CN103034963A - Service selection system and selection method based on correlation - Google Patents

Abstract

The invention relates to a service selection system based on correlation. The service selection system comprises a release module, a retrieval module, an evaluation module, a database module and a selection module, and service information is registered and released through utilization of universal description, discovery and integration (UDDI). The database module sorts atomic service sources according to evaluation data provided by the evaluation module, and the selection module gives preference to the atomic service sources with high evaluation to show to users. The invention relates to the selection method of the service selection system at the same time. The service selection system and the selection method change a traditional service selection mode, break through the limitation that credibility is not high in searching of similar neighbor users in the prior art, avoid possible joint cheating situations through the selection process of correlation considering service between comprehensive services, can enable selected results to be accurate, support dynamic characteristic of an open network environment, can adapt to demands of the open network environment, effectively improves efficiency of service selection, and also avoids conditions that data provided by service providers are not credible.

Description

A kind of service selection system and system of selection based on correlativity

Technical field

The present invention relates to a kind of inquiry system of service class product, concrete, under the complex network environment of an opening, isomery, make up trusted users alliance and carry out the system of services selection according to the correlativity between service, the invention discloses simultaneously this system alternatives.

Background technology

Open networking is used and the theory of " software is as service " will cause Main Morphology, the method for operation, the mode of production and use-pattern based on software systems under the Internet environment that huge variation occurs.The service that a large amount of functions is identical, nonfunctional characteristics is different that distributing on the network does not often have single service can directly satisfy user's demand when user request service, at this moment just need to select some atomic service and carry out polymerization.The polymerization process of service can represent that with atomic service each atomic service is a virtual service module, and its embodies is relation on the service logic, will be replaced by the concrete atomic service of reality in polymerization process.The selection course of service depends on polymerization history, and existing relevant selection scheme great majority all are that the selection course of will serve is independent of polymerization process at present, have ignored the correlativity between service.In a word, existing research or lack the theoretical degree of depth does not perhaps provide the model or the algorithm that gear to actual circumstances, thereby all is not applicable to large-scale open network environment very much.

Summary of the invention

Goal of the invention: the purpose of this invention is to provide a kind ofly opening, provide under the heterogeneous network environment high-level efficiency, the accurate effectively reliable system of services selection, the invention discloses simultaneously this system alternatives.

Technical scheme: the present invention is realized by following technical solution: a kind of service selection system based on correlativity, comprise release module, retrieval module, evaluation module, database module and selection module, retrieval module provides retrieval port for user's access, comprise all relevant atomic service resources in the database and to the evaluating data of these atomic service resources, evaluation module provides evaluation mechanism so that the user estimates the service of accepting or used, evaluating data is sent into database and is preserved, select module to provide close with it Service Source for reference according to the key word of user search, release module provides interface for service provider's issuing service information, complete when information acceptance, release module is stored in information on services in the database.

Described information on services utilizes UDDI to register issue.

Database root sorts to the atomic service resource according to the evaluating data that evaluation module provides, and selects module preferentially to select to estimate high atomic service resource and shows to the user.

A kind of system of selection of selective system comprises the steps:

1) retrieval port that provides by retrieval module of user is submitted the key word of wish inquiry to selective system;

2) key word is sent in the database, selects the Service Source data of coupling from database;

3) according to key word, select module to select and the immediate Service Source of key word;

4) retrieval module feeds back to the user with the resource of coupling by retrieval module.

Also comprise evaluation procedure, after the user accepts service, can write evaluation information to this service by evaluation module, described evaluation mechanism for scoring mechanism with write comment in conjunction with system.

Comprise that also other users are to the evaluation of this resource in the Service Source that the resource that database returns and selection module are recommended.

Beneficial effect: the present invention compared with prior art, changed traditional services selection pattern, break through prior art not high limitation of credibility when seeking similar neighbor user, consider the selection course of service by the correlativity between the integrated service, avoided contingent associating fraud scenario, can make the result of selection more accurate; Support the dynamic perfromance of the dynamic perfromance assurance network enabled of open network environment, adapt to the demand of open network environment; Effectively improve the efficient of services selection, also can avoid the service provider that the generation of the insincere situation of data is provided; Scheme has stronger applicability.

Description of drawings

Fig. 1 is the structural drawing of service selection system of the present invention.

Fig. 2 is the frame construction drawing that the credible alliance of the present invention makes up.

Fig. 3 is that the present invention serves a correlativity synoptic diagram.

Fig. 4 is that the present invention's services set commonly used is extracted synoptic diagram.

Embodiment

Below in conjunction with Figure of description the present invention is described in further detail:

The present invention relates to a kind of service selection system based on correlativity, Fig. 1 has provided the structural drawing of service selection system, comprise release module, retrieval module, evaluation module, database module and select module, wherein release module is that the service provider passes through service description language (sdl), function and the interface message of the service that provides are provided, and will be served the Integration with UDDI(Universal Description Di scovery and) register issue to service register center; Retrieval module is that the service user retrieves the description document of finding own needed service and obtaining respective service in service register center; Evaluation module is that the service user estimates the service quality of using provider in the past, and evaluation information is stored in the database module, and comprehensively these two modules are protected forming a believable service user alliance; Selecting module is that the service user is historical according to its other users' service interaction; under the prerequisite that extracts services set commonly used; select the atomic service that can be complementary with its demand; the correlativity that exists between Reference Services is carried out services selection protect, among the figure shown in the dotted line frame be the components of system as directed of independent and traditional services selection scheme.

Traditional method for service selection is a kind of simple request response process between service user and the service provider, the two binding is often based on service quality QOS's, yet increasing along with quantity of service in the resource pool, service is searched and is selected to be difficult to guarantee the authenticity of data that the service provider provides based on QOS, the credibility of service user's QOS feedback data also can't be protected simultaneously, therefore traditional method Shortcomings aspect the accuracy of services selection.

And according to the evaluation module in the service selection system that proposes and the information in the database module, we can break through traditional method, centered by the service user, have other users that same services is used experience by finding out for the destination service user, be that the destination service user makes up a believable user coalitions according to the trusting relationship between them, service by other users in the alliance carries out service recommendation for the destination service user with history, thereby finish follow-up service selection process, consider services selection from a kind of alliance angle, both can guarantee the credibility of service user's information, that can avoid also that the service provider claims data can not implementations, and the below provides definition and the method basic in the credible alliance process set up:

What 1) credible alliance (Trus tworthy Community) formed in service selection system can guarantee the similarity degree of user aspect the service usage behavior, also can guarantee user's set of its behavior creditability, namely the user in the credible alliance is believable similar neighbor user.

2) attributive character of the attributive character (Service Characteristics) of service service is to embody a series of attributes of service unique characteristics, can be used for personalized service, also can react the user to its satisfaction, comprise credit worthiness and the frequency of utilization of service.The credit worthiness (reputat ion) of service refer to the use of collecting the user of this service to its satisfaction, and frequency of utilization (frequency) refers to that user in the time window in the past uses the number of times of this service.

3) recommend the trust (Trust Of Recommendation) of behavior in service selection system, the trust of recommendation behavior refers to that take the user as core the presentee recommends a kind of subjective determination behavior of honesty, reliability and the validity of behavior to the nominator.

4) recommend the direct trust (Direct Trust Of Recommendat ion) of behavior to recommend user and the targeted customer a kind of trusting relationship by the mutual acquisition of direct information between the two, namely each other to the degree of recognition of the mutual recommended project.

5) recommend trust colony that the indirect trust (Indi rect Trust Of Recommendat ion) of behavior refers to the targeted customer to recommending the transmission of user's credit worthiness, the indirect trust approval relation that namely obtains by the transmission to the direct trusting relationship of recommending behavior.

6) recommending the degree of belief of degree of belief (Trust Degree Of Recommendation) the recommendation behavior of behavior is the tolerance that is present in a kind of point-to-point trusting relationship between presentee and the nominator, is made of the direct degree of belief of the behavior of recommendation and indirect degree of belief two parts of the behavior of recommendation.

Fig. 2 has provided a frame construction drawing that makes up credible alliance, this frame construction drawing mainly is comprised of three parts: at first be the consideration that adds the Service Properties feature, to in the traditional algorithm between the user computing formula of similarity improve, next introduces the trusting relationship between the user, consider the genuine and believable degree of its information by calculating the users to trust degree, the two carries out user assignment last comprehensive similarity and degree of belief, determines targeted customer's similar neighbor user, makes up credible alliance.

A common service recommendation system is by M service requester { U ₁, U ₂... U _mAnd N candidate service { S ₁, S ₂... S _nForm, the user journal by service requester can extract its service evaluation information, and namely service requester can represent with the matrix of a M * N with relation between each service, wherein the element R that is listed as of the capable s of i _{I, s}Expression be service requester i to the evaluation of estimate of the service a certain QoS property value of s (such as response time, success ratio etc.), the higher representative of consumer of scoring is more satisfied to the quality of this service usually, if user i never called service s, so R _{I, s}Then be 0.The set of service that user i and j marked jointly is s _{I, j}, the similarity Sim (i, j) of user i and j can by formula 1 calculate:

$\mathrm{Sim}(i,j)=\frac{\underset{s\∈s_{i,j}}{\mathrm{\Σ}}\sqrt{\mathrm{\μ}\frac{1}{{\mathrm{reputation}}_s^2}+\mathrm{\ω}\frac{1}{{\mathrm{frequency}}_s^2}}(R_{i,s}-\stackrel{\‾}{R_i})\×(R_{j,s}-\stackrel{\‾}{R_j})}{\sqrt{\underset{s\∈s_{i,j}}{\mathrm{\Σ}}{(R_{i,s}-\stackrel{\‾}{R_i})}^2\sqrt{\underset{s\∈s_{i,j}}{\mathrm{\Σ}}{(R_{j,s}-\stackrel{\‾}{R_j})}^2}}}---\left(1\right)$

Wherein

With

Represent that respectively user i and j are to s _{I, j}In the average of scoring of all services.Parameter μ and ω are the factors of influence of two attributive character of service, and span is [0,1].Because the residing applied environment difference of service can cause two attributive character that corresponding variation is also occured the proportion that result of calculation exerts an influence, therefore, variable parameter μ can be so that this similarity calculating method can be applicable to different applied environments better from ω.

Next consider the credibility of user behavior: a kind of channel that effectively obtains the direct degree of belief of recommendation behavior is that the prestige of estimating each other the other side after mutual by user-user information is set up.Calculate the direct degree of belief RDT (i, j) that recommends behavior between two users with the Beta trust model.If user j and i all with service s mutual mistake, R _{I, s}Expression user i is to the scoring of service s, R _{J, s}Expression user j is to the scoring of service s, j is recommended i as independent neighbours of i, and j compared to the true score value of serving to score value and the i of service, if the Error Absolute Value between them is less than a certain fixed value ε, think that then j is correct to the recommendation of i, otherwise think that then it is wrong recommending.The set of service s that jointly marked at both _{I, j}In, user j is designated as pi for the correct total degree of recommending of user i, j, and the total degree that mistake is recommended is designated as ni, j.The Beta probability density function is applicable to describe and comprises the scale-of-two event Posterior probability, to refer to user j be that i makes correct service recommendation to the x event here, the formula of the direct degree of belief of calculated recommendation behavior is as follows:

$\mathrm{RDT}(i,j)=\frac{\mathrm{\Γ}(p_{i,j}+n_{i,j}+2)}{\mathrm{\Γ}(p_{i,j}+1)\mathrm{\Γ}(n_{i,j}+1)}{x}^{p_{i,j}}{(1-x)}^{n_{i,j}}---\left(2\right)$

Wherein Γ is the gamma function, 0＜＜x＜＜1.

During the indirect trust of calculated recommendation behavior, establish set D and be the trust customer group of targeted customer i, by the trust user in the set D to recommending user j to carry out trust evaluation, and in conjunction with separately trust weight as the assessment foundation.RIDT (i, j) expression user i is to the indirect degree of belief of the recommendation behavior of j, and computing formula is as follows:

$\mathrm{RIDT}(i,j)=\frac{\underset{D}{\mathrm{\Σ}}{w}_i\mathrm{RDT}(j,D_i)}{\underset{D}{\mathrm{\Σ}}{w}_i}---\left(3\right)$

In the formula, w _iTrust the user D that gathers in the D for the targeted customer _iTo the trust weight of i recommendation user j, its value is the direct degree of belief that two users recommend behavior, calculates gained by formula 2.RDT (j, M _i) be user D _iAnd the direct degree of belief between the user j.

Obtain recommending after the direct trust and indirect degree of belief of behavior, the user recommends the degree of belief of behavior to be calculated by formula 4:

RT(i,j)=αRDT(i,j)+βRIDT(i,j) （4）

In the formula, the direct degree of belief of RDT (i, j) expression recommendation behavior, the indirect degree of belief of RIDT (i, j) expression recommendation behavior, α, β represent respectively its weight, span is [0,1], and alpha+beta=1.Obviously, RT (i, j) value is larger, illustrates that user j is more credible to the recommendation of targeted customer i.

When integration objective user and candidate neighbours' similarity and degree of belief, the comprehensive weight computing formula is as follows:

$\mathrm{weight}(i,j)=\frac{2\mathrm{sim}(i,j)\·\mathrm{RT}(i,j)}{\mathrm{sim}(i,j)+\mathrm{RT}(i,j)}---\left(5\right)$

When carrying out user assignment because the user is many in the actual network environment, between contact also relatively tight, we take traditional top-k method, choose a front k user according to the comprehensive weight value and form credible alliance.

On the basis of credible coalition formation; the work of Analysis Service user's Relations Among finishes; the work of selecting module in the system namely is historical with reference to the service interaction of user in the credible alliance; the discovery, selection and the polymerization process that comprise each service; analyze and consider existing correlativity between the different services; provide the candidate service that can satisfy its demand for the service user according to this potential correlativity; the process that final completion service is selected, this is need protection another aspect of technology of this patent.The below provides according to basic definition and the method for potential correlativity when carrying out services selection between service:

1) services set s commonly used _cs _cWhat represent is in the certain hour section, historical according to the service aggregating of all users in the credible alliance, and the most frequently used sub-services set of carrying out service aggregating of the user who selects is expressed as s with it _c={ s _C1, s _C2... s _Cm, wherein the demand that other services can both better meet the user is compared in each service.

When 2) candidate service collection c user carried out services selection, the candidate service collection of meeting consumers' demand in the Service Source pond was expressed as c={c ₁, c ₂... c _m.

3) service between correlativity R in service discovery, polymerization and selection course, the specific contact of certain that has between the service or pattern, comprise candidate service collection and service centralized services commonly used and and whole resource pool in service between the size of interrelated degree, this service that the larger explanation candidate service of the value of R is concentrated more can be near user's demand.

4) alliance's principle (Association Rules) is the most important a kind of technology of Data Mining, normally is used for finding the contact between the entity in the large-scale database.Can be expressed as A → B with formula, wherein

And A ∩ B=Ф, the concrete meaning of formula is: A and B belong to a set I, and we have very large probability can find article among the B in finding A in the article.Two very important key concept support and confidence are arranged in alliance's principle.The support of the principle A → B of alliance represents the frequency that this alliance's principle occurs, and what the confidence of the principle A → B of alliance represented is the probability that also can find entity B when finding entity A, and it embodies the contiguity between A and the B.

Fig. 3 has provided the synoptic diagram that has correlativity between service, and two service user U are arranged in the credible alliance ₁And U ₂, they are the same to the atomic service polymerization process of a certain services request S, but work as them to S ₂And S ₅When carrying out concrete sub-services mapping, user U ₁That select is sub-services S ₂₁And S ₅₁, and user U ₂That select is sub-services S ₂₂And S ₅₂, just because of U ₁And U ₂Therefore belong to a user coalitions, they are very high at the similarity degree of service use, can think sub-services S ₂₁, S ₅₁, S ₂₂, S ₅₂Between have certain specific contact or pattern, that is to say the correlativity between potential service, with reference to this correlativity, can realize services selection for the destination service user.

Fig. 4 has provided dynamically, under the heterogeneous network environment, historical by the service aggregating of user in the comprehensive credible alliance, analyze whole service discovery, selection and polymerization process, thereby extract the synoptic diagram of services set commonly used, concrete grammar is expressed as follows:

With user-service call rate as forming s _cReference factor.When a certain user k in the credible alliance selected service j, we recorded this and call, and the invoked total degree of service j is designated as number (s _j, u _k), our 6 rates of calling that calculate service by formula after a period of time are designated as freq (s with it _j, u _k), then user in the credible alliance-service call rate is sorted, select the highest m of numerical value sub-services as the element in the services set commonly used.

$\mathrm{fred}(s_j,u_k)=\left\{\begin{array}{cc}\frac{\mathrm{number}(s_j,u_k)}{\underset{j=1}{\overset{m}{\mathrm{\Σ}}}\mathrm{number}(s_j,u_k)}& \mathrm{ifnumber}(s_j,u_k)\≠0\\ 0& \mathrm{ifnumber}(s_j,u_k)=0\end{array}\right.---\left(6\right)$

A certain service c among the c _iWith s _cBetween correlativity can calculate by alliance's principle that assessment exists between the two.The alliance's principle that exists is more, c _iWith s _cCorrelativity then stronger; We are with c in addition _iWith the correlativity of all set of services in the resource pool also as the important reference factor of another one, c _iAnd the alliance's principle that exists between s is more, c _iWith s _cCorrelativity then more weak.This paper R ₁And R ₂Represent respectively this two kinds of correlativitys, correlativity is defined as R=R between will serving ₁* R ₂, the polymerization history of all services calculates among the candidate service collection c each service and commonly uses services set s in the contact resource pool _cThe support of middle service and confidence value.Get rid of and do not satisfy thresholding confidence (s _c→ c)〉and δ, sup port (s _c→ c)〉candidate service of γ, wherein the value of δ and γ changes according to concrete model and environment.In satisfying the candidate service of threshold value, all calculate c according to formula 7 _iWith s _cThe correlativity R of middle service ₁

${\mathrm{<figure-callout\; id="1"\; label="R"\; filenames="HDA00002485691300012.png,HDA00002485691300021.png"\; state="\{\{state\}\}">R</figure-callout>}}_1=\underset{j=1,s_{\mathrm{cj}}\&Element;S_c}{\overset{m}{\mathrm{\&Sigma;}}}\mathrm{confidence}(s_{\mathrm{cj}}\&RightArrow;c_i)---\left(7\right)$

Calculate c according to formula 8 again _iCorrelativity R with all services ₂, wherein t is the number of all principle AR of alliance of existing in the Service Source pond, num (c _i, AR) expression is to comprise c _iThe number of the principle AR of alliance that participates in.

$R_2=\log \frac{t+1}{\mathrm{num}(c_i,\mathrm{AR})+0.5}---\left(8\right)$

The then correlativity R between calculation services, the candidate service that rreturn value is the highest, user selection its as optimal service.

So far just successfully use alliance's principle in candidate service collection c, to select the service that to meet consumers' demand.

We have realized to select the process of serving for the destination service user by the potential correlativity between Analysis Service on the basis that credible alliance makes up.For the atomic service that selects, upgrade it in the UDDI registration center in service selection system and release news simultaneously, when other service user's request services, can reduce the time overhead of selection course.

For the services selection scheme based on correlativity of the present invention is described, we have provided, and a concrete example---the user elaborates the method to the best tourist service that network request goes to Nanjing to play.

The first step, the service provider that tourism is relevant passes through trusted third party, in the atomic service resource of network issue about Nanjing Tourism, such as: tourist attractions, tourism route, trip mode, accommodation service, food and beverage sevice, financial service and insurance service etc., these Service Sources form whole Service Source pond.

Second step, the user proposes the request of Nanjing Tourism Service Source to the service access platform, searches the candidate service of the demand of satisfying condition.The content of searching comprises the kind (can be categorized as in the present case: commercial, high, medium, common) of service, security (as: trip mode security, food safety grade etc.), reliability (as: prestige of travel agency, the punctuality rate of the vehicles etc.) and the price (common price and favourable price etc.) of service, through some constraints preliminary screening is carried out in service, the candidate service resource pool that formation can supply the user to select.

The 3rd step, the user experiences according to tourism in the past, seek out a part and same or analogous other users of own preference, for example in travel agency, the sight spot, the selection of the aspect such as mode of transportation and lodging all with own identical that a group user, they namely are the similar neighbor user of targeted customer aspect the tourism behavior, if this type of user once carried out direct or indirect tourism recommendation to the targeted customer simultaneously, consider again the trusting relationship that exists between them, in conjunction with before analyze drawn similarity relation, determine not only to the user similar aspect the tourism behavior but also can by the user group of its trust, be decided to be user's credible alliance.

The 4th step, if the user in the credible alliance once arrived the Nanjing Tourism mistake, they had use experience to each tourist attractions, each bar tourism route, various trip mode, each lodging and the food and drink etc. in Nanjing, and they also have comparatively comprehensively evaluation to the q﹠r of these services, according to their feedback, we extract some the most satisfied services of most users in the alliance, formation is about the services set commonly used of tour schedule, considers that as the targeted customer one of service relevance is with reference to the group when carrying out tourist service and choose.

The 5th step, obtain in previous step on the basis of services set commonly used, we carry out related reference with the service in the candidate service resource pool with the service in the services set commonly used, all travel informations in conjunction with the user that once played to Nanjing, seek some the fixing contact that exists between these services, whether all once jointly once called and so on during the journey simultaneously by certain some user such as the service in two services set, utilize alliance's principle to find contact and pattern between these services, that is to say the correlativity that exists between service, the similar correlativity that can find out between serving in candidate service collection and the whole Service Source pond.

The 6th step, by analyzing the also correlativity between calculation services, finally can select that group candidate service that can satisfy targeted customer's demand, the user carries out the Nanjing Tourism plan with reference to this group service, and its request is solved.User's services selection result also is updated to the information that other users can reference, for all kinds of services selection of carrying out faster, more high-quality of the user to Nanjing Tourism from now on provide the reliability guarantee, also improved the efficiency of selection of tourist service in the whole network.

Claims (6)

1. service selection system based on correlativity, it is characterized in that: comprise release module, retrieval module, evaluation module, database module and selection module, retrieval module provides retrieval port for user's access, comprise all relevant atomic service resources in the database and to the evaluating data of these atomic service resources, evaluation module provides evaluation mechanism so that the user estimates the service of accepting or used, evaluating data is sent into database and is preserved, select module to provide close with it Service Source for reference according to the key word of user search, release module provides interface for service provider's issuing service information, complete when information acceptance, release module is stored in information on services in the database.

2. selective system according to claim 1, it is characterized in that: described information on services utilizes UDDI to register issue.

3. selective system according to claim 1, it is characterized in that: database root sorts to the atomic service resource according to the evaluating data that evaluation module provides, and selects module preferentially to select to estimate high atomic service resource and shows to the user.

4. the according to claim 1 system of selection of described service selection system based on correlativity is characterized in that: comprise the steps:

1) retrieval port that provides by retrieval module of user is submitted the key word of wish inquiry to system;

2) key word is sent in the database, selects the Service Source data of coupling from database;

3) according to key word, select module to select and the immediate Service Source of key word;

4) retrieval module feeds back to the user with the resource of coupling by retrieval module.

5. system of selection according to claim 4 is characterized in that: after comprising that also evaluation procedure, user are accepted service, can write evaluation information to this service by evaluation module, described evaluation mechanism for scoring mechanism with write comment in conjunction with system.

6. according to the described system of selection of claim, it is characterized in that: comprise that also other users are to the evaluation of this resource in the Service Source that the resource that database returns and selection module are recommended.

Images