CN103034963B - A kind of service selection system and system of selection based on correlation - Google Patents
A kind of service selection system and system of selection based on correlation Download PDFInfo
- Publication number
- CN103034963B CN103034963B CN201210494558.0A CN201210494558A CN103034963B CN 103034963 B CN103034963 B CN 103034963B CN 201210494558 A CN201210494558 A CN 201210494558A CN 103034963 B CN103034963 B CN 103034963B
- Authority
- CN
- China
- Prior art keywords
- mrow
- service
- msub
- user
- services
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The present invention relates to a kind of service selection system based on correlation, including release module, retrieval module, evaluation module, database module and selecting module, the information on services carries out registration issue using UDDI.The evaluating data that database root is provided according to evaluation module is ranked up to atomic service resource, and selecting module prioritizing selection is evaluated high atomic service resource and shown to user.Present invention simultaneously relates to the selection system alternatives.The present invention changes traditional services selection pattern, break through prior art credible not high limitation when finding similar neighborhood user, the selection course of service is considered by correlation between integrated service, it is to avoid the joint fraud scenario that may occur, the result of selection can be made more accurate；The dynamic characteristic of open network environment is supported, the demand of open network environment is adapted to；The efficiency of services selection is effectively improved, service provider's data presented can also be avoided the occurrence of insincere.
Description
Technical field
The present invention relates to a kind of inquiry system for servicing class product, specifically, in an opening, the complex network ring of isomery
Under border, the system for building trusted users alliance and services selection being carried out according to the correlation between service, while the invention discloses
The system alternatives.
Background technology
Open networked instruments and the theory of " software is used as service " will cause to be based on software under Internet environment
Huge change occurs for Main Morphology, the method for operation, the mode of production and the occupation mode of system.Substantial amounts of work(is dispersed with network
The different service of the identical, nonfunctional characteristics of energy, often use can be directly met when user asks and serviced without single service
The demand at family, is at this moment accomplished by selecting some atomic services and is polymerize.The polymerization process of service can be with atomic service come table
Show, each atomic service is a virtual service module, what it embodied is the relation on service logic, in polymerization process
It is middle to be replaced by the specific atomic service of reality.The selection course of service is dependent on polymerization history, related choosing existing at present
It is all that, independently of polymerization process, have ignored the correlation between service by the selection course of service to select most of scheme.In a word,
Some research either lacks theoretical depth or does not provide the model or algorithm geared to actual circumstances, thus is all less well-suited to big rule
The open network environment of mould.
The content of the invention
Goal of the invention：High efficiency is provided under opening, heterogeneous network environment, accurately have it is an object of the invention to provide one kind
The system for imitating reliable services selection, while the invention discloses the system alternatives.
Technical scheme：The present invention is realized by following technical solution：A kind of service selection system based on correlation,
Including release module, retrieval module, evaluation module, database module and selecting module, retrieval module provides retrieval port and supplied
User is accessed, and all related atomic service resources and the evaluating data to these atomic service resources are included in database, is commented
Valency module provides evaluation mechanism so that user is evaluated the service for having received or having used, and evaluating data feeding database enters
Row is preserved, for reference, release module that selecting module provides close Service Source therewith according to the keyword of user search
Interface is provided and supplies service provider's issuing service information, when information receives to finish, information on services is stored in data by release module
In storehouse.
The information on services carries out registration issue using UDDI.
The evaluating data that database root is provided according to evaluation module is ranked up to atomic service resource, and selecting module is preferentially selected
The high atomic service resource of evaluation is selected to show to user.
One kind selection system alternatives, comprises the following steps：
1）User by retrieve module provide retrieval port to selection system submit the keyword to be inquired about；
2）In keyword feeding database, the Service Source data of matching are selected from database；
3）According to keyword, selecting module is selected and the immediate Service Source of keyword；
4）Module is retrieved by the resource of matching by retrieving module feedback to user.
Also include evaluation procedure, user receives after service, and the evaluation information to the service, institute can be write by evaluation module
Commentary valency mechanism is scoring mechanism and writes comment combination system.
Also the resource is commented including other users in the Service Source that the resource and selecting module that database is returned are recommended
Valency.
Beneficial effect：The present invention compared with prior art, changes traditional services selection pattern, breaks through prior art and exists
Credible not high limitation during similar neighborhood user is found, the selection of service is considered by the correlation between integrated service
Journey, it is to avoid the joint fraud scenario that may occur, can make the result of selection more accurate；Support the dynamic of open network environment
Characteristic ensures to support the dynamic characteristic of network, the demand of adaptation open network environment；The efficiency of services selection is effectively improved,
Service provider's data presented can be avoided the occurrence of insincere；Scheme has stronger applicability.
Brief description of the drawings
Fig. 1 is the structure chart of service selection system of the present invention.
Fig. 2 is the frame construction drawing that credible alliance of the invention builds.
Fig. 3 is dependency diagram between present invention service.
Fig. 4 is that the conventional services set of the present invention extracts schematic diagram.
Embodiment
The present invention is described in further detail with reference to Figure of description：
The present invention relates to a kind of service selection system based on correlation, Fig. 1 gives the structure chart of service selection system,
Including release module, retrieval module, evaluation module, database module and selecting module, wherein release module is that service is provided
Business describes the function and interface message of service provided by service description language (sdl), and by service UDDI（Universal
Description Di scovery and Integration）Registration issue is carried out to service register center；Retrieving module is
Service user retrieves the service required for discovery oneself in service register center and obtains the description file of respective service；Evaluate
Module is that service user is evaluated the service quality for using provider in the past, and evaluation information storage is arrived into database
In module, the two comprehensive modules are protected to forming a believable service user alliance；Selecting module is that service makes
User is according to the service interaction history of its other users, and on the premise of conventional services set is extracted, selecting most can be with its demand
The atomic service matched, carries out services selection to the correlation existed Reference Services and protects, institute in dotted line frame in figure
The components of system as directed for independence and traditional services selection scheme shown.
Traditional method for service selection is a kind of simple request response process between service user and service provider,
Binding between both is often based on service quality QOS, however as in resource pool quantity of service it is increasing, be based on
QOS is searched and selected to be difficult to ensure that the authenticity of service provider's data presented to service, while service user
The credibility of QOS feedback data can not also be protected, therefore traditional method exists not in terms of the accuracy of services selection
Foot.
And the information in the evaluation module and database module in the service selection system proposed, we can dash forward
Broken traditional method, centered on service user, experience is used by being found out for destination service user with same services
Other users, according to the trusting relationship between them be destination service user build a believable user coalitions, pass through
The service usage history of other users to carry out service recommendation for destination service user in alliance, so as to complete follow-up service
Selection course, considers services selection from a kind of alliance's angle, both may insure the credibility of service user's information, and can also keep away
Exempt from service provider claim data can not implementations, be given below and set up definition basic during credible alliance and method：
1) credible alliance（Trus tworthy Community）What is formed in service selection system both can guarantee that user
Similarity degree in terms of usage behavior is serviced, also ensures that user's set of its behavior creditability, you can the use in letter alliance
Family is believable similar neighborhood user.
2) attributive character of service（Service Characteristics）The attributive character of service is the service that embodies itself
A series of attributes of feature, can also react satisfaction of the user to it, include the letter of service for personalized service
Reputation degree and frequency of use.The credit worthiness of service（reputat ion）Refer to that the use being collected into crosses the user of the service to it
Satisfaction, and frequency of use（frequency）Refer to that user in a past time window uses the number of times of the service.
3) trust of behavior is recommended（Trust Of Recommendation）In service selection system, recommend behavior
Trust refers to that using user as core presentee recommends nominator one kind of the honesty, reliability and validity of behavior
Subjective determination behavior.
4) direct trust (the Direct Trust Of Recommendat ion) recommended user and target for recommending behavior use
Family interacts a kind of trusting relationship obtained, i.e., the degree of recognition each other to mutual recommended project by direct information between the two.
5) the indirect trust (Indi rect Trust Of Recommendat ion) of behavior is recommended to refer to targeted customer
Transmission of the trust colony to recommended user's credit worthiness, i.e., by obtained from the transmission of the direct trusting relationship to recommending behavior
Accreditation relation is trusted indirectly.
6) degree of belief of behavior is recommended（Trust Degree Of Recommendation）The degree of belief of recommendation behavior is
It is present in a kind of measurement of point-to-point trusting relationship between presentee and nominator, by the direct degree of belief of recommendation behavior
With the indirect degree of belief two parts composition for recommending behavior.
Fig. 2 gives a frame construction drawing for building credible alliance, and this frame construction drawing is main by three part groups
Into：It is to add the consideration to Service Properties feature first, the calculation formula similarity user in traditional algorithm is improved,
Secondly the trusting relationship introduced between user, the genuine and believable degree of its information is considered by calculating users to trust degree, is finally integrated
Both similarity and degree of belief carry out user's distribution, determine the similar neighborhood user of targeted customer, build credible alliance.
A usual service recommendation system is by M service requester { U1,U2,…UmAnd N number of candidate service { S1,
S2,…SnComposition, its service evaluation information, i.e. service requester can be extracted and each by the user journal of service requester
Relation between individual service can be represented with M × N matrix, wherein the element R of the i-th row s rowi,sWhat is represented is service
Requestor i is to the service a certain QoS property values of s（Such as response time, success rate）Evaluation of estimate, generally scoring it is higher represent use
The quality that family is serviced this is more satisfied, if user i never called service s, then Ri,sIt is then 0.User i and j score jointly
The set of service crossed is sI, j, user i and j similarity Sim (i, j) can be calculated by formula 1：
WhereinWithRepresent user i and j to s respectivelyI, jIn all services scoring average.Parameter μ is service with ω
Two attributive character factor of influence, span is [0,1].Because the application environment difference residing for service can cause two
Corresponding change also occurs for the proportion that attributive character produces influence to result of calculation, and therefore, variable parameter μ can cause with ω
The similarity calculating method can preferably be applied to different application environments.
Next the credibility of user behavior is considered：A kind of channel of effective direct degree of belief for obtaining recommendation behavior is logical
Evaluate the prestige of other side each other to set up after crossing user-user information interaction.Two use are calculated using Beta trust models
Recommend the direct degree of belief RDT (i, j) of behavior between family.If user j and i were interacted with service s, Ri,sRepresent user i to clothes
Business s scoring, Rj,sScorings of the user j to service s is represented, i is recommended using j as mono- single neighbour of i, and by j pairs
Compared with true score value of the score value of service with i to service, if the Error Absolute Value between them is less than a certain fixed value
ε, then it is assumed that recommendations of the j to i be it is correct, it is on the contrary then think to recommend to be wrong.The set of service scored jointly at both
sI, jIn, user j is that the total degree that user i correctly recommends is designated as pi, and j, the total degree that mistake is recommended is designated as ni, j.Beta probability
Density function is applied to description and includes binary eventPosterior probability, here x events refer to that user j makes for i
Correct service recommendation, the formula for calculating the direct degree of belief of recommendation behavior is as follows：
Wherein Γ is gamma functions, 0<<x<<1.
When calculating the indirect trust of recommendation behavior, if set D is targeted customer i trust customer group, the letter in set D
Appoint family and trust evaluation is carried out to recommended user j, and combine respective trust weight as assessment foundation.RIDT (i, j) is represented
User i is to the indirect degree of belief of j recommendation behavior, and calculation formula is as follows：
In formula, wiTrust the user D in set D for targeted customeriTo i recommended users j trust weight, its value is two
User recommends the direct degree of belief of behavior, and gained is calculated by formula 2.RDT(j,Mi) it is user DiDirect letter between user j
Ren Du.
After the direct trust and indirect degree of belief that obtain recommendation behavior, user recommends the degree of belief of behavior can be by formula 4
Calculate：
RT(i,j)=αRDT(i,j)+βRIDT(i,j) （4）
In formula, RDT (i, j) represents the direct degree of belief of recommendation behavior, and RIDT (i, j) represents the indirect trust of recommendation behavior
Degree, α, β represent its weight respectively, and span is [0,1], and alpha+beta=1.Obviously, RT (i, j) value is bigger, illustrates user j to mesh
The recommendation for marking user i is more credible.
When the similarity and degree of belief of integration objective user and candidate neighbor, comprehensive weight calculation formula is as follows：
When carrying out user's distribution, because user is relatively more in actual network environment, between contact also relatively closely,
We take traditional top-k methods, and k user forms credible alliance before being chosen according to synthetic weights weight values.
On the basis of credible coalition formation, the work of relation is between Analysis Service user, is selected in system
The work for selecting module is the service interaction history with reference to user in credible alliance, including the discovery of each service, selection and poly-
Conjunction process, existing correlation, is carried according to this potential correlation for service user between the different services of analysis consideration
For can most meet the candidate service of its demand, the process of services selection is finally completed, this is that this patent needs the another of protection technique
On one side.Basic definition when services selection is carried out according to potentially relevant property between service and method is given below：
1) services set s is commonly usedc scWhat is represented is in certain period of time, to be gathered according to the service of all users in credible alliance
History is closed, the most frequently used sub-services set to carry out service aggregating of user selected is denoted as sc={sc1,sc2,
...scm, wherein each service can better meet the demand of user compared to other services.
2) when candidate service collection c user carries out services selection, the candidate service collection of user's request is met in Service Source pond,
It is expressed as c={ c1,c2,...cm}。
3) correlation R is in service discovery, polymerization and selection course between servicing, certain the specific connection having between service
System or pattern, including candidate service collection and conventional service centralized services and between the service in whole resource pool it is interrelated
The size of degree, this service that the bigger explanation candidate service of R value is concentrated more can close to user demand.
4) alliance's principle（Association Rules）It is a kind of most important technology of Data Mining, typically uses
To find the contact in large-scale database between entity.A → B can be expressed as with formula, wherein And A ∩
B=Ф, the concrete meaning of formula is：A and B belongs to a set I, and we have very big probability while article in finding A
The article in B can be found.There are two critically important basic conception support and confidence in alliance's principle.Alliance
Principle A → B support represents the frequency that alliance principle occurs, what alliance principle A → B confidence was represented be
Also the probability of entity B can be found while finding entity A, it embodies the contiguity between A and B.
Fig. 3, which gives between service to exist in the schematic diagram of correlation, credible alliance, two service user U1And U2, he
To a certain service request S atomic service polymerization process all, but when they are to S2And S5Carry out specific sub-services mapping
When, user U1Selection is sub-services S21And S51, and user U2Selection is sub-services S22And S52, just because of U1And U2Belong to
One user coalitions, they are very high in the similarity degree of service use, it can be considered that sub-services S21,S51,
S22,S52Between there is certain specific contact or pattern, that is to say the correlation between potential service, with reference to this correlation,
Services selection can be realized for destination service user.
Fig. 4 is given under dynamic, heterogeneous network environment, by the service aggregating history of user in comprehensive credible alliance, point
Whole service discovery, selection and polymerization process are analysed, so as to extract the schematic diagram of conventional services set, specific method is expressed as follows：
It regard user-service call rate as formation scReference factor.As a certain user k selections service j in credible alliance
When, we record this and called, and total degree called service j is designated as number (sj,uk), we are by public affairs after a period of time
What formula 6 calculated service calls rate, is designated as freq (sj,uk), then user in credible alliance-service call rate is carried out
Sequence, selects m sub-services of numerical value highest as the element in conventional services set.
A certain service c in ciWith scBetween correlation can be counted by assessing the alliance's principle that exist between the two
Calculate.Alliance's principle of presence is more, ciWith scCorrelation it is then stronger；We are by c in additioniWith all services sets in resource pool
The correlation of conjunction also serves as another important reference factor, ciThe alliance's principle existed between s is more, ciWith scCorrelation
Property is then weaker.R is used herein1And R2Both correlations are represented respectively, correlation between service is defined as R=R1*R2, contact money
The polymerization history of all services in the pond of source calculates each in candidate service collection c and services and commonly use services set scMiddle service
Support and confidence values.Exclusion is unsatisfactory for thresholding confidence (sc→c)>δ, sup port (sc→c)>γ time
Choosing service, wherein δ and γ value change according to specific model and environment.The root in all candidate services for meeting threshold value
C is calculated according to formula 7iWith scThe correlation R of middle service1
C is calculated further according to formula 8iWith the correlation R of all services2, wherein t is all present in Service Source pond
Alliance principle AR number, num (ci, AR) and what is represented is to include ciThe alliance principle AR of participation number.
Then the correlation R between service is calculated, return value highest candidate service, user is selected as optimal service.
So far just successfully be have selected using alliance's principle in candidate service collection c can most meet the service of user's request.
We realize by the potentially relevant property between Analysis Service to take for target on the basis of credible alliance builds
The process of the user that make sures selection service.Simultaneously for the atomic service selected, in the UDDI registrations in service selection system
Update it in the heart to release news, when other service users ask to service, it is possible to reduce the time overhead of selection course.
In order to illustrate the services selection scheme of the present invention based on correlation, a specific example is we illustrated
Son --- user goes the optimal tourist service that Nanjing is played to elaborate this method to network request.
The first step, the related service provider that travels is issued on Nanjing Tourism by trusted third party on network
Atomic service resource, for example：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, user proposes Nanjing Tourism service resource request to service access platform, and lookup meets requirement
Candidate service.The content of lookup includes the species of service（It can be categorized as in present case：It is commercial, high, medium, common）, clothes
The security of business（Such as：Trip mode security, food safety grade etc.）, reliability（Such as：The prestige of travel agency, the vehicles
Punctuality rate etc.）And price（Common price and favourable price etc.）, preliminary screening is carried out to service by some constraints,
Form the candidate service resource pool for being available for user to carry out selection.
3rd step, user undergoes according to conventional tourism, search out a part and oneself preference it is same or analogous other
User, such as at travel agency, sight spot, mode of transportation and stay in terms of selection all with oneself identical that a group user, he
Be similar neighborhood user of the targeted customer in terms of tourist image design, if while such user once carried out to targeted customer it is straight
Connect or indirectly tourism is recommended, consider further that the trusting relationship existed between them, with reference to the drawn similarity relation of analysis before,
It is determined that not only to user in terms of tourist image design it is similar but also can by its trust user group, it is determined as the credible alliance of user.
4th step, if the user in credible alliance once arrived Nanjing Tourism mistake, their each tourist attractions to Nanjing, each bars
Tourism route, various trip modes, each stay and food and drink etc. had an experience that uses, and their quality to these services
Also there is more comprehensive evaluation with reliability, according to their feedback, we extract most users satisfaction the most in alliance
Some services, form the conventional services set on tour schedule, consider to take when carrying out tourist service selection as targeted customer
One of business correlation refers to group.
5th step, on the basis of previous step obtains conventional services set, we by the service in candidate service resource pool with
Service in conventional services set is associated reference, was played all travel informations of user with reference to Nanjing was once arrived, and found these
Service in the contact of some fixations existed between service, such as two services sets whether all once jointly by certain some user simultaneously
Once calling during the journey etc, and finding contact and pattern between these services using alliance's principle, exist between the service of that is to say
Correlation, it is similar can find out candidate service collection and in whole Service Source pond service between correlation.
6th step, by analyzing and calculating the correlation between service, may finally select can most meet targeted customer's demand
That group candidate service, user carries out Nanjing Tourism plan with reference to the service of this group, and its request is addressed.The service of user
Selection result be also updated to other users may be referred to use for reference information, be arrive from now on Nanjing Tourism user it is faster, more excellent
All kinds of services selections of carry out of matter provide guaranteed reliability, also improve the efficiency of selection of tourist service in whole network.
Claims (6)
1. a kind of service selection system based on correlation, it is characterised in that：Including release module, retrieval module, evaluation module,
Database module and selecting module, retrieval module provide retrieval port and accessed for user, comprising all related in database
Atomic service resource and the evaluating data to these atomic service resources, evaluation module provide evaluation mechanism so that user is to having connect
By or the service that has used evaluated, evaluating data feeding database is preserved, and selecting module is according to the pass of user search
Key word provides that close Service Source therewith is for reference, and release module provides interface for service provider's issuing service letter
Breath, when information receives to finish, information on services is stored in database by release module；The selecting module, the work of selecting module
Work is the service interaction history with reference to user in credible alliance, includes discovery, selection and the polymerization process of each service, analysis
Consider existing correlation between different services, it can most be met by being provided according to this potential correlation for service user
The candidate service of demand, is finally completed services selection；The credible alliance builds in the following manner：The trust introduced between user
Relation, the genuine and believable degree of information is considered by calculating users to trust degree, and both last comprehensive similarity and degree of belief are carried out
User distributes, and determines the similar neighborhood user of targeted customer；
A usual service recommendation system is by M user { U1,U2,…UMAnd N number of candidate service { S1,S2,…SNComposition,
Can extract the relation between its service evaluation information, i.e. user and each service by the user journal of user can use one
Individual M × N matrix is represented, wherein the element R of i-th row s rowi,sWhat is represented is that user i is commented the service a certain QoS property values of s
Value, generally scoring be higher, and to represent the quality that user serviced this more satisfied, if user i never called service s, then
Ri,sIt is then 0；The set of service that user i and j scored jointly is si,j, user i and j similarity Sim (i, j) can be by formula 1
Calculated：
<mrow>
<mi>S</mi>
<mi>i</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>s</mi>
<mo>&Element;</mo>
<msub>
<mi>s</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
</mrow>
</munder>
<msqrt>
<mrow>
<mi>&mu;</mi>
<mfrac>
<mn>1</mn>
<mrow>
<msubsup>
<mi>reputation</mi>
<mi>s</mi>
<mn>2</mn>
</msubsup>
</mrow>
</mfrac>
<mo>+</mo>
<mi>&omega;</mi>
<mfrac>
<mn>1</mn>
<mrow>
<msubsup>
<mi>frequency</mi>
<mi>s</mi>
<mn>2</mn>
</msubsup>
</mrow>
</mfrac>
</mrow>
</msqrt>
<mrow>
<mo>(</mo>
<msub>
<mi>R</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>s</mi>
</mrow>
</msub>
<mo>-</mo>
<mover>
<msub>
<mi>R</mi>
<mi>i</mi>
</msub>
<mo>&OverBar;</mo>
</mover>
<mo>)</mo>
</mrow>
<mo>&times;</mo>
<mrow>
<mo>(</mo>
<msub>
<mi>R</mi>
<mrow>
<mi>j</mi>
<mo>,</mo>
<mi>s</mi>
</mrow>
</msub>
<mo>-</mo>
<mover>
<msub>
<mi>R</mi>
<mi>j</mi>
</msub>
<mo>&OverBar;</mo>
</mover>
<mo>)</mo>
</mrow>
</mrow>
<msqrt>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>s</mi>
<mo>&Element;</mo>
<msub>
<mi>s</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
</mrow>
</munder>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>R</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>s</mi>
</mrow>
</msub>
<mo>-</mo>
<mover>
<msub>
<mi>R</mi>
<mi>i</mi>
</msub>
<mo>&OverBar;</mo>
</mover>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<msqrt>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>s</mi>
<mo>&Element;</mo>
<msub>
<mi>s</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
</mrow>
</munder>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>R</mi>
<mrow>
<mi>j</mi>
<mo>,</mo>
<mi>s</mi>
</mrow>
</msub>
<mo>-</mo>
<mover>
<msub>
<mi>R</mi>
<mi>j</mi>
</msub>
<mo>&OverBar;</mo>
</mover>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
</mrow>
</msqrt>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
WhereinWithRepresent user i and j to s respectivelyi,jIn all services scoring average；Parameter μ is the two of service with ω
The factor of influence of individual attributive character, span is [0,1]；Because the application environment difference residing for service can cause two attributes
Corresponding change also occurs for the proportion that feature produces influence to result of calculation, therefore, and variable parameter μ causes the similarity with ω
Computational methods can preferably be applied to different application environments；
Next the credibility of user behavior is considered：It is a kind of it is effective obtain recommendation behavior direct degree of belief channel be by using
The prestige of other side is evaluated after the information interaction of family each other to set up；Calculated using Beta trust models two users it
Between recommend behavior direct degree of belief RDT (i, j)；If user j and i were interacted with service s, Ri,sRepresent user i to service s
Scoring, Rj,sScorings of the user j to service s is represented, j as i a single neighbours are recommended i, and by j pairs
Compared with true score value of the score value of service with i to service, if the Error Absolute Value between them is less than a certain fixed value
ε, then it is assumed that recommendations of the j to i be it is correct, it is on the contrary then think to recommend to be wrong；The set of service scored jointly at both
si,jIn, user j is that the total degree that user i correctly recommends is designated as pi, and j, the total degree that mistake is recommended is designated as ni, j；Beta probability
Density function is applied to description and includes binary eventPosterior probability, here x events refer to that user j makes for i
Correct service recommendation, the formula for calculating the direct degree of belief of recommendation behavior is as follows：
<mrow>
<mi>R</mi>
<mi>D</mi>
<mi>T</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<mi>&Gamma;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>p</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
<mo>+</mo>
<msub>
<mi>n</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
<mo>+</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>&Gamma;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>p</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mi>&Gamma;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>n</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<msup>
<mi>x</mi>
<msub>
<mi>p</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
</msup>
<msup>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>n</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein Γ is gamma functions, 0<x<1；
When calculating the indirect trust of recommendation behavior, if set D is targeted customer i trust customer group, the trust in set D is used
Family carries out trust evaluation to recommended user j, and combines respective trust weight as assessment foundation；RIDT (i, j) represents user i
To the indirect degree of belief of j recommendation behavior, calculation formula is as follows：
<mrow>
<mi>R</mi>
<mi>I</mi>
<mi>D</mi>
<mi>T</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mi>D</mi>
</munder>
<msub>
<mi>w</mi>
<mi>i</mi>
</msub>
<mi>R</mi>
<mi>D</mi>
<mi>T</mi>
<mrow>
<mo>(</mo>
<mi>j</mi>
<mo>,</mo>
<msub>
<mi>D</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<munder>
<mo>&Sigma;</mo>
<mi>D</mi>
</munder>
<msub>
<mi>w</mi>
<mi>i</mi>
</msub>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
In formula, wiTrust the user D in set D for targeted customeriTo user i recommended users j trust weight, its value is two
User recommends the direct degree of belief of behavior, and gained is calculated by formula 2；RDT(j,Di) it is user DiDirect letter between user j
Ren Du；
After the direct degree of belief and indirect degree of belief that obtain recommendation behavior, user recommends the degree of belief of behavior to be counted by formula 4
Draw：
RT (i, j)=α RDT (i, j)+β RIDT (i, j) (4)
In formula, RDT (i, j) represents the direct degree of belief of recommendation behavior, and RIDT (i, j) represents the indirect degree of belief of recommendation behavior,
α, β represent its weight respectively, and span is [0,1], and alpha+beta=1；RT (i, j) value is bigger, illustrates user j to targeted customer i
Recommendation it is more credible；
When the similarity and degree of belief of integration objective user and candidate neighbor, comprehensive weight calculation formula is as follows：
<mrow>
<mi>w</mi>
<mi>e</mi>
<mi>i</mi>
<mi>g</mi>
<mi>h</mi>
<mi>t</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<mn>2</mn>
<mi>s</mi>
<mi>i</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>&CenterDot;</mo>
<mi>R</mi>
<mi>T</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>s</mi>
<mi>i</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>R</mi>
<mi>T</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>5</mn>
<mo>)</mo>
</mrow>
</mrow>
When carrying out user's distribution, because user is relatively more in actual network environment, between contact also relatively closely, we
Traditional top-k methods are taken, k user forms credible alliance before being chosen according to synthetic weights weight values；
The work of selecting module is the service interaction history with reference to user in credible alliance, includes discovery, the choosing of each service
Select and polymerization process, existing correlation, makes according to this potential correlation for service between the different services of analysis consideration
User, which provides, can most meet the candidate service of its demand, be finally completed services selection, specifically chosen method is expressed as follows：
It regard user-service call rate as formation scReference factor；As a certain user k selections service j in credible alliance, I
Record this and call, total degree called service j is designated as number (sj,uk), we are based on formula 6 after a period of time
Calculate service calls rate, is designated as freq (sj,uk), then user in credible alliance-service call rate is ranked up,
M sub-services of numerical value highest are selected as the element in conventional services set；
<mrow>
<mi>f</mi>
<mi>r</mi>
<mi>e</mi>
<mi>q</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>s</mi>
<mi>j</mi>
</msub>
<mo>,</mo>
<msub>
<mi>u</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mfrac>
<mrow>
<mi>n</mi>
<mi>u</mi>
<mi>m</mi>
<mi>b</mi>
<mi>e</mi>
<mi>r</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>s</mi>
<mi>j</mi>
</msub>
<mo>,</mo>
<msub>
<mi>u</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>m</mi>
</munderover>
<mi>n</mi>
<mi>u</mi>
<mi>m</mi>
<mi>b</mi>
<mi>e</mi>
<mi>r</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>s</mi>
<mi>j</mi>
</msub>
<mo>,</mo>
<msub>
<mi>u</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
<mi> </mi>
<mi>n</mi>
<mi>u</mi>
<mi>m</mi>
<mi>b</mi>
<mi>e</mi>
<mi>r</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>s</mi>
<mi>j</mi>
</msub>
<mo>,</mo>
<msub>
<mi>u</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>&NotEqual;</mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
<mi> </mi>
<mi>n</mi>
<mi>u</mi>
<mi>m</mi>
<mi>b</mi>
<mi>e</mi>
<mi>r</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>s</mi>
<mi>j</mi>
</msub>
<mo>,</mo>
<msub>
<mi>u</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>6</mn>
<mo>)</mo>
</mrow>
</mrow>
Conventional services set scWhat is represented is in certain period of time, according to the service aggregating history of all users in credible alliance, choosing
The most frequently used sub-services set to carry out service aggregating of user gone out, is denoted as sc={ sc1,sc2,...scm, wherein often
Individual service can better meet the demand of user compared to other services；It is full in Service Source pond when user carries out services selection
The candidate service collection of sufficient user's request, is expressed as c={ c1,c2,...cm}；
A certain service c in ciWith scBetween correlation can be calculated by assessing the alliance's principle that exist between the two；Deposit
Alliance's principle it is more, ciWith scCorrelation it is then stronger；We are by c in additioniWith the phase of all set of services in resource pool
Closing property also serves as another important reference factor, ciThe alliance's principle existed between all set of services is more, ciWith sc's
Correlation is then weaker；R is used herein1And R2Both correlations are represented respectively, correlation between service is defined as R=R1*R2,
The polymerization history of all services calculates each in candidate service collection c and services and commonly use services set s in contact resource poolcMiddle clothes
Support the and confidence values of business；Exclusion is unsatisfactory for thresholding confidence (sc→ c) ＞ δ, support (sc→ c) ＞
The value of γ candidate service, wherein δ and γ changes according to specific model and environment；In all candidate's clothes for meeting threshold value
C is calculated according to formula 7 in businessiWith scThe correlation R of middle service1
<mrow>
<msub>
<mi>R</mi>
<mn>1</mn>
</msub>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
<mo>,</mo>
<msub>
<mi>s</mi>
<mrow>
<mi>c</mi>
<mi>j</mi>
</mrow>
</msub>
<mo>&Element;</mo>
<msub>
<mi>S</mi>
<mi>c</mi>
</msub>
</mrow>
<mi>m</mi>
</munderover>
<mi>c</mi>
<mi>o</mi>
<mi>n</mi>
<mi>f</mi>
<mi>i</mi>
<mi>d</mi>
<mi>e</mi>
<mi>n</mi>
<mi>c</mi>
<mi>e</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>s</mi>
<mrow>
<mi>c</mi>
<mi>j</mi>
</mrow>
</msub>
<mo>&RightArrow;</mo>
<msub>
<mi>c</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>7</mn>
<mo>)</mo>
</mrow>
</mrow>
C is calculated further according to formula 8iWith the correlation R of all services2, wherein t is that all alliances present in Service Source pond are former
Then AR number, num (ci, AR) and what is represented is to include ciThe alliance principle AR of participation number；
<mrow>
<msub>
<mi>R</mi>
<mn>2</mn>
</msub>
<mo>=</mo>
<mi>l</mi>
<mi>o</mi>
<mi>g</mi>
<mfrac>
<mrow>
<mi>t</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>n</mi>
<mi>u</mi>
<mi>m</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>c</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<mi>A</mi>
<mi>R</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mn>0.5</mn>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>8</mn>
<mo>)</mo>
</mrow>
</mrow>
Then the correlation R between service is calculated, return value highest candidate service, user is selected as optimal service；
So far just successfully be have selected using alliance's principle in candidate service collection c can most meet the service of user's request.
2. selection system according to claim 1, it is characterised in that：The information on services carries out registration hair using UDDI
Cloth.
3. selection system according to claim 1, it is characterised in that：The evaluating data that database root is provided according to evaluation module
Atomic service resource is ranked up, selecting module prioritizing selection is evaluated high atomic service resource and shown to user.
4. a kind of system of selection of the service selection system based on correlation according to claim 1, it is characterised in that：Including
Following steps：
1) user by retrieve module provide retrieval port to system submit the keyword to be inquired about；
2) in keyword feeding database, the Service Source data of matching are selected from database；
3) according to keyword, selecting module is selected and the immediate Service Source of keyword；
4) module is retrieved by the resource of matching by retrieving module feedback to user.
5. system of selection according to claim 4, it is characterised in that：Also include evaluation procedure, user receives after service, can
The evaluation mechanism inside the evaluation information to the service, evaluation module is write by evaluation module to combine with writing comment for marking
System.
6. system of selection according to claim 4, it is characterised in that：What the resource and selecting module that database is returned were recommended
Also include evaluation other users to the resource in Service Source.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210494558.0A CN103034963B (en) | 2012-11-28 | 2012-11-28 | A kind of service selection system and system of selection based on correlation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210494558.0A CN103034963B (en) | 2012-11-28 | 2012-11-28 | A kind of service selection system and system of selection based on correlation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103034963A CN103034963A (en) | 2013-04-10 |
| CN103034963B true CN103034963B (en) | 2017-10-27 |
Family
ID=48021832
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210494558.0A Active CN103034963B (en) | 2012-11-28 | 2012-11-28 | A kind of service selection system and system of selection based on correlation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103034963B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104111959B (en) * | 2013-04-22 | 2017-06-20 | 浙江大学 | Service recommendation method based on social networks |
| CN103412918B (en) * | 2013-08-08 | 2016-07-06 | 南京邮电大学 | A kind of service trust degree appraisal procedure based on service quality and reputation |
| CN105225175A (en) * | 2015-10-21 | 2016-01-06 | 刘锐光 | C2C, B2C, B2B social platform internet traveling method and system |
| CN105303497A (en) * | 2015-12-01 | 2016-02-03 | 浪潮电子信息产业股份有限公司 | Nursing worker query system and method based on cloud computing |
| CN106209978B (en) * | 2016-06-22 | 2019-04-09 | 安徽大学 | A kind of alliance Services Composition selection system and selection method |
| CN108629345B (en) * | 2017-03-17 | 2021-07-30 | 北京京东尚科信息技术有限公司 | High-dimensional image feature matching method and device |
| CN106961356B (en) * | 2017-04-26 | 2020-01-10 | 中国人民解放军信息工程大学 | Web service selection method and device based on dynamic QoS and subjective and objective weight |
| CN110196796B (en) * | 2019-05-15 | 2023-04-28 | 无线生活（杭州）信息科技有限公司 | Effect evaluation method and device for recommendation algorithm |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1679037A (en) * | 2002-08-28 | 2005-10-05 | 国际商业机器公司 | Network system, provider, management site, requestor, and program |
| CN101895547A (en) * | 2010-07-16 | 2010-11-24 | 浙江大学 | Uncertain service-based recommender system and method |
| CN101909055A (en) * | 2010-07-19 | 2010-12-08 | 东南大学 | Multithread Web service negotiation method based on QoS |
| CN102523247A (en) * | 2011-11-24 | 2012-06-27 | 合肥工业大学 | Cloud service recommendation method and device based on multi-attribute matching |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004288118A (en) * | 2003-03-25 | 2004-10-14 | Hewlett Packard Co <Hp> | Method of managing service registration data |
-
2012
- 2012-11-28 CN CN201210494558.0A patent/CN103034963B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1679037A (en) * | 2002-08-28 | 2005-10-05 | 国际商业机器公司 | Network system, provider, management site, requestor, and program |
| CN101895547A (en) * | 2010-07-16 | 2010-11-24 | 浙江大学 | Uncertain service-based recommender system and method |
| CN101909055A (en) * | 2010-07-19 | 2010-12-08 | 东南大学 | Multithread Web service negotiation method based on QoS |
| CN102523247A (en) * | 2011-11-24 | 2012-06-27 | 合肥工业大学 | Cloud service recommendation method and device based on multi-attribute matching |
Non-Patent Citations (1)
| Title |
|---|
| 基于偏好推荐的可信服务选择;朱锐等;《软件学报》;20110531;第22卷(第5期);第852-864页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103034963A (en) | 2013-04-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103034963B (en) | A kind of service selection system and system of selection based on correlation | |
| Bindu et al. | From conventional governance to e-democracy: Tracing the evolution of e-governance research trends using network analysis tools | |
| Al-Gasawneh et al. | The relationship between perceived destination image, social media interaction and travel intentions relating to Neom city | |
| Yang et al. | Friend or frenemy? Predicting signed ties in social networks | |
| CN103745105B (en) | Method and system for predicting user property in social network | |
| CN107220365A (en) | Accurate commending system and method based on collaborative filtering and correlation rule parallel processing | |
| US20140143332A1 (en) | Discovering signature of electronic social networks | |
| CN102567900A (en) | Method for recommending commodities to customers | |
| CN106447463A (en) | Commodity recommendation method based on Markov decision-making process model | |
| CN109241341A (en) | A kind of video display of knowledge based map preferentially recommender system and method | |
| CN106951471A (en) | A kind of construction method of the label prediction of the development trend model based on SVM | |
| KR20090000814A (en) | Method and system for creating advertisement | |
| US20110184815A1 (en) | System and method for sharing profits with one or more content providers | |
| CN105761154A (en) | Socialized recommendation method and device | |
| CN107194721A (en) | Service recommendation person based on reputation record analysis has found method | |
| CN114817946A (en) | Credible execution environment-based federated learning gradient boosting decision tree training method | |
| Vidgen et al. | Invited viewpoint: how well does the information systems discipline fare in the Financial Times’ top 50 journal list? | |
| CN106021428A (en) | KNN and three-way decision-based movie recommendation method | |
| CN107767280A (en) | A kind of high-quality node detecting method based on element of time | |
| Liu et al. | Heterogeneous graph neural networks for large-scale bid keyword matching | |
| Chikweche et al. | Targeting the floating middle of pyramid consumers in Africa: Revisiting marketing mix considerations | |
| CN109544271A (en) | A kind of trade managing system | |
| CN106446191A (en) | Logistic regression based multi-feature network popular tag prediction method | |
| KR20200130767A (en) | Method and device for evaluating whether cryptocurrency is listed on cryptocurrency market using artificial neural network | |
| CN109559169B (en) | Method for identifying sharp users based on online user scoring |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |