CN101383850A - Internet service selection system and method based on QoS body - Google Patents

Abstract

The invention discloses an Internet (Web) service selection system based on a service quality (QoS) body. The system comprises a QoS body expansion unit, a QoS request vector composition unit, a QoS advertisement vector composition unit and a Web service selection realization unit; the Web service selection realization unit is used for obtaining and calculating the QoS information comprehensive similarity degree between QoS information in a QoS request vector and QoS information in a QoS advertisement vector, and being matched with the demands of a service request terminal and services provided by a service provision terminal. The invention also discloses a Web service selection method based on the QoS body, and the method comprises the following steps: obtaining and calculating the QoS information comprehensive similarity degree between the QoS information in the QoS request vector and the QoS information in the QoS advertisement vector, and being matched with the demands of the service request terminal and the services provided by the service provision terminal. The needs of Web service selection based on semantics can be satisfied by adopting the system and the method.

Description

A kind of internet service selection system and method based on QoS body

Technical field

The present invention relates to the Internet (Web) services selection technology, relate in particular to a kind of Web service selective system and method based on service quality (QoS) body.

Background technology

Develop rapidly along with Internet, the application of network has been deep into a lot of aspects of society, the last obtainable service of Web sharply increases, and network information is exponential growth, make simultaneously from the user in each field to information search, visit, represent and safeguard become more and more difficult.How from huge bundle of services, to obtain the Web service of required function effectively and how in intimate Web service, to find the problem that allows the most satisfied service of user become be worth to be concerned about, and these Web service tasks of selecting just.Web service selection technology refers to: according to the QoS information of Web service, select one in the technology that meets most the Web service of service requester individual demand aspect the Qos from all multi-functional identical Web services.Here, so-called Web service refers to: the enough programming techniques of a kind of energy distributed application program that call by Web, interoperable.

As seen, along with the sharp increase of quantity of service on the network, more selection when using Web service, the user has been arranged.Because the QoS information of Web service is the important evidence of carrying out the Web service ordering and selecting.Therefore, QoS information is important consideration particularly in Web service selection technology.And existing Web service selection technology is on the basis of syntactic description, mates by keyword and realizes.Such as, import father in the search engine of user on network, can return the return results that is complementary with this keyword of father.Even if father and father are the speech that is associated semantically, semantically be actually a meaning, adopt existing Web service to select technology, be to return the return results that this speech that is associated with father of father is complementary.Thereby cause user's search experience, search efficiency and accuracy all to reduce greatly.That is to say that existing Web services selection technology lacks the description that Web service is had semantic information, intelligent degree is not high, still all can't satisfy the growing demand of people in accuracy on return rate.Obtain better return results, need more efficient and more perfect Web service to select technology.Solution is exactly to make that by the QoS body QoS information of all Web services all has semanteme in the network, increases the description that Web service is had semantic information, and computer can be understood and handle.

Below at selecting relevant present Research to set forth based on the Web service of semanteme.

Main research point concentrates on three aspects: the one, and the QoS modeling, the 2nd, the coupling of QoS information, three are based on the services selection of QoS.

At the QoS modeling, initial model is described the QoS of service with limited several information such as response time, reliability, availability and prices.This does not have semantic, and all QoS information are all expressed with real number, and extensibility is very poor.After QoS ontology describing language occurs, consider to utilize QoS ontology describing language to study the QoS Ontology Modeling.Such as DAML-QoS body and WSMO-QoS body.The DAML-QoS body is divided into three layers-QoS configuration layer, QoS information definition layer and QoS measure layer with the QoS body.The QoS configuration layer is used to carry out the QoS coupling.QoS attribute definition layer is described the domain of definition and the span of attribute.QoS measure layer has been stipulated the standard of tolerance.What the DAML-QoS body was paid close attention to is the integrally-built division of model, has set forth the relation between the QoS information emphatically.What the WSMO-QoS body was paid close attention to is each QoS information, has specifically described each attribute of QoS information, the QoS Information Level of corresponding DAML-QoS body.The WSMO-QoS body thinks that between each QoS information be equality, between do not have dependence.Aspect QoS tolerance, the WSMO-QoS body thinks that each metric all is a monodrome, can not support the tolerance mode of fuzzy monodrome and Interval Type.

At the coupling of QoS information, consider to use description logic to calculate the matching degree of two service QoS, comprise two kinds of solutions.First kind of solution be whole QoS with a service as matching unit, Subsume, Exact, PlugIn, Intersection and Disjoint Pyatyi matching degree have been proposed.And second kind of solution is to be matching unit with each QoS information, calculates earlier the level Four matching degree of corresponding QoS information respectively, and weighting summation then calculates the matching degree of whole QoS information.But second kind of solution compared with first kind of solution, and difference is that the algorithm granularity of second kind of solution is too thick.The characteristics of the coupling of QoS information are the couplings of numerical information.To this, first kind of solution with parameter value, semantic concept and＜=,=etc. mathematical operator combine, as the measure word of description logic, as＜10storage.MBMetric.Yet the computer understanding this type of information is inconvenient.Second kind of solution represented number range with an attribute of QoS information, numerical information and semantic information branch come treat, and helps computer understanding to handle.But this can only calculate between the QoS of same concept information, can not solve the matching problem of isomery QoS information.Here, so-called isomery QoS information refers to: with different forms QoS information is described, thus service impacting coupling.

At the services selection based on QoS, existing Web services selection algorithm is paid close attention to the calculating of qos value more, and by qos value being sorted the quality of rating services, the service that qos value is big more is outstanding more.Some Web service selection technology has been considered the demand of a little service requester, allows service requester to propose the restriction of some coarsenesses, under the prerequisite that satisfies restriction, again qos value is calculated and gets excellent.In fact, the service of qos value maximum not necessarily service requester needs most, and be only with the immediate service of service requester demand should be selecteed.In addition, existing Web services selection algorithm is not considered the completeness of service requester demand, the lowest term of service execution is not made judgement, and then can't guarantee the quality of selection result.

In sum, adopt existing Web services selection correlation technique, all there is certain defective aspect QoS modeling, QoS coupling and the services selection, accurate inadequately to the description of the semantic information of Web service, can not satisfy relying on the needs of QoS information well based on the Web service selection of semanteme.Still on return rate, all can't satisfy the growing demand of people in accuracy.

Summary of the invention

In view of this, main purpose of the present invention is to provide a kind of Web service selective system based on the QoS body, can satisfy relying on the needs that QoS information is selected based on the Web service of semanteme well, obtain more accurate return results, improve the service efficiency that technology is selected in Web service.

Another object of the present invention is to provide a kind of Web service system of selection based on the QoS body, can satisfy well and rely on the needs that QoS information is selected based on the Web service of semanteme, obtain more accurate return results, improve the service efficiency that technology is selected in Web service.

For achieving the above object, technical scheme of the present invention is achieved in that

A kind of Web service selective system based on the QoS body, this system comprises:

Service quality QoS body expanding element, QoS request vector constitute the unit, QoS advertisement vector constitutes the unit and the Internet Web service selects to realize the unit;

Wherein, QoS body expanding element is used to expand the QoS body;

The QoS request vector constitutes the unit, is used for extracting QoS information from described QoS body expanding element, constitutes the QoS request vector that meets the service request terminal demand;

QoS advertisement vector constitutes the unit, is used for extracting QoS information from described QoS body expanding element, constitutes the QoS advertisement vector that meets service that service providing end provides;

Web service selects to realize the unit, be used for obtaining and calculating the QoS informix similarity between the QoS information in described QoS request vector QoS information and the described QoS advertisement vector, mate the service that described service request terminal demand and described service providing end provide, select to realize the Web service that meets described service request terminal demand.

Wherein, this system also comprises the screening unit, is used for adopting default experience threshold values that the QoS information of described QoS advertisement vector is screened, and the screening back keeps compatible QoS information in QoS advertisement vector.

Wherein, described Web service selects to realize that the unit comprises: QoS informix similarity calculated is used for calculating the QoS informix similarity between the QoS information in described QoS request vector QoS information and the described QoS advertisement vector.

Wherein, described Web service selects to realize that the unit also comprises: service matching degree computing unit, be used for obtaining QoS informix similarity matrix from described QoS informix similarity calculated, with each the row weighting summation in the described QoS informix similarity matrix, constitute service matching degree vector, obtain the pairing service of element of numerical value maximum in the described service matching degree vector, finish the service matching degree calculating that Web service is selected.

A kind of Web service system of selection based on the QoS body, this method may further comprise the steps:

A, expansion QoS body;

B, extract QoS information, constitute QoS request vector that meets the service request terminal demand and the QoS advertisement vector that meets service that service providing end provides respectively from the QoS body of described expansion;

C, obtain and calculate in the described QoS request vector QoS informix similarity between the QoS information in the QoS information and described QoS advertisement vector, mate the service that described service request terminal demand and described service providing end provide, select to realize the Web service that meets described service request terminal demand.

Wherein, also comprise before the step C: adopt default experience threshold values that the QoS information in the described QoS advertisement vector is screened, the screening back keeps compatible QoS information in QoS advertisement vector.

Wherein, among the step C, described QoS informix calculation of similarity degree formula is:

Qsim(q _a，q _r)＝Qsims(q _a，q _r)×Qsimn(q _a，q _r)；

Wherein, Qsim (q _a, q _r) be QoS informix similarity; Qsims (q _a, q _r) be the semantic similarity of QoS information; Qsimn (q _a, q _r) be the numerical value similarity of QoS information; q _a, q _rBe respectively QoS information.

Wherein, the computing formula of the semantic similarity of described QoS information is:

Qsims(q _a，q _r)＝Sim(q _a，q _r)×C(q _a，q _r)；

Wherein, Sim (q _a, q _r) be the similarity of describing the common semantic concept of QoS information; C (q _a, q _r) for describing the function of QoS information compatibility.

Wherein, the numerical value calculation of similarity degree formula of described QoS information is:

$\mathrm{Qsimn}(q_a,q_r)=\sqrt{\frac{f^2(q_a.l,q_r.l)+f^2(q_a.u,q_r.u)}{2}};$

Wherein, q _a.l, q _a.u be respectively the first interval value QoS information q _aThe upper bound and lower bound, q _r.l, q _r.u be respectively the second interval value QoS information q _rThe upper bound and lower bound;

F (q _a.l, q _r.l) or f (q _a.u, q _r.u) with function f (a, r) expression, f (a r) is numerical value similarity function between metric a, the r of QoS information, and when a, r are increase type qos parameter, f (a, r)=fInc (a, r); When a, r are decrease type qos parameter, and f (a, r)=fDec (a, r).

Wherein, described QoS request vector is Q _R, comprise n the QoS information that service request terminal is paid close attention to; Described QoS advertisement vector is Q _A, Q _ABe the set of QoS advertisement vector, comprise the QoS advertisement vector of x the alternative Web service that all service providing end provide; Described default experience threshold values is Q _D, Q _DBe the QoS vector; Step C is specially:

C1, described Q to importing _ACarry out preliminary treatment, and with Q _ABe converted into the services selection matrix M of the capable m row of x _{X * m}

C2, obtain and according to described Q _DTo described M _{X * m}Screening filters out y the Web service of satisfying the execution basic demand of described Web service selection from x alternative Web service, the screening back constitutes the field selection matrix M of the capable m row of y ^D

C3, obtain and according to described Q _RFrom described M ^DIn extract the QoS information that service request terminal is paid close attention to, from M ^DSelect and Q in the QoS information of every row _RThe middle compatible QoS information of QoS information constitutes effective selection matrix M ^E

C4, calculate described M ^EIn every capable QoS information, with Q _RIn QoS informix similarity between the corresponding QoS information, constitute QoS informix similarity matrix M ^S

C5, with described M ^SIn each row weighting summation, constitute service matching degree vector S, obtain the element S of numerical value maximum among the described S _iThe service matching degree calculating that Web service is selected is finished in pairing service.

The present invention introduces the QoS body and describes QoS information, and the QoS body of describing QoS information is expanded; Employing is represented close between the QoS information and the degree that is associated based on the QoS comprehensive similarity of semantic matches, has made full use of the semantic factor in the QoS information, can handle isomery QoS information, has strengthened flexibility and accuracy that Web service is selected.Thereby guarantee the quality of Web service selection result, can realize the personalization that Web service is selected.And then be the center with the demand of service requester, with the primary condition that domain expert's suggestion is selected as Web service, can avoid the service requester demand incomplete and cause selected Web service to carry out the situation of failure.Here, so-called domain expert refers to: specialize in Web service, especially the expert of area research is selected in Web service QoS, Web service.Some evaluation criterions that the domain expert provides, the minimum standard that can carry out of Web service often at Web service QoS.In sum, be the value of QoS information because prior art pays close attention to, rather than meet the semantic factor in the QoS information of service requester demand, and the QoS body is exactly to be used to solve how to make QoS information have semantic technology.Therefore adopt the present invention, can satisfy relying on the needs that QoS information is selected based on the Web service of semanteme.

Employing the present invention have the following advantages:

First point, the present invention have made full use of the semantic factor of QoS information, can handle the situation of isomery QoS information.Specifically, existing Web services selection technology concentrate on that qos value calculates and the coupling of numeric representation scope on, do not have or seldom use semantic factor in the QoS information, therefore can not handle the Web service selection under the isomery QoS information state.And the present invention adopts the QoS body to describe QoS information, employing is represented close degree between the QoS information based on the QoS comprehensive similarity of semantic matches, made full use of the semantic factor in the QoS information, can carry out the services selection under the isomery QoS information state, strengthen the flexibility that Web service is selected.

Second point, the present invention are the center with the demand of service requester, can realize personalized Web service selection.Specifically, existing Web services selection technical concerns be the calculating of qos value, the foundation that Web service is selected is the size of qos value, does not consider whether the demand of alternative Web service and service requester approaching.And the present invention can select the Web service near the service requester demand, the foundation that adopts the QoS comprehensive similarity to select as Web service, can select and the immediate Web service of the demand of service requester, strengthen the personalized adaptive capacity that Web service is selected.

Thirdly, the present invention is the basis of services selection with domain expert's suggestion qos value, can avoid the service requester demand incomplete and cause selected Web service to carry out the failure phenomenon.Specifically, service requester is not QoS expert, may ignore some important QoS information, and therefore satisfying the selected Web service of service requester demand may not satisfy the executable primary condition of Web service.Existing Web services selection technology is not considered the executable minimum requirements of Web service, and therefore selected Web service might can not be carried out.And the present invention is to serve as the executable minimum requirements of service with domain expert's suggestion as Web, thereby avoids selecting Web service inferior, has improved the quality that Web service is selected.

Description of drawings

Fig. 1 is the composition structural representation of system of the present invention;

Fig. 2 is a realization schematic diagram of the QoS body specific implementation of expansion;

Fig. 3 is another realization schematic diagram of the QoS body specific implementation of expansion;

Fig. 4 is the another realization schematic diagram of the QoS body specific implementation of expansion;

Fig. 5 is a realization schematic diagram again of the QoS body specific implementation of expansion;

Fig. 6 is the realization flow schematic diagram of the inventive method;

Fig. 7 is the schematic diagram of the QoS numerical value similarity function of increase type QoS measure information value;

Fig. 8 is the schematic diagram of the QoS numerical value similarity function of decrease type QoS measure information value;

Fig. 9 inherits for the class of the QoS body of the present invention's expansion and realizes a realization schematic diagram of implementing.

Embodiment

Core concept of the present invention is: introduce the QoS body and describe QoS information, and the QoS body of describing QoS information is expanded; Employing is represented close between the QoS information and the degree that is associated based on the QoS comprehensive similarity of semantic matches, has made full use of the semantic factor in the QoS information, can handle isomery QoS information, has strengthened flexibility and accuracy that Web service is selected.

For making the purpose, technical solutions and advantages of the present invention clearer, by the following examples and with reference to accompanying drawing, the present invention is described in more detail.

As shown in Figure 1, a kind of Web service selective system based on the QoS body, this system comprises: QoS body expanding element 1, QoS request vector constitute unit 2, QoS advertisement vector constitutes unit 3 and Web service selects to realize unit 4.Wherein, QoS body expanding element 1 is used to expand in order to describe the QoS body of semantic relation between the QoS information.The QoS request vector constitutes unit 2 and is used for extracting QoS information from QoS body expanding element 1, constitutes the QoS request vector that meets the service request terminal demand.QoS advertisement vector constitutes unit 3 and is used for extracting QoS information from QoS body expanding element 1, constitutes the QoS advertisement vector that meets service that service providing end provides.Web service selects to realize that unit 4 is used for obtaining and calculating the QoS informix similarity between the QoS information in QoS request vector QoS information and the QoS advertisement vector, mate the service that service request terminal demand and service providing end provide, select to realize the Web service that meets this service request terminal demand.

Here, this system also comprises screening unit 5, be used for adopting default experience threshold values that the QoS information of QoS advertisement vector is screened, the screening back keeps compatible QoS information in QoS advertisement vector, and the QoS advertisement vector after will screening afterwards sends to Web service and selects to realize that unit 4 carries out subsequent treatment.

Here, Web service is selected to realize comprising in the unit 4: QoS informix similarity calculated is used for calculating the QoS informix similarity between the QoS information in QoS request vector QoS information and the QoS advertisement vector.Web service selects to realize that unit 4 also comprises: service matching degree computing unit, be used for obtaining QoS informix similarity matrix from QoS informix similarity calculated, with each the row weighting summation in the QoS informix similarity matrix, constitute service matching degree vector, obtain the pairing service of element of numerical value maximum in the service matching degree vector, finish the service matching degree calculating that Web service is selected.Wherein, the pairing service of element of numerical value maximum is in the service matching degree vector: be complementary with the service request terminal demand, by alternative Web service that service providing end provided.

As shown in Figure 6, a kind of Web service system of selection based on the QoS body, this method may further comprise the steps:

Step 101, by QoS body expanding element, expansion is in order to describe the QoS body of semantic relation between the QoS information.

Here, at the QoS body of expansion, the QoS body of this expansion is a series of clear and definite, formal, the shared notion that is used to describe QoS information, and the relation between these notions, such as inheritance.As the realization schematic diagram of Fig. 2 to the several specific implementations of QoS body that Figure 5 shows that this expansion.The QoS body of this expansion mainly comprises QoS profile (QoSProfile), QoS attribute (QoSAttribute), QoS trend (Tendency), tolerance (Metric) and five big classes of value (Value), and the inheritance between big class and the subclass.Wherein, QoSAttribute comprises three public QoS attributes of subclass (CommonQoSAttribute), field QoS attribute (DomainQoSAttribute), special QoS attribute (SpecialQoSAttribute) again, can define the QoS information of different levels by these three subclasses, the ISP also can expand this body by defining other subclasses.The Tendency class has two values of increase and decrease in order to the good and bad trend of expression QoS information.If the Tendency class is the increase type, what then this qos parameter rose whole QoS information is positive effect, is the bigger the better, as reliability.If the Tendency class is the decrease type, then this qos parameter plays negative effect to whole QoS information, and is the smaller the better, such as price.In fact, central QoS information may be an accurate value, also may be a scope, also may be a fuzzy value, and the tolerance mode that therefore is necessary for QoS information provides powerful expression ability.Therefore the OWL-QoS body has designed the tolerance mode that the Metric class defines QoS information.The OWL-QoS body that the present invention introduces can be supported monodrome, fuzzy monodrome and three kinds of tolerance modes of interval value.Value such as QoS information can be a numeric type, such as 5 yuan, 500 milliseconds; The value of QoS information can be the language description type, such as very high, very big.This demand realizes by the Value class in the OWL-QoS body.By the Value class, QoS information can be described by numerical value or language.

Step 102, QoS request vector constitute the unit and obtain the QoS body of expansion from QoS body expanding element, and extract QoS information, constitute the QoS request vector that meets the service request terminal demand.

Step 103, QoS advertisement vector constitute the unit and obtain the QoS body of expansion from QoS body expanding element, and extract QoS information, constitute the QoS advertisement vector that meets service that service providing end provides.

Step 104, Web service select to realize that the unit constitutes unit and QoS advertisement vector formation unit from the QoS request vector respectively, obtain QoS information in the QoS request vector and the QoS information in the QoS advertisement vector, QoS informix similarity between the QoS information in QoS information and the QoS advertisement vector in the calculating QoS request vector, mate the service that service request terminal demand and service providing end provide, select to realize the Web service that meets the service request terminal demand.

Here it is to be noted, before step 104, also comprise: by the screening unit, adopt default experience threshold values that the QoS information in the QoS advertisement vector is screened, the screening back keeps compatible QoS information in QoS advertisement vector, and the QoS advertisement vector after will screening afterwards sends to Web service and selects to realize that the unit carries out subsequent treatment.Like this, with the primary condition that the experience threshold values of domain expert suggestion is selected as Web service, can avoid the service request terminal demand incomplete and cause selected Web service to carry out the situation of failure.

Here, in the step 104, QoS informix calculation of similarity degree formula is following formula (1).

Qsim(q _a，q _r)＝Qsims(q _a，q _r)×Qsimn(q _a，q _r) (1)

Wherein, Qsim (q _a, q _r) be QoS informix similarity, be to be used to characterize two close degree between the QoS information; Qsims (q _a, q _r) be the semantic similarity of QoS information; Qsimn (q _a, q _r) be the numerical value similarity of QoS information; q _a, q _rBe respectively QoS information.

In the formula (1), the computing formula of the semantic similarity of QoS information is following formula (2).

Qsims(q _a，q _r)＝Sim(q _a，q _r)×C(q _a，q _r) (2)

Wherein, Sim (q _a, q _r) be the similarity of describing the common semantic concept of QoS information.C (q _a, q _r) for describing the function of QoS information compatibility, the value of this function is a Boolean, description is compatible with the QoS information of one side QoS attribute, C (q _a, q _r) value be 1 the expression compatible, otherwise incompatible.

In the formula (1), the numerical value calculation of similarity degree formula of QoS information is following formula (3).

$\mathrm{Qsimn}(q_a,q_r)=\sqrt{\frac{f^2(q_a.l,q_r.l)+f^2(q_a.u,q_r.u)}{2}}---\left(3\right)$

Wherein, q _a.l, q _a.u be respectively the first interval value QoS information q _aThe upper bound and lower bound, q _r.l, q _r.u be respectively the second interval value QoS information q _rThe upper bound and lower bound.F (q _a.l, q _r.l) or f (q _a.u, q _r.u) ((a r) is QoS numerical value similarity function between metric a, the r of QoS information to f for a, r) expression with function f; F (a r) has different expressions, when a, r are increase type qos parameter, f (a, r)=fInc (a, r); When a, r are decrease type qos parameter, and f (a, r)=fDec (a, r).

Wherein, (a r), is illustrated in figure 7 as the schematic diagram of the QoS numerical value similarity function of increase type QoS measure information value at fInc.(a, character r) is as follows for fInc.

a1、q _min<a，r<q _max；

a2、0≤fInc(a，r)≤100％；

A3, q _MinDuring＜a＜r, and fInc ' (a, r)〉0; FInc ＂ (a, r)〉0;

A4, r＜a＜q _MaxThe time, fInc ' (a, r)＜0; FInc ＂ (a, r)＜0.

Wherein, (a r), is illustrated in figure 8 as the schematic diagram of the QoS numerical value similarity function of decrease type QoS measure information value at fDec.(a, character r) is as follows for fDec.

b1、q _min<a，r<q _max；

b2、0≤fDec(a，r)≤100％；

B3, q _MinDuring＜a＜r, and fDec ' (a, r)〉0; FDec ＂ (a, r)＜0;

B4, a＜r＜q _MaxThe time, fDec ' (a, r)＜0; FDec ＂ (a, r)〉0.

Below come the concrete processing procedure of illustrative step 104 with an example, the step 104 here mainly comprises domain expert's selection course and personalized service selection course.In domain expert's selection course, by the screening unit, experience threshold values with domain expert's suggestion is reference, this experience threshold values is a qos value, the alternative Web service set that service providing end provides is filtered, guarantees that selected alternative Web service can satisfy the execution lowest term requirement of Web service selection.Like this, can avoid the service request terminal demand incomplete and cause selected Web service to carry out the situation of failure.In the personalized service selection course, be the center with the individual demand of service request terminal, be devoted to select the Web service that meets the service request terminal demand most.

Specifically, when the QoS request vector be Q _R, comprise n the QoS information that service request terminal is paid close attention to, and Q _R=(q _R1, q _R2..., q _Rn), q wherein _RiBe i QoS information.QoS advertisement vector is Q _A, Q _ABe the set of QoS advertisement vector, comprise the QoS advertisement vector of x the alternative Web service that all service providing end provide, and $\left\{Q_A\right\}=\{Q_A^1,Q_A^2,...,Q_A^x\},$ Wherein

Represent the QoS advertisement vector of j service, promptly $Q_A^j=(q_{a1}^j,q_{a2}^j,...,q_{\mathrm{am}}^j),$ Each QoS advertisement vector

Comprise m QoS information, wherein

K QoS information representing the advertisement vector of j alternative services.It is pointed out that language shape parameter value in all QoS advertisement vectors all passes through the process of de-fuzzy and be converted into the parameter value of numeric type.Here, at the process of de-fuzzy, usually the language shape parameter refers to: with the parameter of nonumeric formal description, such as the evaluation to certain qos parameter be divided into fine, good, general, bad, differ from five grades.With respect to describing with numerical value, this is a kind of fuzzy form of presentation.And the process of de-fuzzy is exactly of this sort process, such as above five grades being mapped to [0,1] space, fine corresponding 1, good corresponding 0.8, general corresponding 0.6, bad corresponding 0.4, poor corresponding 0.2.

Default experience threshold values is Q _D, Q _DBe the QoS vector.The concrete processing procedure of step 104 is so:

Step 1041, Q to importing _ACarry out preliminary treatment, and with Q _ABe converted into the services selection matrix M of the capable m row of x _{X * m}

Here, the Q of input _AFor $\left\{Q_A\right\}=\left\{\begin{array}{ccccc}(q_{a1}^1,& \&CenterDot;\&CenterDot;\&CenterDot;& q_{\mathrm{ak}}^1,& \&CenterDot;\&CenterDot;\&CenterDot;& q_{\mathrm{am}}^1)\\ \&CenterDot;\&CenterDot;\&CenterDot;& & \&CenterDot;\&CenterDot;\&CenterDot;& & \&CenterDot;\&CenterDot;\&CenterDot;\\ (q_{a1}^j,& & q_{\mathrm{ak}}^j,& & q_{\mathrm{am}}^j)\\ \&CenterDot;\&CenterDot;\&CenterDot;& & \&CenterDot;\&CenterDot;\&CenterDot;& & \&CenterDot;\&CenterDot;\&CenterDot;\\ (q_{a1}^x,& \&CenterDot;\&CenterDot;\&CenterDot;& q_{\mathrm{ak}}^x,& \&CenterDot;\&CenterDot;\&CenterDot;& q_{\mathrm{am}}^x)\end{array}\right\};$

The M that after preliminary treatment, obtains _{X * m}For

M _{X * m}Each the row $(q_{a1}^j,q_{a2}^j,...,q_{\mathrm{am}}^j)$ Represent the QoS advertisement vector of some alternative services

Each row $(q_{\mathrm{ak}}^1,q_{\mathrm{ak}}^2,...,q_{\mathrm{ak}}^x)$ Represent the same QoS information of different services, promptly $M[j,k]=q_{\mathrm{ak}}^j.$ Like this, follow-up passing through to matrix M _{X * m}The ranks computing can offer convenience for the QoS informix calculation of similarity degree between the same class QoS information of difference service, also make the calculating of the whole QoS information of same service clear.

Step 1042, obtain and according to Q _DTo M _{X * m}Screening filters out y the Web service of satisfying the execution basic demand of Web service selection from x alternative Web service, the screening back constitutes the field selection matrix M of the capable m row of y ^D

Step 1043, obtain and according to Q _RFrom M ^DIn extract the QoS information that service request terminal is paid close attention to, from M ^DSelect and Q in the QoS information of every row _RThe middle compatible QoS information of QoS information constitutes effective selection matrix M ^E

Step 1044, calculating M ^EIn every capable QoS information, with Q _RIn QoS informix similarity between the corresponding QoS information, constitute QoS informix similarity matrix M ^S

Step 1045, with M ^SIn each row weighting summation, constitute service matching degree vector S, obtain the element S of numerical value maximum among the S _iThe service matching degree calculating that Web service is selected is finished in pairing service.

Here, in the step 1045, the element S of numerical value maximum among the S _iPairing service is: be complementary with the service request terminal demand, by alternative Web service that service providing end provided.

Here it is pointed out that by above-mentioned steps 1042 and realize that domain expert's selection course is very important.If do not carry out domain expert's selection course, and the demand that just relies on service request terminal is selected, then because the requestor of service request terminal is not QoS expert, may not understanding some is enough to influence the importance that the QoS information of implementation effect is selected in Web service.For unapprehended QoS information, the requestor ignores it probably, and only the QoS information of its care is proposed demand.Therefore, even satisfy all demands of service request terminal, the execution failure that also may cause Web service to be selected because some important QoS information is left in the basket.Yet the domain expert can be a threshold value to the experience threshold values that each QoS information is provided suggestion rule of thumb, can satisfy the lowest term that execution is selected in Web service to guarantee the alternative Web service that service providing end was provided.This step just is being based on the consideration of this respect, the M after screening ^DCan avoid the service request terminal demand incomplete and cause selected Web service to carry out the situation of failure.M ^DThe meaning and the M of middle element _{X * m}Identical, only line number reduces to some extent.

Here it is pointed out that by step 1043 to step 1045 and can realize the personalized service selection course, to guarantee that in alternative Web service selected Web service is near the demand of service request terminal, to realize personalized services selection.

At step 1043, step 1043 also can be understood as the extraction process of effective QoS information.Below the realization of step 1043 is specifically addressed.

Because Q _RBe the individual demand of service request terminal, wherein comprise n the QoS information that service request terminal is paid close attention to.M ^DM the QoS information that one certain service providing end of line display is issued.Yet the number of m and n may be not exclusively the same, M ^DIn certain delegation also not necessarily comprise all effective QoS information, describe with the employed notion of QoS attribute can be not in full accord yet on the one hand.Therefore, need be according to Q _RTo M ^DAdjust, from M ^DSelect and Q in the QoS information of every row _RMiddle QoS information q _RiCompatible effective QoS information is rejected and all q _RiAll inconsistent invalid QoS information forms the effective QoS information vector.If Q _RIn certain QoS information q _RiWith M ^DIn certain M of delegation ^D[j, ^*] in all QoS information all incompatible, this means that the pairing service of this journey can not be q _RiPairing QoS attribute is given security, and can not satisfy service request terminal at q _RiThe individual demand of aspect, so this class service is with disallowable, promptly from M ^DMiddle row of rejecting effective QoS vector length less than n is selected z from y service, constitute effective selection matrix M ^EM ^EIn each element represent an effective QoS information, every row represent n of a service effectively QoS information vector, each row is represented and Q _RIn certain QoS information q _RiCompatible effective QoS information vector.

Be described as with the corresponding specific algorithm of the realization of above-mentioned steps 1043:

The matrix M of c1, a y * m of initialization ^E

C2, from up to down travel through M ^DEach the row M ^D[j, ^*]; J=1...y wherein; At certain M of delegation ^D[j, ^*] in, consider each QoS information M from left to right one by one ^D[j, k], wherein k=1...m; If can find Q _RIn certain QoS information q _RiWith M ^D[j, k] is compatible, then makes M ^E[j, i]=M ^D[j, k]; Otherwise make M ^E[j, i]=null, here, at q _Ri, i=1...n;

C3, from up to down check M ^EEach the row, if certain M of delegation ^E[j, ^*] in comprise certain value and be the element M of null ^E[j, i] is then with M ^E[j, ^*] Delete Entire Row.

C4, finally obtain the M of z * m ^E

With the corresponding code of above-mentioned arthmetic statement be:

for(j＝1 to y){

for(k＝1 to m){

for(i＝1 to n){

if(C(q _ri，M ^D[j，k])＝＝true)M ^E[j，i]＝M ^D[j，k]；

else M ^E[j，i]＝null；

}

}

M ^E＝M ^E remove M ^E[j， ^*]where M ^E[j，i]＝＝null；

}

At step 1044, step 1043 also can be understood as QoS informix calculation of similarity degree process.Below the realization of step 1044 is specifically addressed.

Because M ^EIn the QoS information of every row

With Q _RIn QoS information q _RiAll be corresponding one by one compatible, this is Q _AIn QoS information

With Q _RIn corresponding QoS information q _RiBetween coupling created condition.Therefore, can this QoS informix calculating formula of similarity of application of formula (1) calculate

And the QoS informix similarity between the qri, and then form QoS informix similarity matrix M ^SM ^SEach element is a percent value, and this value is the effective QoS information in the QoS advertisement vector of this alternative services

With its corresponding with service QoS request q _RiBetween QoS informix similarity, the QoS informix similarity vector of a service of each behavior.M ^SEssential meaning be the similarity degree of the QoS information QoS advertisement vector and the QoS request vector of having portrayed from the different aspect of QoS attribute.

Be described as with the corresponding specific algorithm of the realization of above-mentioned steps 1044:

The matrix M of d1, a z * m of initialization ^S

D2, from up to down travel through M ^EEach the row M ^E[j, ^*], j=1...z wherein; At certain M of delegation ^E[j, ^*] in, calculate each QoS information M from left to right one by one ^E[j, i] and Q _RIn i QoS information q _RiQoS informix similarity Qsim (M ^E[j, i], q _Ri), and make M ^S[j, i]=Qsim (M ^E[j, i], q _Ri), at M ^E[j, i], wherein i=1...n.

D3, finally obtain the M of z * m ^S

With the corresponding code of above-mentioned arthmetic statement be:

for(j＝1 to z){

for(i＝1 to n){

M ^S[j，i]＝Qsim(M ^S[j，i]，q _ri)；

}

}

At step 1045, step 1045 also can be understood as service matching degree computational process.Below the realization of step 1045 is specifically addressed.

M ^SPreserved each QoS advertisement vector Q _AIn each effective QoS information

With QoS request vector Q _RMiddle corresponding QoS information q _RiBetween QoS informix similarity.On the essence, M ^SSimilarity degree between the Web service of the alternative Web service that is provided of service providing end and service request terminal request has been provided from the different aspect of QoS.What QoS advertisement vector embodied is the alternative Web service that is provided of service providing end, what the QoS request vector embodied is the Web service of service request terminal request, and final services selection result need compare between each advertisement vector, and selecting accord with Q oS request vector is an alternative Web service of service request terminal individual demand.This just need be with M ^SEach the row change into a value unique, that can measure and compare, then this value is compared, sorts, and then obtain the result.Here, this value unique, that can measure and compare is service matching degree S.

And, because different service request terminal requestors' individual demand difference, therefore to QoS request vector Q _RMiddle QoS information q _RiDegree of concern also different.Can provide the ability of expressing degree of concern difference for the service request terminal requestor by the form of weight w.Therefore, definition service matching degree S is: Q _AIn effective QoS information

With Q _RMiddle corresponding QoS information q _RiBetween the weighting summation sum of QoS informix similarity, i.e. M ^SEach row element weighting summation, and then form service matching degree vector S.Each element of S is represented the matching degree S of a service _iTherefore, among the S in the element S of numerical value maximum _iPairing service matching degree is the highest, so S _iPairing service is: be complementary with the service request terminal demand, by alternative Web service that service providing end provided.

Be described as with the corresponding specific algorithm of the realization of above-mentioned steps 1045:

E1, an i dimension of initialization service matching degree vector S.

E2, from up to down travel through M ^SEach the row M ^S[j, ^*], j=1...z wherein; At certain M of delegation ^S[j, ^*] in, calculate each QoS similarity M from left to right one by one ^SI weight component w among [j, i] and the W _iProduct w _i* M ^S[i, j], and order $s_i=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}(w_i\&times;M^S[i,j]),$ At M ^S[j, i], wherein i=1...n.

E3, obtain i dimension service matching degree vector S.

E4, from S, select the element of numerical value maximum $s=\underset{i=1}{\overset{z}{\mathrm{Max}}}\left(s_i\right),$ Then the pairing service of s is selected service.

With the corresponding code of above-mentioned arthmetic statement be:

for(j＝1 to z){

$s_i=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}(w_i\&times;M^S[i,j]);$

}

$s=\underset{i=1}{\overset{z}{\mathrm{Max}}}\left(s_i\right)$

In sum, enforcement of the present invention needs four partial data supports, empirical value, QoS request vector and the QoS advertisement vector of the QoS body of expansion, domain expert's suggestion.Wherein, the QoS body of expansion is the expansion to the OWL-QoS body, also can be called field QoS body.Field QoS body is the specific implementation of OWL-QoS body in certain field, and it has embodied the semantic relation between each QoS information in this field, and field QoS body comprises a series of QoS information.These QoS information are all directly or indirectly inherited QoSAttrribute class in the OWL-QoS body, have semantic relation between separately.The empirical value of domain expert suggestion refers to: what the domain expert rule of thumb provided is guaranteeing that Web service selects the threshold value of each QoS information under the executable primary condition.For the QoS information of increase type, the empirical value of domain expert's suggestion is a minimum value.For the parameter of decrease type, the empirical value of domain expert's suggestion is a maximum.The QoS request vector is the vector that one group of QoS information is formed, and the expression service request terminal is to the individual demand of QoS information.QoS advertisement vector also is the vector that one group of QoS information is formed, and its is described is the QoS information of the alternative services that provides of service providing end.QoS information in QoS request vector and the QoS advertisement vector is all from field QoS body.

Method embodiment: adopt field QoS body shown in Figure 9 to realize Web service system of selection based on the QoS body.Specifically, the class that adopts extensibility (E), price (P), time delay (L), scale extensibility (SE) and these five of functional expansionaries (FE) to describe QoS information is formed field QoS body.Wherein E, SE, FE are the increase type.P, L are the decrease type.Semantic inheritance between all kinds of as shown in Figure 9.E, P, L directly inherit from QoSAttrribute class.SE, FE are the subclass of E.This means that this domain body can describe the QoS information of three aspects, correspond respectively to E, P, three QoS information of L.And E, SE, FE describe is the QoS information of autgmentability aspect, so is compatible mutually between E, SE, the FE.Be inconsistent mutually between E, P, the L.

Experience threshold values for domain expert's suggestion as shown in table 1, at E, P, L, SE and five QoS information of FE, the domain expert rule of thumb provides empirical value as shown in table 1 to suggestion.Wherein E, SE, FE are the increase parameter, and what table 1 provided is minimum threshold, and the Web service that is lower than this value can't be satisfied the primary condition that execution is selected in Web service.P, L are the decrease shape parameter, and what table 1 provided is max-thresholds, and the Web service that is higher than this value can't be satisfied the primary condition that execution is selected in Web service.

QoS information	E	SE	FE	P	L
						Empirical value	0.7	0.7	0.7	10	11

Table 1

QoS request vector of asking for service request terminal as shown in table 2 supposes that service request terminal is only interested in FE, P, these three QoS information of L, proposes the individual demand of oneself respectively.Utilize the powerful statement ability of OWL-QoS body, the requestor of service request terminal can propose the QoS information of three kinds of tolerance such as fuzzy monodrome, monodrome and interval value mode.And the requestor of service request terminal is identical to the degree of concern of these three parameters, and promptly the weight of being distributed is identical all is 1/3, as shown in table 2.

QoS information (weight)	FE(1/3)	P(1/3)	L(1/3)
				Value request	0.93 about	[9.88，9.90]	8.99

Table 2

QoS advertisement vector for alternative services that service providing end provides as shown in table 3 supposes to have the supplier of five service providing end can provide function identical but service that QoS is different.To some the QoS information with the aspect, the used notion of different service provider also may be different.As shown in table 3, for the QoS information of E aspect, what Service1, Service4 adopted is the E notion.That Service2 adopts is the sub-notion SE of E.And that Service3, Service5 employing is the sub-notion FE of E.

QoS information	E	SE	FE	P	L
						Service1	0.99	--	--	9.23	9.32

Service2	--	0.78	--	9.51	10.51
						Service3	--	--	0.9	10.32	9.21
Service4	0.95	--	--	9.91	9.81
						Service5	--	--	0.93	9.89	8.99

Table 3

Above field QoS body has been arranged, the experience threshold values of domain expert's suggestion, QoS request vector and QoS advertisement vector just can be implemented the Web service system of selection based on the QoS body.

At first, carry out domain expert's selection course.This domain expert's selection course is specially: select according to listed alternative Web service in the listed domain expert's empirical value his-and-hers watches 3 of table 1.For the P parameter, domain expert's recommended value is 10.And P is the QoS information of decrease type, and the P parameter maximum of alternative Web service must be lower than 10.Therefore, the P value is that 10.32 Service3 will fail to be elected.This expression Service3 does not satisfy Web service and selects executable lowest term, must carry out Times.And Service1, Service2, Service4, Service5 can satisfy Web service and select the lowest term carried out, by domain expert's selection, enter next step personalized selection course.

Then, carry out personalized selection course.Through previous step screening, Service1, Service2, Service4, Service5 enter this and take turns based on the personalized service of service request terminal demand and select.At first, will be according to the request of service request terminal, i.e. QoS request vector in the table 2: FE is about 0.93; P is in [9.88,9.90]; L constitutes the QoS request vector 8.99 by FE, P and L.Calculate the QoS informix similarity of Service1, Service2, four each QoS information of alternative Web service of Service4, Service5.This personalization selection course is specially: will consider semantic and the ambilateral factor of numerical value as a whole.Semantic aspect, for the E parameter, the Web service that has directly is described with the E notion, and the Web service that has is then described with its sub-notion SE, FE.Calculate and know that the semantic similarity of FE and E, SE is respectively 0.85,0.92.The numerical value aspect can define the numerical value similarity that obtain alternative Web service according to QoS numerical value similarity.At last, calculate the QoS informix similarity between each alternative Web service and the service request terminal individual demand, as shown in table 4 is QoS informix similarity.

Table 4

At last, the weight calculation of being distributed for each QoS information of paying close attention to according to QoS informix similarity and service request terminal the service matching degree of alternative Web service Service1, Service2, Service4, Service5.What matching degree was the highest promptly is the service of satisfying the service request terminal demand most, is the final result who selects.According to shown in the table 5, the service matching degree of Service5 is the highest, so Service5 is selected service.

Alternative Web service	Service1	Service2	Service4	Service5
					The QoS comprehensive similarity	0.61	0.42	0.76	0.98

Table 5

The above is preferred embodiment of the present invention only, is not to be used to limit protection scope of the present invention.

Claims (10)

1, a kind of Web service selective system based on the QoS body is characterized in that this system comprises:

Service quality QoS body expanding element, QoS request vector constitute the unit, QoS advertisement vector constitutes the unit and the Internet Web service selects to realize the unit;

Wherein, QoS body expanding element is used to expand the QoS body;

The QoS request vector constitutes the unit, is used for extracting QoS information from described QoS body expanding element, constitutes the QoS request vector that meets the service request terminal demand;

QoS advertisement vector constitutes the unit, is used for extracting QoS information from described QoS body expanding element, constitutes the QoS advertisement vector that meets service that service providing end provides;

Web service selects to realize the unit, be used for obtaining and calculating the QoS informix similarity between the QoS information in described QoS request vector QoS information and the described QoS advertisement vector, mate the service that described service request terminal demand and described service providing end provide, select to realize the Web service that meets described service request terminal demand.

2, system according to claim 1 is characterized in that, this system also comprises the screening unit, is used for adopting default experience threshold values that the QoS information of described QoS advertisement vector is screened, and the screening back keeps compatible QoS information in QoS advertisement vector.

3, system according to claim 1 and 2, it is characterized in that, described Web service selects to realize that the unit comprises: QoS informix similarity calculated is used for calculating the QoS informix similarity between the QoS information in described QoS request vector QoS information and the described QoS advertisement vector.

4, system according to claim 3, it is characterized in that, described Web service selects to realize that the unit also comprises: service matching degree computing unit, be used for obtaining QoS informix similarity matrix from described QoS informix similarity calculated, with each the row weighting summation in the described QoS informix similarity matrix, constitute service matching degree vector, the service matching degree calculating that Web service is selected is finished in the pairing service of element of obtaining numerical value maximum in the described service matching degree vector.

5, a kind of Web service system of selection based on the QoS body is characterized in that this method may further comprise the steps:

A, expansion QoS body;

B, extract QoS information, constitute QoS request vector that meets the service request terminal demand and the QoS advertisement vector that meets service that service providing end provides respectively from the QoS body of described expansion;

C, obtain and calculate in the described QoS request vector QoS informix similarity between the QoS information in the QoS information and described QoS advertisement vector, mate the service that described service request terminal demand and described service providing end provide, select to realize the Web service that meets described service request terminal demand.

6, method according to claim 5 is characterized in that, also comprises before the step C: adopt default experience threshold values that the QoS information in the described QoS advertisement vector is screened, the screening back keeps compatible QoS information in QoS advertisement vector.

According to claim 5 or 6 described methods, it is characterized in that 7, among the step C, described QoS informix calculation of similarity degree formula is:

Qsim(q _a，q _r)＝Qsims(q _a，q _r)×Qsimn(q _a，q _r)；

Wherein, Qsim (q _a, q _r) be QoS informix similarity; Qsims (q _a, q _r) be the semantic similarity of QoS information; Qsimn (q _a, q _r) be the numerical value similarity of QoS information; q _a, q _rBe respectively QoS information.

8, method according to claim 7 is characterized in that, the computing formula of the semantic similarity of described QoS information is:

Qsims(q _a，q _r)＝Sim(q _a，q _r)×C(q _a，q _r)；

Wherein, Sim (q _a, q _r) be the similarity of describing the common semantic concept of QoS information; C (q _a, q _r) for describing the function of QoS information compatibility.

9, method according to claim 8 is characterized in that, the numerical value calculation of similarity degree formula of described QoS information is:

$\mathrm{Qsimn}(q_a,q_r)=\sqrt{\frac{f^2(q_a.l,q_r.l)+f^2(q_a.u,q_r.u)}{2}};$

Wherein, q _a.l, q _a.u be respectively the first interval value QoS information q _aThe upper bound and lower bound, q _r.l, q _r.u be respectively the second interval value QoS information q _rThe upper bound and lower bound;

F (q _a.l, q _r.l) or f (q _a.u, q _r.u) with function f (a, r) expression, f (a r) is numerical value similarity function between metric a, the r of QoS information, and when a, r are increase type qos parameter, f (a, r)=fInc (a, r); When a, r are decrease type qos parameter, and f (a, r)=fDec (a, r).

10, method according to claim 9 is characterized in that, described QoS request vector is Q _R, comprise n the QoS information that service request terminal is paid close attention to; Described QoS advertisement vector is Q _A, Q _ABe the set of QoS advertisement vector, comprise the QoS advertisement vector of x the alternative Web service that all service providing end provide; Described default experience threshold values is Q _D, Q _DBe the QoS vector; Step C is specially:

C1, described Q to importing _ACarry out preliminary treatment, and with Q _ABe converted into the services selection matrix M of the capable m row of x _{X * m}

C2, obtain and according to described Q _DTo described M _{X * m}Screening filters out y the Web service of satisfying the execution basic demand of described Web service selection from x alternative Web service, the screening back constitutes the field selection matrix M of the capable m row of y ^D

C3, obtain and according to described Q _RFrom described M ^DIn extract the QoS information that service request terminal is paid close attention to, from M ^DSelect and Q in the QoS information of every row _RThe middle compatible QoS information of QoS information constitutes effective selection matrix M ^E

C4, calculate described M ^EIn every capable QoS information, with Q _RIn QoS informix similarity between the corresponding QoS information, constitute QoS informix similarity matrix M ^S

C5, with described M ^SIn each row weighting summation, constitute service matching degree vector S, obtain the element S of numerical value maximum among the described S _iThe service matching degree calculating that Web service is selected is finished in pairing service.

Images