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

Abstract

The invention relates to a correlation-based service selection system, which includes a release module, a search module, an evaluation module, a database module and a selection module, and the service information is registered and published by using UDDI. The database sorts the atomic service resources according to the evaluation data provided by the evaluation module, and the selection module preferentially selects the atomic service resources with high evaluation to display to the user. The invention also relates to the selection method of the selection system. The present invention changes the traditional service selection mode, breaks through the limitation of the existing technology that the credibility is not high when looking for similar neighbor users, considers the service selection process through the correlation between comprehensive services, and avoids possible joint fraud , can make the selection result more accurate; support the dynamic characteristics of the open network environment, adapt to the needs of the open network environment; effectively improve the efficiency of service selection, and can also avoid the occurrence of untrustworthy data provided by service providers.

Description

A correlation-based service selection system and selection method

technical field

The present invention relates to a query system for service products, specifically, in an open, heterogeneous and complex network environment, a system for building trusted user alliances and selecting services according to the correlation between services. At the same time, the present invention discloses The system selection method.

Background technique

Open network applications and the concept of "software as a service" will inevitably lead to tremendous changes in the main form, operation mode, production mode and use mode of software systems based on the Internet environment. There are a large number of services with the same function and different non-functional characteristics distributed on the network. When users request services, there is often no single service that can directly meet the needs of users. At this time, some atomic services need to be selected for aggregation. The aggregation process of services can be represented by atomic services. Each atomic service is just a virtual service module, which embodies the logical relationship of services, and will be replaced by actual specific atomic services during the aggregation process. The service selection process depends on the aggregation history. Most existing correlation selection schemes separate the service selection process from the aggregation process, ignoring the correlation between services. In short, the existing research either lacks theoretical depth, or does not give a practical model or algorithm, so it is not suitable for large-scale open network environment.

Contents of the invention

Purpose of the invention: The purpose of the invention is to provide a system for providing high-efficiency, accurate, effective and reliable service selection in an open and heterogeneous network environment, and the invention discloses a selection method of the system at the same time.

Technical solution: the present invention is realized through the following technical solutions: a service selection system based on correlation, including a publishing module, a retrieval module, an evaluation module, a database module and a selection module, the retrieval module provides a retrieval port for users to access, and the database contains All relevant atomic service resources and the evaluation data of these atomic service resources, the evaluation module provides an evaluation mechanism for users to evaluate the services they have accepted or used, and the evaluation data is sent to the database for storage, and the selection module is based on the keywords retrieved by users Provide similar service resources for users' reference. The publishing module provides an interface for service providers to publish service information. When the information is received, the publishing module stores the service information in the database.

The service information uses UDDI to register and release.

The database sorts the atomic service resources according to the evaluation data provided by the evaluation module, and the selection module preferentially selects the atomic service resources with high evaluation to display to the user.

A selection method for a selection system, comprising the steps of:

1) The user submits the keywords to be queried to the selection system through the search port provided by the search module;

2) The keyword is sent to the database, and the matching service resource data is selected from the database;

3) According to the keyword, the selection module selects the service resource closest to the keyword;

4) The retrieval module feeds back the matched resources to the user through the retrieval module.

It also includes an evaluation step. After the user accepts the service, the user can write evaluation information for the service through the evaluation module. The evaluation mechanism is a combination system of scoring and writing reviews.

The resources returned by the database and the service resources recommended by the selection module also include other users' comments on the resources.

Beneficial effects: Compared with the prior art, the present invention changes the traditional service selection mode, breaks through the limitation of the prior art that the credibility is not high when looking for similar neighbor users, and considers the service selection through the correlation between comprehensive services The process avoids possible joint deception and makes the selection result more accurate; supports the dynamic characteristics of the open network environment to ensure the support of the dynamic characteristics of the network and adapts to the needs of the open network environment; effectively improves the efficiency of service selection, and also It can avoid the occurrence of untrustworthy data provided by the service provider; the scheme has stronger applicability.

Description of drawings

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

Fig. 2 is a frame structure diagram of the trusted alliance construction of the present invention.

Fig. 3 is a schematic diagram of inter-service correlation in the present invention.

Fig. 4 is a schematic diagram of extracting common service sets in the present invention.

Detailed ways

Below in conjunction with accompanying drawing, the present invention is further described in detail:

The present invention relates to a service selection system based on correlation. Fig. 1 provides a structural diagram of the service selection system, including a publishing module, a retrieval module, an evaluation module, a database module and a selection module, wherein the publishing module is a service provider through a service Describe the language, describe the function and interface information of the provided service, and use UDDI (Universal Description Discovery and Integration) to register and publish the service to the service registration center; the retrieval module is what the service user searches and finds in the service registration center. service and obtain the description file of the corresponding service; the evaluation module is for service users to evaluate the service quality of previously used providers, and store the evaluation information in the database module, and combine these two modules to form a credible service use The selection module is based on the service interaction history of other users, and on the premise of extracting the common service set, the service user selects the atomic service that best matches its needs, and the correlation between the reference services Carry out service selection for protection, and the dotted box in the figure shows the system part of the independent and traditional service selection scheme.

The traditional service selection method is a simple request-response process between the service user and the service provider. The binding between the two is often based on the quality of service (QOS). However, with the increasing number of services in the resource pool, Searching and selecting services based on QOS is difficult to guarantee the authenticity of the data provided by the service provider, and the credibility of the QOS feedback data of service users cannot be guaranteed, so the traditional method is insufficient in the accuracy of service selection .

According to the information in the evaluation module and database module in the proposed service selection system, we can break through the traditional method, center on the service user, and find other users with the same service experience for the target service user. Build a credible user alliance for the target service users according to the trust relationship between them, and make service recommendations for the target service users through the service usage history of other users in the alliance, so as to complete the subsequent service selection process, from a Considering service selection from the perspective of the alliance can not only ensure the credibility of service user information, but also avoid the situation where the service provider claims that the data is unreliable. The basic definitions and methods in the process of establishing a trusted alliance are given below:

1) Trustworthy Community (Trustworthy Community) is a set of users formed in the service selection system that can not only ensure the similarity of users in terms of service usage behavior, but also ensure the credibility of their behavior, that is, the users in the trusted community are Trusted similar neighbor users.

2) Service Characteristics Service characteristics are a series of attributes that reflect the characteristics of the service itself, which can be used to personalize the service, and can also reflect the user's satisfaction with it, including the reputation of the service and the frequency of use. Service reputation (reputation) refers to the satisfaction degree of users who have used the service collected, and frequency of use (frequency) refers to the number of times users use the service in a past time window.

3) Trust of Recommendation (Trust Of Recommendation) In the service selection system, trust of recommendation refers to a subjective judgment of the recommender on the honesty, reliability and effectiveness of the recommender's recommendation, with the user at the core. Behavior.

4) Direct trust of recommendation behavior (Direct Trust Of Recommendation) is a trust relationship obtained through direct information interaction between the recommended user and the target user, that is, the degree of mutual recognition of each other's recommended items.

5) Indirect Trust Of Recommendation (Indirect Trust Of Recommendation) refers to the transmission of the credibility of the recommended user by the trust group of the target user, that is, the indirect trust and recognition relationship obtained through the transmission of the direct trust relationship of the recommendation behavior.

6) Trust Degree of Recommended Behavior (Trust Degree Of Recommendation) Trust Degree of Recommended Behavior is a measure of a point-to-point trust relationship between the recommender and the recommender. The indirect trust degree consists of two parts.

Figure 2 shows a frame structure diagram for building a trusted alliance. This frame structure diagram is mainly composed of three parts: first, the consideration of service attribute characteristics is added, and the calculation formula of the similarity between users in the traditional algorithm is improved. Secondly, the trust relationship between users is introduced, and the real credibility of the information is considered by calculating the user's trust degree. Finally, the user is assigned based on the combination of similarity and trust, and the similar neighbor users of the target user are determined to build a trusted alliance.

Usually a service recommendation system is composed of M service requesters {U ₁ , U ₂ ,…U _m } and N candidate services {S ₁ ,S ₂ ,…S _n }, and the user logs of service requesters can extract The service evaluation information, that is, the relationship between the service requester and each service can be expressed by an M×N matrix, where the element R _i,s in the i-th row and s column represents the service requester i’s relationship with the service s The evaluation value of a certain QoS attribute value (such as response time, success rate, etc.), usually the higher the score, the more satisfied the user is with the quality of the service. If the user i has never called the service s, then R _i,s is 0 . The set of services rated by users i and j together is s _{i, j} , and the similarity Sim(i, j) between users i and j can be calculated according to formula 1:

$\mathrm{<span\; class="notranslate">\; Sim</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\sqrt{\mathrm{<span\; class="notranslate">\; \&mu;</span>}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; reputation</span>}}_{\mathrm{<span\; class="notranslate">\; the\; <figure-callout\; id="2"\; label="s"\; filenames="HDA00002485691300012.png"\; state="\{\{state\}\}">s</figure-callout></span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&omega;</span>}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; frequency</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}<span\; class="notranslate">\; )</span>}{\sqrt{\underset{\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\sqrt{\underset{\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

和

分别表示用户i和j对s_i，j中所有服务的评分的均值。参数μ与ω是服务的两个属性特征的影响因子，取值范围是[0,1]。由于服务所处的应用环境不同会导致两个属性特征对计算结果产生影响的比重也发生相应的变化，因此，可变的参数μ与ω可以使得该相似度计算方法能更好地适用于不同的应用环境。in

and

denote the average ratings of all services in si _,j by users i and j, respectively. The parameters μ and ω are the influencing factors of the two attribute characteristics of the service, and the value range is [0,1]. Since the application environment of the service is different, the proportion of the influence of the two attribute features on the calculation result will also change accordingly. Therefore, the variable parameters μ and ω can make the similarity calculation method more suitable for different application environment.

Next, consider the credibility of user behavior: a channel to effectively obtain direct trust in recommended behavior is established by evaluating each other's reputation after information interaction between users. The Beta trust model is used to calculate the direct trust degree RDT(i,j) between two users in the recommended behavior. If both users j and i have interacted with service s, R _i,s represents the rating of user i on service s, R _j,s represents the rating of user j on service s, and j is recommended as a single neighbor of i , and compare the rating value of j pair of services with the real rating value of i pair of services, if the absolute value of the error between them is less than a certain fixed value ε, it is considered that j’s recommendation for i is correct, otherwise, the recommendation it is wrong. In the service set si _,j that both have scored together, the total number of correct recommendations for user i by user j is recorded as pi,j, and the total number of incorrect recommendations is recorded as ni,j. The Beta probability density function is suitable for describing binary events The posterior probability of , where the x event refers to the user j making a correct service recommendation for i, the formula for calculating the direct trust degree of the recommended behavior is as follows:

$\mathrm{<span\; class="notranslate">\; RDT</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; \&Gamma;</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; \&Gamma;</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; \&Gamma;</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; x</span>}}^{{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}^{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

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

When calculating the indirect trust of the recommended behavior, set D as the trusted user group of the target user i, and the trusted users in the set D evaluate the trust of the recommended user j, and combine their respective trust weights as the evaluation basis. RIDT(i,j) represents the indirect trust degree of user i to j's recommendation behavior, and the calculation formula is as follows:

$\mathrm{<span\; class="notranslate">\; RIDT</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; D.</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; w</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}\mathrm{<span\; class="notranslate">\; RDT</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>}{\underset{\mathrm{<span\; class="notranslate">\; D.</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; w</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

In the formula, w _i is the trust weight of user D _i in the target user trust set D recommending user j to i, and its value is the direct trust degree of the two user recommendation behaviors, which is calculated by formula 2. RDT(j,M _i ) is the direct trust degree between user D _i and user j.

After obtaining the direct trust and indirect trust degree of the recommended behavior, the trust degree of the user's recommended behavior can be calculated by formula 4:

RT(i,j)=αRDT(i,j)+βRIDT(i,j) （4）

In the formula, RDT(i,j) represents the direct trust degree of the recommended behavior, RIDT(i,j) represents the indirect trust degree of the recommended behavior, α and β represent their weights respectively, and the value range is [0,1], and α+β=1. Obviously, the larger the value of RT(i, j), the more trustworthy the user j's recommendation to the target user i is.

When synthesizing the similarity and trust between the target user and the candidate neighbors, the formula for calculating the comprehensive weight is as follows:

$\mathrm{<span\; class="notranslate">\; weight</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; sim</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&CenterDot;</span>\mathrm{<span\; class="notranslate">\; RT</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; sim</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; RT</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$

When performing user allocation, since there are many users in the actual network environment and the connections between them are relatively close, we adopt the traditional top-k method and select the top k users according to the comprehensive weight value to form a trusted alliance.

On the basis of the formation of the trusted alliance, the work of analyzing the relationship between service users has been completed. The work of the selection module in the system is to refer to the service interaction history of users in the trusted alliance, including the discovery, selection and aggregation process of each service , analyze and consider the correlation between different services, provide service users with candidate services that best meet their needs according to this potential correlation, and finally complete the service selection process, which is another aspect of the technology that needs to be protected in this patent one aspect. The basic definitions and methods for service selection based on the potential correlation between services are given below:

1) The commonly used service set s _c s _c represents the sub-service set that is most commonly used for service aggregation by selected users according to the service aggregation history of all users in the trusted alliance within a certain period of time, which is denoted as s _c ={s _c1 ,s _c2 ,...s _cm }, where each service can better meet the user's needs than the other services.

2) Candidate service set c When the user selects a service, the candidate service set in the service resource pool that meets the user's needs is expressed as c={c ₁ ,c ₂ ,...c _m }.

3) Inter-service correlation R In the process of service discovery, aggregation, and selection, there is a specific connection or pattern between services, including the interaction between candidate service sets and commonly used services, and services in the entire resource pool. The size of the degree of association, the larger the value of R, the closer the service in the candidate service set is to the user's needs.

并且A∩B=Ф，公式的具体含义是：A与B同属于一个集合I，我们在找到A中物品的同时有很大的概率可以找到B中的物品。在联盟原则中有两个很重要的基本概念support和confidence。联盟原则A→B的support表示该联盟原则出现的频率，联盟原则A→B的confidence表示的是在找到实体A的同时也能找到实体B的概率，它体现出A与B之间的联系性。4) Association Rules (Association Rules) is the most important technology in the field of data mining, usually used to discover the relationship between entities in a large-scale database. The formula can be expressed as A→B, where

And A∩B=Ф, the specific meaning of the formula is: A and B belong to the same set I, when we find the items in A, we have a high probability of finding the items in B. There are two very important basic concepts support and confidence in the alliance principle. The support of the alliance principle A→B indicates the frequency of the alliance principle, and the confidence of the alliance principle A→B indicates the probability that the entity B can also be found when the entity A is found, which reflects the connection between A and B .

Figure 3 shows a schematic diagram of the correlation between services. There are two service users U ₁ and U ₂ in the trusted alliance. They have the same atomic service aggregation process for a service request S, but when they request S ₂ When mapping specific sub-services with _S5 , user U ₁ chooses sub-services S ₂₁ and S ₅₁ , while user U ₂ chooses sub-services S ₂₂ and S ₅₂ , because U ₁ and U ₂ belong to the same user Alliance, they have a high degree of similarity in service use, so it can be considered that there is a certain connection or pattern among the sub-services S ₂₁ , S ₅₁ , S ₂₂ , and S ₅₂ , that is, a potential inter-service Dependencies, with reference to which, service selection can be achieved for target service consumers.

Figure 4 shows a schematic diagram of extracting commonly used service sets by analyzing the entire process of service discovery, selection, and aggregation by synthesizing the service aggregation history of users in trusted alliances in a dynamic and heterogeneous network environment. The specific methods are expressed as follows:

The user-service call rate is used as a reference factor to form _sc . When a user k in the trusted alliance chooses service j, we record this call, and the total number of times service j is called is recorded as number(s _j , u _k ). After a period of time, we calculate the service value according to formula 6 Call rate, which is recorded as freq(s _j ,u _k ), and then sort the user-service call rates in the trusted alliance, and select the m sub-services with the highest value as the elements of the common service set.

$\mathrm{<span\; class="notranslate">\; fred</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}\frac{\mathrm{<span\; class="notranslate">\; number</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; m</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; number</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}& \mathrm{<span\; class="notranslate">\; ifnumber</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; ifnumber</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; u</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}\end{array}\right.<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$

The correlation between a certain service _ci and s _c in c can be calculated by evaluating the alliance principle existing between them. The more alliance principles exist, the stronger the correlation between c _i and s _c is; in addition, we also take the correlation between c _i and all service sets in the resource pool as another important reference factor, there is a relationship between c _i and s The more coalition principles there are, the weaker the correlation between _ci and s _c will be. In this paper, R ₁ and R ₂ are used to represent these two kinds of correlations respectively, and the correlation between services is defined as R=R ₁ *R ₂ , and the aggregation history of all services in the resource pool is used to calculate each service in the candidate service set c The support and confidence values of the services in the common service set s _c . Exclude candidate services that do not meet the threshold confidence(s _c →c)>δ, support port(s _c →c)>γ, where the values of δ and γ vary according to specific models and environments. In all candidate services that meet the threshold value, calculate the correlation R ₁ between _ci and the service in s _c according to formula 7

${\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="R"\; filenames="HDA00002485691300012.png,HDA00002485691300021.png"\; state="\{\{state\}\}">R</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; cj</span>}}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; S</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}}{\overset{\mathrm{<span\; class="notranslate">\; m</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; confidence</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; cj</span>}}<span\; class="notranslate">\; \&Right\; Arrow;</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; )</span>$

Then calculate the correlation R ₂ between _ci and all services according to formula 8, where t is the number of all alliance principles AR existing in the service resource pool, and num( _ci ,AR) represents the alliance principles including _ci participating The number of ARs.

${\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="R"\; filenames="HDA00002485691300012.png"\; state="\{\{state\}\}">R</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; log</span>}\frac{\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; num</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; AR</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>$

Then calculate the correlation R between services, return the candidate service with the highest value, and the user selects it as the optimal service.

So far, the coalition principle has been successfully used to select the service that best meets the user's needs from the candidate service set c.

Based on the construction of trusted alliance, we realize the process of selecting services for target service users by analyzing the potential correlation between services. At the same time, for the selected atomic service, update its release information in the UDDI registration center in the service selection system. When other service users request services, the time cost of the selection process can be reduced.

In order to illustrate the correlation-based service selection scheme of the present invention, we give a specific example—the user requests the best travel service for visiting Nanjing from the network to illustrate the method in detail.

In the first step, tourism-related service providers publish atomic service resources about Nanjing tourism on the Internet through a trusted third party, such as: tourist attractions, travel routes, travel methods, accommodation services, catering services, financial services and insurance services etc. These service resources form the entire service resource pool.

In the second step, the user submits a request for Nanjing tourism service resources to the service access platform, and searches for candidate services that meet the requirements. The search content includes the type of service (in this case, it can be classified as: business, high-level, medium, ordinary), service safety (such as: travel mode safety, catering safety level, etc.), reliability (such as: travel agency reputation , transportation punctuality rate, etc.) and price (normal price and preferential price, etc.), after some conditional constraints, the service is initially screened to form a candidate service resource pool for users to choose from.

In the third step, the user finds some other users who have the same or similar preferences as their own based on their past travel experience, for example, the group of users who have the same choices as themselves in terms of travel agencies, scenic spots, transportation methods, and accommodation. The target user’s similar neighbor users in terms of travel behavior, and if such users have made direct or indirect travel recommendations to the target user, then consider the trust relationship between them, combined with the similarity relationship obtained from the previous analysis, determine that both The user group that is similar to the user in terms of travel behavior and can be trusted by the user is defined as the user's trusted alliance.

The fourth step, if the users in the trusted alliance have traveled to Nanjing, they have experience in various tourist attractions, various tourist routes, various travel modes, various accommodation and catering in Nanjing, and they are familiar with Nanjing. The quality and reliability of these services have also been comprehensively evaluated. Based on their feedback, we have extracted some services that most users in the alliance are most satisfied with, and formed a set of commonly used services for travel itineraries, which will be considered by target users when selecting travel services. A reference group for service dependencies.

The fifth step, on the basis of the commonly used service set obtained in the previous step, we associate and refer to the services in the candidate service resource pool and the services in the commonly used service set, and combine all the travel information of users who have visited Nanjing to find out which services are the best. There are some fixed connections between them, such as whether the services in the two service collections have been called by some users during a trip at the same time, and the alliance principle is used to find the connections and patterns between these services, that is, between services Existing correlation is similar to finding out the correlation between the candidate service set and the services in the entire service resource pool.

In the sixth step, by analyzing and calculating the correlation between services, a group of candidate services that can best meet the needs of target users can be finally selected, and users can refer to this group of services to plan Nanjing travel, and their requests can be resolved. The user's service selection results are also updated with information that other users can refer to, which provides a reliable guarantee for users who travel to Nanjing in the future to choose various services more quickly and with better quality, and also improves the selection of travel services in the entire network efficiency.

Claims (6)

1. A service selection system based on correlation, characterized in that: it includes a publishing module, a retrieval module, an evaluation module, a database module and a selection module, the retrieval module provides a retrieval port for user access, and the database contains all relevant atomic service resources And for the evaluation data of these atomic service resources, the evaluation module provides an evaluation mechanism so that users can evaluate the services they have accepted or used, and the evaluation data is sent to the database for storage, and the selection module provides similar services based on the keywords retrieved by users The resource is for user's reference, and the publishing module provides an interface for the service provider to publish service information. When the information is received, the publishing module stores the service information in the database. 2. The selection system according to claim 1, characterized in that: said service information uses UDDI to register and release. 3. The selection system according to claim 1, wherein the database sorts the atomic service resources according to the evaluation data provided by the evaluation module, and the selection module preferentially selects the atomic service resources with high evaluation to display to the user. 4. A selection method of the correlation-based service selection system according to claim 1, characterized in that: comprising the steps of: 1) The user submits the keywords to be queried to the system through the search port provided by the search module; 2) The keyword is sent to the database, and the matching service resource data is selected from the database; 3) According to the keyword, the selection module selects the service resource closest to the keyword; 4) The retrieval module feeds back the matched resources to the user through the retrieval module. 5. The selection method according to claim 4, characterized in that: it also includes an evaluation step, after the user accepts the service, the user can write the evaluation information of the service through the evaluation module, and the evaluation mechanism is a combination system of scoring system and writing comments . 6. The selection method according to claim, wherein the resources returned by the database and the service resources recommended by the selection module also include other users' comments on the resources.

Images