQ

119895

UR119896(119895) minusQ

119871

+ 06 UR119896(119895) gt Q

119895

0 UR119896(119895) le Q

119895

(1 le 119895 le 119869)

(17)



BS119896(119895) =

Q119895minus UR119896(119895)

Q119898minus UR119896(119895)

+ 06 UR119896(119895) lt Q

119895

0 UR119896(119895) ge Q

119895

(1 le 119895 le 119869)

(18)



TDQU (119895)

=

1119870

119870

sum

119896=1(1minus

1003816100381610038161003816

(TD119896(119895) times 120583

119896(119895)) minus BS

119896(119895)

1003816100381610038161003816

BS119896(119895)

)

(1 le 119895 le 119869)

(19)


119865

TD119865=

1119870

119870

sum

119896=1(TD119896timesUE119892)

=

1119870

119870

sum

119896=1(

sum

119869

119895=1 TD119896 (119895)119869

timesUE119892)

(20)









WS =WS119865+WSQ


119865+

119868

sum


+

119869

sum


(21)


119872 times (119868 times 119870 + 119869))

5 Case Study



119868

sum

119894=1(119882Q119894 timesQ119894) (22)









119875 =

119873119878

4 times 50

(23)













6 Conclusion





Acknowledgments


References

























Volume 2014




Journal of











Journal of


Function Spaces






Algebra













WS =WS119865+WSQ


119865+

119868

sum


+

119869

sum


(21)


119872 times (119868 times 119870 + 119869))

5 Case Study



119868

sum

119894=1(119882Q119894 timesQ119894) (22)









119875 =

119873119878

4 times 50

(23)













6 Conclusion





Acknowledgments


References

























Volume 2014




Journal of











Journal of


Function Spaces






Algebra



















6 Conclusion





Acknowledgments


References

























Volume 2014




Journal of











Journal of


Function Spaces






Algebra

























Volume 2014




Journal of











Journal of


Function Spaces






Algebra
















Volume 2014




Journal of











Journal of


Function Spaces






Algebra









Documents

Research Article A Novel Trust-Aware Composite Semantic ...downloads.hindawi.com/journals/mpe/2015/928193.pdf · Research Article A Novel Trust-Aware Composite Semantic Web Service