CN101175090B

CN101175090B - Service synthesizing system and method thereof based on service characteristic

Info

Publication number: CN101175090B
Application number: CN2007100315529A
Authority: CN
Inventors: 顾宁; 刘铁江; 周树蕙; 卢暾; 顾庆; 李彦; 石晓旻
Original assignee: Fudan University; Huawei Technologies Co Ltd
Current assignee: Fudan University; Huawei Technologies Co Ltd
Priority date: 2007-11-21
Filing date: 2007-11-21
Publication date: 2011-01-19
Anticipated expiration: 2027-11-21
Also published as: CN101175090A

Abstract

The invention discloses a service synthesis system based on service characteristics, which comprises a memory module, an initialization module and a synthesis module. The a memory module is used for memorizing path tables and synthesis vector tables; the initialization module initializes the path tables and synthesis vector tables in the memory module through implementing modeling process to Web service group; the synthesis module is used for selecting a path as the best path from the multiple paths of input and output according to scheme selecting rules, and upgrading the path tables and synthesis vector tables in the memory module using the best path. Correspondently, the invention provides a service synthesis method based on service characteristics. According to the technology scheme of the invention, satisfactory service synthesis can be provided to users within good time complexity.

Description

Service synthesis system and service synthetic method based on service characteristic

Technical field

The present invention relates to electronic communication field, relate in particular to a kind of service synthesis system and service synthetic method based on service characteristic.

Background technology

That Web service has is self-contained, self-described and modular characteristics, can issue and call by Internet.It can handle simple request, also can be by simple Web service being combined the complicated service process of carrying out.The loose couplings that it possesses and the good characteristic of platform independence are fit to the interoperability between the isomery application and integrated under the Internet environment very much.

Along with increasing of Web service supplier quantity, redundant phenomenon can appear inevitably, and promptly different ISPs has developed the identical or similar Web service of function.Like this, the user has a plurality of candidate's Web services usually when using Web service or utilizing simple Web service to create compound Web service.In order to help the user to realize choosing automatically of Web service and to synthesize, on the basis of satisfying the user function demand, often with service quality (Quality of Service, QoS) standard of conduct differentiation and evaluate candidate Web service.Described QoS is a major criterion of weighing synthetic schemes, has reflected the satisfaction of user to synthetic schemes to a certain extent.

The service quality of Web service is made of multiple quality of service attribute (QoS Properties), can describe the nonfunctional characteristics of Web service from many aspects such as performance, reliability, availability and fail safes.

Web Service is on a kind of Web of being deployed in, self-contained, modular application program, and it can be described in network, issues, searches and call.Web Services has all advantages that object technology is promised to undertake; Simultaneously, therefore the foundation stone of Web Services has than any existing object technology better opening based on the Web specification technique of XML, opening, is the new platform of setting up the distributed application program of interoperable.

Web Services is synthetic to be exactly that Web Services with a plurality of autonomies is according to application need, provide Web Services new, with better function by the interface between service search and the service is integrated, the Web Services of some value-added is provided in other words.Synthetic Web Services be some independently, the gathering of mutual Web Services mutually, and the Web Services that it is assembled treated as the assembly of oneself, thereby acquisition has the function of renewal than original Web Services.On " granularity " meaning, Web Services is synthetic to be that Web Services is carried out more most encapsulation and this encapsulation back is exposed to the external world as a Web Service; And on " in proper order " meaning, synthetic is the order of calling of a series of Web Services of definition.

Service quality and Web Services composition algorithm aspect have obtained some achievements at present, but also good without comparison algorithm well combines both.The Web Services synthetic method that an existing quality drives, mainly comprise and carry out cost (price), time of implementation criterions such as (duration), on this basis, propose a global service system of selection, but specifically do not provided the process of finding the solution synthetic schemes.Other composition algorithms can be obtained the synthetic schemes of cost optimum in the complexity of O (n3) under the service cost has only the condition of single attribute.But when the cost of service is one when vectorial, the relation between them be interpreted as simply one and, promptly

Q (cs) = Σ_{i = 1}^{5} w_{i} \cdot Q_{i},

But actual conditions are more complicated.Liu has proposed a kind of dynamic synthesis method, and this method is at first carried out modeling to Web Services, adopts composition algorithm dynamically to generate synthetic schemes then.Corresponding to a plurality of outputs of customer requirements, this method is carried out finding the solution of a synthetic schemes for each output, obtains all schemes corresponding to single output.And then resultant scheme is done nature connect.Naturally connect and be meant that the hypothesis customer requirements is output as { A, B, C}, synthetic schemes corresponding to A output is 4, and the synthetic schemes of B output is 5, and the synthetic schemes of C output is 6, according to this algorithm, the synthetic schemes of respective user output will have 4 * 5 * 6=120, and all schemes that it is exhaustive draws according to user's input and output feed back to the user according to the scheme that the QoS method is found out a QoS optimum then.The result here is the QoS optimum, but exhaustive owing to carrying out, the time complexity of whole process is huge unusually, is unfavorable in practice realizing.

Summary of the invention

The embodiment of the invention is to provide a kind of service synthesis system and service synthetic method based on service characteristic, can automatically synthesize in less time complexity, obtains the optimum synthetic schemes of user preference.

In order to realize above-mentioned effect, the embodiment of the invention has proposed a kind of service synthesis system based on service characteristic, and it comprises:

Memory module is used for store path table and composite vector table;

Initialization module comes routing table and composite vector table in the initialization memory module by the Web service set being carried out modeling processing (as: production modeling, attribute binding modeling);

Synthesis module is used for selecting a paths as optimal path according to the Scheme Choice rule from the mulitpath that is input to output, and uses described optimal path to upgrade routing table in the memory module and composite vector table.

Correspondingly, the present invention also provides a kind of service synthetic method based on service characteristic, it is characterized in that, comprising:

Obtain initial routing table and composite vector table by the Web service set being carried out modeling (comprising production modeling or attribute binding modeling) processing;

From the mulitpath that is input to output, select a paths as optimal path according to the Scheme Choice rule, and use described optimal path to upgrade described initial routing table and composite vector table.

A kind of quality-of-service based service synthesis system and service synthetic method according to embodiment of the invention proposition, by the modeling processing input and output of Web service are represented with path model, the synthetic process of service is exactly the routing update process, by choosing the path according to Scheme Choice rule (user is to the ordering of optimization preference and the discrimination of each qualitative attribute of service), and do not need exhaustive all synthetic schemes, reduce time complexity, improved system works efficient.

Description of drawings

Fig. 1 is a kind of application scenarios schematic diagram of the embodiment of the invention;

Fig. 2 is a kind of structural representation of serving synthesis system of the embodiment of the invention;

Fig. 3 is the structural representation of the described initialization module of Fig. 2;

Fig. 4 is the structural representation of the described synthesis module of Fig. 2;

Fig. 5 is a kind of flow chart of serving synthetic method of the embodiment of the invention;

Fig. 6 is the particular flow sheet of the step S1 of the described method of Fig. 5;

Fig. 7 is the particular flow sheet of the step S2 of the described method of Fig. 5;

Fig. 8 is the particular flow sheet of the step S3 of the described method of Fig. 5;

Fig. 9 is the network structure of a specific embodiment provided by the invention;

Figure 10 is the path schematic diagram corresponding to network structure shown in Figure 9;

Figure 11 is a kind of schematic diagram of binding modeling in the embodiment of the invention.

Embodiment

A kind of service synthesis system and service synthetic method that the embodiment of the invention proposes based on service characteristic, by the modeling processing input and output of Web service are represented with path model, the synthetic process of service is exactly the routing update process, the shortest path that obtains at last is the optimal case of user preference.

Below in conjunction with accompanying drawing, elaborate a kind of that the embodiment of the invention proposes based on the service synthesis system of service characteristic and the technical scheme of service synthetic method.

With reference to figure 1, be a kind of application scenarios of the embodiment of the invention.

Operation list 1 has been stored the log-on message and the corresponding descriptor of professional or service, comprises incidence relation between interface message, the Web service, QoS information etc.;

QoS dimension synthetic standards provides server 3 to be responsible for safeguarding the general QoS weights computational logic of current standard definition;

The weights computational logic that ordering of optimization preference that service synthesis system 2 provides according to user 3 and discrimination and QoS dimension synthetic standards provide server to provide synthesizes the multiple service that operation list provides, at last the input set that provides according to user 3 and the requirement of output collection provide shortest path or the corresponding scheme of optimum QoS information to the user.

Wherein, described ordering of optimization preference be the user according to self preference to a plurality of QoS attributes, an ordering from high to low;

Described discrimination is in the service building-up process, for several alternative paths (path is corresponding to scheme) when comparing, at certain concrete QoS attribute, the percentage of user's acceptable performance difference.As: the execution cost (Price) of 4 paths is respectively 1.0,1.1,1.2,1.5 o'clock, the user area calibration is 15%, because 1.0 is current best, other 3 (1.1,1.2,1.5) and 1.0 comparisons, the user have only 1.1 and 1.0 gap not exceed 15%, so can accept further next attribute relatively.

As shown in Figure 2, the described service synthesis system 2 of Fig. 1 specifically can comprise:

Memory module 21 is used for store path table and composite vector table;

Initialization module 22 comes routing table and composite vector table in the initialization memory module 21 by the Web service set being carried out modeling processing (as: production modeling, attribute binding modeling);

Synthesis module 23 is used for selecting a paths as optimal path according to the Scheme Choice rule from the identical mulitpath that is input to identical output, and uses routing table and composite vector table in the described optimal path updated stored module 21;

Many service processing module 24 are used for that synthesis module 23 is upgraded the routing table and the composite vector table that obtain and carry out many service processing, obtain optimal path table and optimum composite vector table at last.

Enquiry module 25, the input set and the output set that are used for providing according to the user are inquired about through composite vector table and routing table after

synthesis module

23 and 24 processing of many service processing module, to obtain the path and the vector value of the QoS cost vector optimum corresponding with described input set and output set.

Wherein, described production modeling is that a Web service is expressed as a production, and the body of production and head are represented the input and output of Web service respectively, as: I-＞O, wherein, I represents input, O represents output; That is: production, reachable path, Web service three concern one to one;

Described routing table is used for storing in the current building-up process of obtaining the pairing routing information of optimum cost vector that each is input to each output, and is as shown in table 1:

Table 1

Vertical expression input of table, as: { A, B} represent to be input as A and B; Laterally expression output, as: { C} represents to be output as C; { (1) } expression in the form has path 1 to reach, 1 expression path number, and the blank in the form represents that corresponding path wouldn't reach; As: be input as A and B, being output as C has path 1 to reach, the production that it is corresponding: { A, B}-＞{ C}, an also corresponding a kind of Web service of reachable path simultaneously or synthetic schemes.

Described composite vector table is used for storing that each is input to the optimum cost vector information of each output in the current building-up process of obtaining, and is as shown in table 2:

Table 2

Vertical expression input of table, as: { A, B} represent to be input as A and B; Laterally expression output, as: { C} represents to be output as C, and the data representation composite vector in the parentheses in the form is the cost vector of this synthetic schemes, as: (0.5,0.8,2,46,2.23,2) represented A, B}-＞the cost vector of the synthetic schemes of C} correspondence, the expression of 0 in form input and output are identical, blank in the form represents that the cost vector is infinitely great, does not promptly have reachable path.

Wherein, described Scheme Choice rule is specially:

Obtain the ordering of optimization preference of user to the QoS attribute, as: cost, time of implementation, reliability, availability, reputation carried out;

Choose first attribute in the described ordering, as: cost carried out;

Whether the difference of judging the described attribute of two schemes surpasses discrimination;

If described judged result is for being then according to the property value of two schemes, to choose one of them scheme as optimal case according to first rule;

If described judged result is chosen the next attribute (time of implementation) in the described ordering for not, whether the difference of returning the described attribute of judging two schemes surpasses the step of discrimination.

Wherein, described first rule is at concrete QoS attribute, and the selection rule that sets in advance particularly, is chosen the scheme of a more excellent attribute correspondence.With five above-mentioned attributes is example:

For carrying out cost, the user wishes that cost is the smaller the better, so first rule predetermining is chosen the less scheme of weights of this attribute;

For the time of implementation, the user wishes that the processing time is the smaller the better, so first rule predetermining is chosen the less scheme of weights of this attribute;

For reliability, the user expectation reliability is the bigger the better, so first rule predetermining is chosen the bigger scheme of weights of this attribute;

For availability, the user expectation availability is the bigger the better, so first rule predetermining is chosen the bigger scheme of weights of this attribute;

For reputation, the better scheme of user expectation reputation is so first rule predetermining is chosen the bigger scheme of weights of this attribute.

In the specific implementation, as shown in Figure 3, the described initialization module 22 of Fig. 2 comprises:

Set of service unit 221 is used for storing the relation information of Web service set and this each Web service of set, and described relation information comprises the incidence relation between each Web service at least;

Serve weights aggregation units 222, be used to store the QoS weights information of described each Web service;

Many standard evaluations unit 223 is used for the QoS weights information according to service weights aggregation units 222, by the method that adopts the quality vector set of the Web service in the set of service unit 221 is estimated, and is generated the set of service that contains the QoS vector;

Production modeling unit 224 is used for the set of service that contains the QoS vector that many standard evaluations unit 223 generates is carried out the production modeling, generates the modeling table, and is as shown in table 3.

Table 3

Wherein, “ ﹠amp; " expression and relation, the vertical production of expression is laterally represented the execution cost weights of each quality standard of each production correspondence, i.e. the QoS weights.

Synthetic initialization unit 225 is used for routing table and composite vector table according to the described memory module 21 of modeling table initialization of production modeling unit 223 generations.

In the specific implementation, as shown in Figure 4, the described synthesis module 23 of Fig. 2 comprises:

Weights computational logic unit 231, be used to store the weights computational logic, this weights computational logic is corresponding to the computational methods to each quality standard (or QoS attribute), as: for carrying out cost and time of implementation: directly addition, for reliabilty and availability: get its exponential form and multiply each other, for reputation: add and remove after the value of current path in current path sum mean value etc. relatively.

Ordering of optimization preference and discrimination unit 232 are used to store the ordering of optimization preference and the discrimination of each service quality standard or Service Properties, and described ordering of optimization preference is that the user sets the preference ordering to each quality standard; Described discrimination is used for representing the difference on each quality standard of user's acceptable;

Synthesis unit 233 is used for synthesizing calculating according to described weights computational logic and ordering of optimization preference and discrimination, upgrades routing table and composite vector table in the described memory module, obtains complete set of paths.

More than a kind of service synthesis system based on service characteristic of the embodiment of the invention is described in detail, elaborate a kind of technical scheme of serving synthetic method that the embodiment of the invention proposes below in conjunction with Fig. 5, Fig. 6, Fig. 7 and Fig. 8.

With reference to figure 5, a kind of service synthetic method based on service characteristic comprises:

Step S1 obtains initial routing table and composite vector table by the Web service set being carried out modeling processing (as: production modeling, attribute binding modeling);

Step S2 selects a paths as optimal path according to the Scheme Choice rule from the identical mulitpath that is input to identical output, and uses described optimal path to upgrade described initial routing table and composite vector table.

When detecting the situation that has many services, after step S2, also comprise:

Step S3 carries out many service processing to upgrading the routing table and the composite vector table that obtain by the cost vector, obtains optimal path table and optimum composite vector table.

Wherein, be modeled as example with production, as shown in Figure 6, the described step S1 of Fig. 5 specifically comprises:

Step S101 generates the set of service that contains the QoS vector according to Web service aggregate information and QoS weights information;

Step S102 carries out the production modeling to the set of service of the described QoS of containing vector, generates the modeling table;

Step S103 generates initial routing table and composite vector table according to described modeling table.

As shown in Figure 7, the described step S2 of Fig. 5 is specially: each row at described composite vector table is all carried out following step one time:

S201 generates two node set, that is: processing node is gathered and the node set that is untreated;

S202, all output nodes of the described composite vector table of initialization are in the node set that is untreated;

S203 joins processing node set with described composite vector table when the input node of pre-treatment row input side;

S204 judges whether described input node exists reachable path to the node set that is untreated, if, execution in step S205 then, otherwise the node in the node set that directly will be untreated joins processing node set, and empties the node set that is untreated;

S205 selects first path according to the Scheme Choice rule from described reachable path, and the output node in first path is joined processing node set, the output node in described first path of deletion in the processing node set never;

S206 judges whether the output node in described first path exists reachable path to the node set that is untreated, if, execution in step S207 then;

S207 selects second path according to the Scheme Choice rule from described reachable path;

S208 synthesizes the Third Road footpath with described first path and second path;

S209 judges whether the input node exists reachable path to described Third Road output node directly, if, execution in step S210 then, otherwise, execution in step S211;

S210 selects the 4th path according to the Scheme Choice rule from described reachable path and Third Road footpath, and with described composite vector table of the 4th routing update and routing table;

S211 directly upgrades described composite vector table and routing table with described Third Road;

S212, judge whether the node set that is untreated is empty, if not, then returns execution in step S204: judge whether described input node exists the step of reachable path to the node set that is untreated, otherwise, finish this step.

Wherein, described in embodiments of the present invention Scheme Choice rule is specially:

Choose first attribute in the described ordering, as: cost carried out;

If described judged result is for being, then, choose one of them scheme as optimal case according to first rule according to the weights of the attribute of two schemes;

Wherein, described first rule is at concrete QoS attribute, chooses the scheme of more excellent attribute correspondence.

As shown in Figure 8, the described step S3 of Fig. 5 is specially: each row at described composite vector table is all carried out following step one time:

S301 judges whether the input when the pre-treatment row is a set, if then carry out next step; As: be input as that { C, D} are set;

S302 judges whether to exist node to arrive described set and forms reachable path, if then carry out next step; As: there is A-＞C, B-＞D;

S303 is generated as input set with described node, and described set is generated as the output collection, adds in described composite vector table and the routing table; That is: input set is that { the output collection is { C, D} for A, B};

S304, when described input set when there is reachable path in other node, at each QoS attribute, choose attribute more excellent in the attribute of described reachable path according to second rule and upgrade the QoS attribute of described input set to the path of described node;

Wherein, described second rule is at concrete QoS attribute, and the selection rule that sets in advance particularly, is chosen more excellent attribute.To carry out the example that is chosen for of cost, time of implementation, reliability, availability, reputation:

For carrying out cost, the user wishes that cost is the smaller the better, so, second rule predetermining: choose the less attribute of weights;

For the time of implementation, the user wishes that the processing time is the smaller the better, so second rule predetermining is chosen the less attribute of weights;

For reliability, the user expectation reliability is the bigger the better, so second rule predetermining is chosen the bigger attribute of weights;

For availability, the user expectation availability is the bigger the better, so second rule predetermining is chosen the bigger attribute of weights;

For reputation, the better scheme of user expectation reputation is so second rule predetermining is chosen the bigger attribute of weights.

As: exist A, B}-＞X then compare A, B}-＞C, D}-＞X, A-＞X, B-＞X gets every attribute optimal value and upgrades { A, B}-＞X;

S305, when other node when there is reachable path in described output collection, at each QoS attribute, according to the opposite rule of second rule choose attribute relatively poor in the attribute of described reachable path upgrade described node to described output collection the QoS attribute in path;

Wherein, described and the opposite rule of second rule, that is: the choosing method with second rule predetermining is opposite, particularly, chooses relatively poor attribute.Carrying out the example that is chosen for of cost, time of implementation, reliability, availability, reputation,

For carrying out cost, the user wishes that cost is the smaller the better, described and the second regular opposite rule predetermining: choose the bigger attribute of weights;

For the time of implementation, the user wishes that the processing time is the smaller the better, described and the opposite rule predetermining of second rule: choose the bigger attribute of weights;

For reliability, the user expectation reliability is the bigger the better, described and the opposite rule predetermining of second rule: choose the less attribute of weights;

For availability, the user expectation availability is the bigger the better, described and the opposite rule predetermining of second rule: choose the less attribute of weights;

For reputation, the better scheme of user expectation reputation, described and the opposite rule predetermining of first rule: choose the less attribute of weights.

As: exist X-＞C, D}, X-＞C relatively then, X-＞D gets every attribute and upgrades X-＞{ C, D} than difference.

The foregoing description has been done detailed elaboration to a kind of service synthesis system and method based on service characteristic that the present invention proposes, in order to make those skilled in the art can further understand and use the present invention, below in conjunction with Fig. 9 and Figure 10, specifically set forth the technical scheme of the embodiment of the invention.

In the present embodiment, the attribute of service quality (QoS) comprising: carry out cost (Price), time of implementation (Duration), reliability (Reliability), availability (Availability) and reputation (Reputation), certainly comprise other attribute in practice, as: fail safe, disaster tolerance etc.

Its QoS vector representation is:

Q(cs)＝(Qprice(cs)，Qdu(cs)，Qrel(cs)，Qav(cs)，Qrep(cs)

The user to the ordering of optimization preference of QoS attribute is: carry out cost, time of implementation, reliability, availability and reputation, these five attributes are provided discrimination be respectively: 10%, 20%, 30%, 40%, 50%.

Fig. 9 is a kind of Web service network structure in the present embodiment, wherein, 7 Web services of

numeral

1,2,3,4,5,6,7 expressions in the circle, letter A, B, C, D, E, F represent the input and output of each service, " ∧ " expression and relation, " ∨ " expression or relation, as: for service 6: input C and D, output E.For the ease of understanding the technical scheme of the embodiment of the invention, Fig. 9 is converted into traditional path profile, as shown in figure 10, letter representation input node or output node in the circle, limit (straight line of illustrated band arrow) expression weight vector, first digit on the limit is a service-number, two letters (node) if between have the limit, then represent the service of corresponding input and output.

The service synthesis system treatment step of the embodiment of the invention is as follows:

Step 1, carry out the production modeling:

According to the relation information of all Web services among Fig. 9 and weights information generate contain the QoS vector set of service (as shown in figure 10,), carry out the production modeling according to each node shown in Figure 10 and vector relations, be about to exist the mid-side node relation table to be shown: I-＞O form, generate the modeling table, as shown in table 4:

Table 4

Step 2 generates initial composite vector table and routing table:

Generate initial composite vector table according to above-mentioned modeling table, as shown in table 5, vertically be the input of production, laterally be the output of production, have in the form of reachable path and insert the QoS vector, as: the form that is input as A and is output as the C correspondence inserts (0.5,0.8,1.65,2.23,2); The input node set comprises output node when set, and form inserts 0, as: be input as that { C, D} and the form that is output as the C correspondence insert 0.When list data was not empty, there was a path that is input to output in expression, a promptly corresponding Web service or synthetic schemes, and when list data was sky, expression wouldn't reach, and promptly wouldn't have reachable path.

Table 5

Generate initial routing table according to above-mentioned modeling table, as shown in table 6:

Table 6

Wherein, this routing table be the input of production vertically, be the output of production laterally, have service-number in the form of reachable path, list data is that 0 expression input node set comprises the output node set, and list data is that empty expression wouldn't exist reachable path.

Step 3 is synthesized processing, brings in constant renewal in composite vector table and routing table:

From the identical mulitpath that is input to identical output, select a paths according to the Scheme Choice rule as optimal path, and use described optimal path to upgrade described routing table and composite vector table, particularly, carry out following operation at each row of described composite vector table:

To first row, generates two node set, that is: the processing node S set and the node set N that is untreated, initially handled S set and be: A}, and all the other horizontal output nodes are gathered N to being untreated, and promptly N is: { C, D, E, F};

Judge whether input node A exists reachable path to the set N that is untreated, if exist, then selects optimal path from described reachable path, is specially:

According to table 5 first row, find out all production that are input as A, take out all weight vectors (representing) with V:

A-＞C：V1(0.5，0.8，1.65，2.23，2)，

A-＞D：V2(0.85，1.0，1.49，2.23，2)，

A-＞F：V3(1.5，1.0，2.46，2.34，3)，

According to the Scheme Choice rule, compare for the first attribute Price, Price is the smaller the better, this moment V1 the Price optimum, be 0.5, the user area calibration is 10%, and other two Price have all surpassed 0.5 * (1+10%)=0.55, promptly exceeded the scope that the user accepts, so choose the path of V1 correspondence, corresponding service (production and service are one to one) is A-＞C, and join C and handle S set this moment, be that S is { A, C} gathers deletion of node C the N from being untreated simultaneously, and promptly N is { D, E, F}.

Judge whether to exist the reachable path of C point to the set N that is untreated, through inquiry composite vector table (table 5), there is the production of C-＞F in discovery, promptly has the path of C to F, and A-＞C and C-＞F are synthesized A-＞C-＞F, handle according to given QoS weights computational logic, in this example, the addition of preceding two attributes (carrying out cost and time of implementation) value, middle two attributes (reliabilty and availability) value multiplies each other, last attribute (reputation) value addition is averaged, that is:

A-＞C：V1(0.5，0.8，1.65，2.23，2)

C-＞F：V(0.3，2.0，1.82，1.49，2.5)

A-＞C-＞F：V4(0.8，2.8，3.0，3.32，2.25)

Judge whether to exist the reachable path of A, find to exist through question blank 5: A-＞F:V3 (1.5,1.0,2.46,2.34,3) to F;

According to the Scheme Choice rule, from A-＞C-＞F:V4 (0.8,2.8,3.0,3.32,2.25) and A-＞F:V3 (1.5,1.0,2.46,2.34,3) in select more optimal way, be specially: the first attribute Price is compared, Price is the smaller the better, the Price optimum of V4 is 0.8 at this moment, and the user area calibration is 10%, and the Price of V3 1.5 has surpassed 0.8 * (1+10%)=0.88, promptly exceed the user and accepted scope, therefore, chosen the path of V4 correspondence, be production: A-＞C-＞F, and upgrade routing table and composite vector table, be specially: renewal A-＞F is A-＞C-＞F, replaces V3 with V4 simultaneously, this moment, the composite vector table was as shown in table 7, and routing table is as shown in table 8.

Table 7

Table 8

Judge whether the node set that is untreated is empty, at this moment, the processing node S set is { A, C}, the node set that is untreated N is { D, E, F}, be that N is not sky, continue then to judge whether input node A exists reachable path to the set N that is untreated, and there is the path of following production correspondence in composite vector table 7 discovery after looking into renewal:

A-＞D：V6(0.85，1.0，1.49，2.23，2)

A-＞F：V7(0.8，2.8，3.0，3.32，2.25)，

According to the Scheme Choice rule, the first attribute Price is compared, Price is the smaller the better, this moment V7 the Price optimum, be 0.8, the user area calibration is 10%, and the Price of V6 is 0.85 less than 0.8 * (1+10%)=0.88, and promptly the user can accept, then according to ordering of optimization preference, further compare the second attribute Duration, Duration is the smaller the better, at this moment the Duration optimum of V6, be 1.0, the user area calibration is 20%, and the Duration of V7 2.8 has surpassed 1 * (1+20%)=1.2, has promptly exceeded the scope that the user accepts, therefore, (delete among the N of processing node set never simultaneously in the path that A-＞D) is corresponding, and D joined processing node S set to choose the production of V6 correspondence.

At this moment, the processing node S set be A, C, D}, the node set that is untreated N is { E, F};

Judge whether to exist the reachable path of D, 7 find, do not have be untreated to other production of node of D, promptly do not have any reachable path through tabling look-up to the node set N that is untreated;

Judge whether the node set N that is untreated is empty, find that N is that { E, F} promptly are not empty, continue then to judge whether input node A exists reachable path to the node set N that is untreated, through 7 discoveries of tabling look-up, only there is production: an A-＞F:V8 (0.8,2.8,3.0,3.32,2.25) and corresponding path, do not have other production by comparison this moment, then the path of A-＞F correspondence is optimal path, therefore F is joined processing node S set, and delete among the N of processing node set never, that is: S is { A, C, D, F}, N are { E};

Judge whether to exist the reachable path of node F to the set N that is untreated, find through question blank 7, there is F-＞E, therefore A-＞F and F-＞E are synthesized A-＞F-＞E, be specially: handle according to given QoS weights computational logic, in this example, the addition of preceding two attributes (carrying out cost and time of implementation) value, middle two attributes (reliabilty and availability) value multiplies each other, and the addition of last attribute (reputation) value is averaged, that is:

A-＞F：V8(0.8，2.8，3.0，3.32，2.25)

F-＞E：V(1.2，2.5，2.23，2.46，2)

A-＞F-＞E：V9(2.0，5.3，6.69，8.26，2.13)

Judge whether to exist the reachable path of A, find that through question blank 7 A is unreachable to E to E, therefore, with A-＞F-＞E:V9 (2.0,5.3,6.69,8.26,2.13) upgrading composite vector table and routing table, the composite vector table after the renewal is as shown in table 9, and routing table is as shown in table 10.

Table 9

Table 10

[0180]Judge whether the node set that is untreated is empty, and at this moment, the processing node S set is { A, C, D, F}, the node set that is untreated N for E}, and promptly N be the sky, then continue to judge whether to exist the reachable path of A to the node set N that is untreated, table look-up and 9 find that have the reachable path of A to E, promptly A-＞E:V10 (2.0,5.3,6.69,8.26,2.13), therefore E is joined processing node S set, and delete among the N of processing node set never, that is: the processing node S set is { A, C, D, E, F}, the node set that is untreated N is for empty;

Judge whether to exist the reachable path of E,, then do not have reachable path, needn't upgrade because N is an empty set to the node set N that is untreated;

Node set N is an empty set owing to be untreated, and shows to find to be input as shortest path or the optimal path of A to other all nodes.

Aforesaid operations is all carried out in the input of each row one time at other then, and the composite vector table after the renewal is as shown in table 11, and it is as shown in table 12 to upgrade the rear path table.

Table 11

Table 12

Step 4 is many service processing of carrying out of a set to input:

Each row at the composite vector table (shown in the table 11) after the above-mentioned renewal is carried out a following operation:

Judge whether the input when the pre-treatment row is a set, if then carry out next step operation, otherwise, handle next line;

When handling the 4th row, found that input is a set: { C, D} are output as E,, then, judge whether to exist node arrive described set C, D} forms reachable path, question blank 11 is found to exist: A-＞C, and therefore B-＞D and A-＞D enter next step;

Node A and B are generated input set { A, B}, node A generation input set { A, A} is with { C, D} are generated as the output collection, promptly there is following production: { A, B}-＞{ C, D}, { A, A}-＞{ C, D}, and newly-generated input set and output collection added in composite vector table and the routing table, composite vector table and routing table upgraded.

At first to A, B}-＞C, D} carry out vector to be upgraded, and renewal process is as follows:

A-＞C：(0.5，0.8，1.65，2.23，2)

B-＞D：(0.6，1.5，2.46，2.34，3)

Because A-＞C and B-＞D are concurrency relation (but not the serial of " A-＞C and C-＞D " formula relations), every attribute just can be represented { A, B}-＞{ C, QoS attribute of D} than difference, so every here attribute is got than difference, rather than calculate according to QoS weights computational logic;

As: big more poor more for the first attribute Price, therefore, A, B}-＞C, the Price of D} gets 0.6, for the second attribute Duration, big more poor more, therefore, A, B}-＞C, the Price of D} gets 1.5, and be that to be worth more little expression poor more for reliability, availability and reputation, so remove smaller value respectively: 1.65,2.23,2;

That is: { A, B}-＞{ C, the QoS weight vector of D} are (0.6,1.5,1.65,2.23,2)

Then, to { A, B}-＞{ C, D} is with { C, D}-＞E synthesize renewal, that is:

{C，D}-＞E：(0.8，2.0，1.49，1.65，2.5)

{A，B}-＞{C，D}：(0.6，1.5，1.65，2.23，2)

Calculate according to QoS weights computational logic:

{A，B}-＞{C，D}-＞E：(1.4，3.5，2.46，3.68，2.25)，

Judge whether to exist A or B to the E reachable path, table look-up and 11 find, have A-＞E:(2.0,5.3,6.69,8.26,2.13), according to the scheme selection rule from A, B}-＞C, choose optimal path among D}-＞E and the A-＞E:

To the first attribute Price, the smaller the better, { A, B}-＞{ C at this moment, the Price optimum of D}-＞E is 1.4, and the user area calibration is 10%, and the Price of A-＞E 2.0 has surpassed 1.4 * (1+10%)=1.54, so choose A, B}-＞C, D}-＞E, and upgrade composite vector table and routing table;

For { A, B} be to the optimal path of other each node (X), and each attribute is according to second rule, from A-＞X, B-＞X and { A in the attribute of B}-＞X, chooses a more excellent attribute right value update { A, composite vector table and the routing table of B}-＞X;

For other node (X) to C, the optimal path of D}, each attribute according to the opposite rule of second rule, from X-＞C, X-＞D and X-＞{ C chooses a relatively poor attribute right value update X-＞{ C, composite vector table and routing table of D} in the attribute of D};

The composite vector table that obtains behind the update all is as shown in table 13:

Table 13

Then based on the composite vector table shown in the table 13, to A, A}-＞C, D} carry out vector to be upgraded, step of updating and above-mentioned to A, B}-＞C, the renewal process of D} is in full accord:

According to A-＞C, D}:(0.85,1.0,1.49,1.35,2),

{C，D}-＞E：(0.8，2.0，1.49，1.65，2.5)

Calculate according to QoS weights computational logic that { A}-＞{ C, D}-＞{ vector value of E} is (1.65,3.0,2.22,2.23,2.25)

Will A}-＞and C, D}-＞{ vector value of E} is (1.65,3.0,2.22,2.23,2.25) compare with the optimal case (2.0,5.3,6.69,8.26,2.13) of current A-＞E, find (1.65,3.0,2.22,2.23,2.25) be better than (2.0,5.3,6.69,8.26,2.13), the optimal case that upgrades A-＞E is that A-＞{ the refreshing weight vector is (1.65 for C, D}-＞E, 3.0,2.22,2.23,2.25).

Judge to find to exist A, the optimal case of B}-＞E (1.4,3.5,2.46,3.68,2.25) compares with the optimal case (1.65,3.0,2.22,2.23,2.25) of A-＞E:

The first priority price is compared, and price is the smaller the better, at this moment { A, the price optimum of B}-＞E, be 1.4, the user area calibration is 10%, and another price has all surpassed 1.4 * (1+10%)=1.54, exceed the user and accept scope, therefore, current { A, B}-＞{ C, D}-＞{ E} has been optimal case, needn't upgrade.

Obtain optimum composite vector table after upgrading this moment, as shown in table 14:

Table 14

Routing table after the renewal, as shown in Table 15:

Table 15

At last, when the user input query condition, as: input set and output set, the service synthesis system is question blank 14 or table 15 automatically, will offer the user with the corresponding shortest path of querying condition (i.e. Zui You Web service synthetic schemes).

What deserves to be explained is, carry out the modeling processing with the production modeling in the foregoing description and obtain composite vector table and routing table, be the preferred embodiments of the present invention, those skilled in the art should understand other modeling pattern in addition, as attribute binding modeling, also can obtain composite vector table and routing table.Described attribute binding modeling is that Web Service is regarded as a class, its input I, and output O is as the attribute of class, and (I O), adds before each input attributes name one " $ ", represents that this attribute is an input variable to be expressed as S.As shown in figure 11, a service 1 has two input parameter A and B, and output parameter is C, with the input and output attribute of " $ " Differentiated Services 1.

In sum, a kind of quality-of-service based service synthesis system and service synthetic method according to embodiment of the invention proposition, by the production modeling input and output of Web service are represented with path model, the synthetic process of service is exactly the routing update process, by choosing the path according to Scheme Choice rule (user is to the ordering of optimization preference and the discrimination of each qualitative attribute of service), and do not need exhaustive all synthetic schemes, and reduced time complexity, improved system works efficient.

The above is a preferred implementation of the present invention; should be pointed out that for those skilled in the art, under the prerequisite that does not break away from the principle of the invention; can also make some improvements and modifications, these improvements and modifications also are considered as protection scope of the present invention.

Claims

1. the service synthesis system based on service characteristic is characterized in that, comprising:

Memory module is used for store path table and composite vector table;

Initialization module handles routing table and composite vector table in the initialization memory module by Web service set being carried out modeling;

Synthesis module is used for selecting a paths according to the Scheme Choice rule from the mulitpath that is input to output, and uses routing table and composite vector table in the described memory module of selected routing update;

Wherein, describedly from the mulitpath that is input to output, select a paths according to the Scheme Choice rule and be specially:

Obtain the ordering of optimization preference of user to the QoS attribute;

Choose first attribute in the described ordering;

Whether the difference of judging the described attribute of two paths surpasses discrimination;

If described judged result is for being, then according to the property value of two paths, according to choosing a wherein paths at first rule of Qos attribute setting in advance;

If described judged result is then chosen the next attribute in the described ordering for not, return the step whether difference of carrying out the described attribute of judging two paths surpasses discrimination.

2. service synthesis system as claimed in claim 1 is characterized in that,

Described routing table, at least one is input to the pairing routing information of QoS cost vector of at least one output to be used for storing the current building-up process of obtaining;

Described composite vector table, at least one is input to the cost vector information of at least one output to be used for storing the current building-up process of obtaining.

3. service synthesis system as claimed in claim 1 is characterized in that, described initialization module comprises:

The set of service unit is used for storing the relation information of Web service set and this each Web service of set, and described relation information comprises the incidence relation between each Web service at least;

Serve the weights aggregation units, be used to store the QoS weights information of described each Web service;

Many standard evaluations unit is used for the QoS weights information according to service weights aggregation units, by the method that adopts the quality vector set of the Web service in the set of service unit is estimated, and is generated the set of service that contains the QoS vector;

The production modeling unit is used for the set of service that contains the QoS vector that many standard evaluations unit generates is carried out the production modeling, generates the modeling table;

Synthetic initialization unit is used for routing table and composite vector table according to the described memory module of modeling table initialization of production modeling unit generation.

4. service synthesis system as claimed in claim 1 is characterized in that, described synthesis module comprises:

Weights computational logic unit is used to store the weights computational logic, the computational methods of corresponding each QoS attribute of this weights computational logic;

Ordering of optimization preference and discrimination unit are used to store the ordering of optimization preference and the discrimination of each QoS attribute, and described ordering of optimization preference is that the user sets the preference ordering to each QoS attribute; Described discrimination is used for representing the difference on each QoS attribute of user's acceptable;

Synthesis unit is used for synthesizing calculating according to described weights computational logic and ordering of optimization preference and discrimination, upgrades routing table and composite vector table in the described memory module, obtains set of paths.

5. as any described service synthesis system of claim 1 to 4, it is characterized in that described service synthesis system also comprises:

Many service processing module are used for that described synthesis module is upgraded the routing table and the composite vector table that obtain and carry out many service processing to upgrade routing table and composite vector table.

6. service synthesis system as claimed in claim 5 is characterized in that, described service synthesis system further comprises:

Enquiry module, the input set and the output set that are used for providing according to the user are inquired about through composite vector table and routing table after synthesis module and the processing of many service processing module, to obtain and described input set and corresponding path of output set and vector value.

7. the service synthetic method based on service characteristic is characterized in that, comprising:

By being carried out the modeling processing, the Web service set obtains initial routing table and composite vector table;

From the mulitpath that is input to output, select a paths according to the Scheme Choice rule, and use described routing table of selected routing update and composite vector table;

Obtain the ordering of optimization preference of user to the QoS attribute;

Choose first attribute in the described ordering;

If described judged result is for being, then according to two property value, according to choosing a wherein paths at first rule of Qos attribute setting in advance;

8. service synthetic method as claimed in claim 7 is characterized in that, described modeling is handled and comprised:

Production modeling or attribute binding modeling.

9. service synthetic method as claimed in claim 8 is characterized in that, described modeling is treated to the production modeling when handling, and describedly handles the step that obtains initial routing table and composite vector table and specifically comprises by the Web service set being carried out modeling:

Generate the set of service that contains the QoS vector according to Web service aggregate information and QoS weights information;

Set of service to the described QoS of containing vector is carried out the production modeling, generates the modeling table;

Generate initial routing table and composite vector table according to described modeling table.

10. service synthetic method as claimed in claim 7, it is characterized in that, describedly from the mulitpath that is input to output, select a paths, and use the step of described routing table of selected routing update and composite vector table specifically to comprise according to the Scheme Choice rule:

Each row at described composite vector table is carried out following operation:

Generate the processing node set and the node set that is untreated;

All output nodes of the described composite vector table of initialization are in the node set that is untreated;

The pre-treatment row input node of working as of described composite vector table is joined processing node set;

Judge whether described input node exists reachable path to the node set that is untreated, if, then from described reachable path, select first path according to the Scheme Choice rule, and the output node in first path joined processing node set, the output node in described first path of deletion in the processing node set never;

Whether the output node of judging described first path exists reachable path to the node set that is untreated, if then select second path according to the Scheme Choice rule from described reachable path;

Described first path and second path are synthesized the Third Road footpath;

Judge whether the input node exists reachable path to the output node in described Third Road footpath, if, then from described reachable path and Third Road footpath, select the 4th path according to the Scheme Choice rule, and with described composite vector table of the 4th routing update and routing table, otherwise, directly upgrade described composite vector table and routing table with described Third Road;

Judge whether the node set that is untreated is empty, if not, then return to carry out to judge whether described input node exists the step of reachable path to the node set that is untreated, otherwise, this step finished.

11. as any described service synthetic method in the claim 7 to 10, it is characterized in that, from the mulitpath that is input to output, select a paths according to the Scheme Choice rule, and using the step of described routing table of selected routing update and composite vector table also to comprise afterwards:

Described routing table and composite vector table are carried out many service processing further upgrade routing table and composite vector table.

12. service synthetic method as claimed in claim 11 is characterized in that, describedly described routing table and composite vector table are carried out many service processing further upgrades the step of routing table and composite vector table and specifically comprises:

Judge whether the input when the pre-treatment row is a set, if then carry out next step;

Judge whether to exist node to arrive described set and form reachable path, if then carry out next step;

Described node is generated as input set, and described set is generated as the output collection, adds in described composite vector table and the routing table;

When there was reachable path in three nodes of the node to the in the described input set, at each QoS attribute, the attribute of choosing described reachable path according to second rule upgraded the QoS attribute of described input set to the path of described the 3rd node;

When there was reachable path in the node of concentrating to described output when the 4th node, at each QoS attribute, the attribute of choosing described reachable path according to the rule opposite with second rule upgraded the QoS attribute of described the 4th node to the path of described output collection.