CN102270140B - Time-constrained Web service flow mining method - Google Patents

Abstract

The invention discloses a time-constrained Web service flow mining method in the technical field of Web service flow mining. This method first establishes a time-constrained Web service process graph description for all candidate Web service processes and requesting Web service processes in the Web process library; then calculates the task compatibility value between the candidate and requesting Web service processes; and then respectively Establish a standardized matrix for the requesting Web service flowchart description and each candidate Web service flowchart description compatible with its tasks, and calculate the ratio between the requesting Web service flowchart description and each candidate Web service flowchart description compatible with its tasks Finally, select the candidate Web service flowchart description with the smallest difference value from the requested Web service flowchart description, and reuse the corresponding Web service process in the Web process library. The invention can ensure that the excavated service process can meet the basic task requirements required by customers and the time constraint requirements expected by users.

Description

A Time-Constrained Web Service Process Mining Method

technical field

The invention belongs to the technical field of web service flow mining, in particular to a time-constrained web service flow mining method.

Background technique

With the continuous development of Service Oriented Architecture (SOA) and technology, Web Service (Web Service) has played an increasingly important role in e-commerce, enterprise application integration and other fields. Web services are externally available distributed application components accessed through the Web. Through the combination technology of Web services, enterprises can combine existing relatively simple Web services into Web service processes according to certain business process logic. In order to adapt to the complex, changeable and fierce market competition needs of today's enterprises, on the one hand, when the enterprise implements specific Web service processes, various departments within the enterprise must work together as effectively as possible to meet the service quality and time expected by enterprise customers In order to remain invincible in the fierce market competition. Therefore, various Web service processes provided by enterprises for different business needs are often time-constrained. This is especially true for e-commerce businesses. On the other hand, enterprise application systems must be able to implement business processes as quickly and flexibly as possible according to business needs, so as to save service costs and improve efficiency. Therefore, enterprises need to use process mining technology to find out the Web service process that can meet the current latest user needs from the existing time-constrained Web service process. By realizing the repeated use of the Web service process, the service cost can be saved as much as possible, and the value-added of the Web service process can be realized.

The description languages used to describe the process of Web services include BPEL4WS, WSFL, XLANG, WSCI, WSCL, BPML, etc., which have the syntax of XML language and are similar to each other in description ability. Many systems and methods supporting Web service process modeling have been developed and designed at home and abroad for several years. Representative systems include the eFlow system developed by HP Labs and the SERV system developed by the University of New South Wales in Australia. The former supports graphical process modeling and dynamic service composition, and the latter uses state diagrams to model service process logic. However, the current Web service process research and application mainly focus on service process modeling, verification, simulation and execution based on workflow technology, rather than service process mining. The current tools and methods for process mining, such as the ProM process mining framework and DAGAMA process discovery tool, are mainly to ensure that the mined service process can meet the basic task requirements required by customers, but cannot ensure that the mined service process can meet The demand for efficient services expected by users, that is, to meet the time constraints expected by customers. For e-commerce and other enterprises whose survival is based on service quality and efficiency, this is undoubtedly an important problem to be solved urgently.

Contents of the invention

Aiming at the disadvantages mentioned in the background art above that the current process mining technology cannot meet the requirements of users, the present invention proposes a time-constrained Web service process mining method.

The technical solution of the present invention is a time-constrained Web service process mining method, which is characterized in that the method comprises the following steps:

Step 1: Create a graph description of the time-constrained Web service process for all Web service processes and request Web service processes in the Web process library;

Step 2: Calculate the task compatibility value between the request Web service flowchart description and each candidate Web service flowchart description, and when the compatibility value is greater than the specified threshold, the two tasks are compatible;

Step 3: On the basis of step 2, establish a standardized matrix for each candidate Web service flowchart description that is compatible with the task of the request Web service flowchart description;

Step 4: Based on the standardized matrix, calculate the difference value between the request Web service flowchart description and each candidate Web service flowchart description compatible with its task;

Step 5: Based on the difference value, select the candidate Web service flowchart description with the smallest difference value from the requested Web service flowchart description, and reuse the corresponding Web service process in the Web process library.

The graph description of the Web service process can be determined by a set of nodes and a set of edges, where the node is the name of the Web service in the service process, and the edge is a directed edge with a time constraint mark between two nodes.

The calculation formula of the task compatibility value is:

$\mathrm{<span\; class="notranslate">\; taskcomp</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; \&cap;</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{{\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}<span\; class="notranslate">\; |</span>}{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; \&cup;</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{{\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}<span\; class="notranslate">\; |</span>}$

in:

R is the description of the flow chart of requesting Web services;

Q _i is the flow chart description of the i-th candidate Web service;

taskcomp(R, Q _i ) is the task compatibility value of R and Q _i ;

V _R is the set of nodes contained in R;

为Q_i中所包含的结点集合。

is the set of nodes contained in Q _i .

The formula for calculating the normalization matrix described in the flow chart of requesting Web services is:

in:

M(i, j) is the element of the standardized matrix described by the request Web service flowchart;

v _i is the i-th node in the union set of Web service name sets contained in R and Q _i ;

v _j is the jth node in the union set of Web service name sets contained in R and Q _i .

The calculation formula of the normalization matrix described by the candidate Web service flowchart is:

in:

M'(i, j) is the element of the standardized matrix described by the candidate Web service flowchart;

v _i is the i-th node in the union set of Web service name sets contained in R and _Qr ;

v _j is the jth node in the union set of Web service name sets contained in R and Q _r ;

[a, b] represents the time constraint mark corresponding to the edge (v _i , v _j ) in R;

[a', b'] represents the time constraint mark corresponding to the side (v _i , v _j ) in Q _r ;

[a, b] ∩ [a', b'] means that part of the time constraints where [a, b] and [a', b'] overlap in time

Note, such as [2,5]∩[3,7]=[3,5];

Dur([a, b]) represents the interval time value of the time stamp [a, b], namely Dur([a, b])=b-a;

If [a,b]∩[a′,b′] has no overlap in time, then

Dur([a,b]∩[a′,b′])=0.

The calculation formula of the difference degree value is:

d _r =DiagonalSum((MM′)×(MM′) ^T )

in:

d _r is the difference value;

M-M' is the matrix obtained after subtracting the corresponding element values of matrix M and matrix M';

(MM') ^T represents the transpose matrix of matrix MM'.

The invention can not only ensure that the excavated service process can meet the basic task requirements required by customers, but also ensure that the excavated service processes can meet the time constraint requirements expected by users.

Description of drawings

Figure 1 is a schematic diagram of a time-constrained Web service process mining algorithm;

Figure 2 is a diagram description of requesting Web service process and two candidate Web service processes;

Figure 3 shows the matrix difference calculation between the requesting Web service process and the first candidate Web service process;

Figure 4 shows the matrix difference calculation between the requesting Web service process and the second candidate Web service process.

Detailed ways

The preferred embodiments will be described in detail below in conjunction with the accompanying drawings. It should be emphasized that the following description is only exemplary and not intended to limit the scope of the invention and its application.

The purpose of the present invention is to propose a time-constrained Web service process mining method for the deficiencies of the existing Web service process mining technology, so as to ensure that the mined service process can meet the basic task requirements required by customers and the expected requirements of customers. Time-constrained demand, so as to promote effective collaborative work among various business-related departments within the enterprise, save enterprise service operating costs and improve enterprise economic benefits while fulfilling customer needs with high quality.

The technical scheme of the invention is a time-constrained Web service process mining method. This method converts the Web service process into a graph description form, and then converts it into a corresponding standardized process matrix for difference calculation, and further uses the difference value between Web service processes to select the Web service process that is most likely to meet customer needs for reuse. .

A kind of time-constrained Web service process mining method proposed by the present invention mainly includes the following steps:

1. Establish a graph description of the time-constrained Web service process for all Web service processes in the Web process library and request Web service processes.

The graph description of a Web service process can be determined by its node set and edge set, where the node described in the graph is the name of the Web service in the service process, and the edge described in the graph is the directed edge between two nodes , which is used to indicate the activation relationship between the Web services corresponding to these two nodes, and the time constraint mark of [a, b] on the directed edge indicates that the Web service at the start end needs to be completed within a to b time units In order to further activate the terminal Web service, a≤b. The node set and edge set described in the graph can be stored through two database tables respectively.

Specifically include the following sub-steps:

1.1 Directly use graph description to represent the process of requesting Web services. Requesting Web service process is an ideal Web service process formulated for given customer requirements (including basic task requirements and time constraint requirements).

1.2 For all the candidate Web service processes described by the process description language in the Web service process library, use the currently commonly used XML parser tool to extract all the Web service names and service activation relationships from the Web service process description, and then according to the Web The system records the log of the service process executed before, and extracts the time constraint relationship between service activations, so as to establish and store its graph description for each candidate Web service process. Use Q ₀ , Q ₁ , ... Q _m-1 to represent the graph descriptions of all candidate Web service processes.

2. Calculate the task compatibility value of the request Web service flowchart description R and each candidate Web service flowchart description Q ₀ , Q ₁ , . . . Q _m-1 .

Specifically include the following sub-steps:

2.1 Set the initial value of the counter i to 0.

2.2 The calculation formula of task compatibility value taskcomp is:

$\mathrm{<span\; class="notranslate">\; taskcomp</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; \&cap;</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{{\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}<span\; class="notranslate">\; |</span>}{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; \&cup;</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{{\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}<span\; class="notranslate">\; |</span>}$

in:

R is the flow chart description of requesting Web services;

Q _i is the i-th candidate Web service flowchart description (0≤i≤m-1);

taskcomp(R, Q _i ) is the task compatibility value of R and Q _i ;

分别表示R和Q_i中所包含的结点集合。 _VR and

represent the node sets contained in R and Q _i respectively.

Set the task compatibility threshold to 0.6. Task compatibility is mainly to check the ratio of the same service name contained in two service processes. If the taskcomp values of R and Q _i are greater than or equal to this threshold, they are task compatible; otherwise, they are task incompatible.

2.3 If R and Q _i are task-compatible, then add _Qi to the task-compatible Web service flowchart description queue.

2.4 If i<m-1 at this time, then set i=i+1, and then go to sub-2.2, select the next candidate Web service process Q _i from the Web service process library to compare task compatibility with R. Otherwise, if i≥m-1, stop the compatibility calculation described by the flowchart.

3. Establish a standardized matrix for the requesting Web service flowchart description and each candidate Web service flowchart description compatible with its task.

The service flow diagram descriptions in the queue of Web service flow diagram descriptions compatible with the requesting Web service flow diagram description R task are denoted as Q ₀ , Q ₁ , ... Q _k-1 , 0≤k≤m.

即R和Q_r的标准化矩阵中的行和列元素来自R和Q_r的节点集的并集。该并集中的n个元素记为v₀，v₁，…v_n-1。The normalized matrices of R and Q _r (0≤r≤k-1) are denoted by M and M' respectively, and they are both n×n matrices. in,

That is, the row and column elements in the normalized matrices of R and _Qr come from the union of the node sets of R and _Qr . The n elements in this union are denoted as v ₀ , v ₁ , . . . v _n-1 .

Specifically include the following sub-steps:

3.1 Based on the graph description of the Web service process, establish a standardized matrix R for requesting the Web service flowchart description.

The standardized matrix of requesting Web service flow chart description R is denoted as M, and the value of each element M(i, j) is calculated as follows:

in:

)中的第i个结点和第j个结点，且i和j满足条件0≤i，j≤n-1，且

即并集的基数。v _i and v _j represent the union of the Web service name sets contained in R and Q _r (ie

) in the i-th node and j-th node, and i and j satisfy the conditions 0≤i, j≤n-1, and

That is, the cardinality of the union.

3.2 Based on the graph description of the Web service process, a standardized matrix is established for each candidate Web service flowchart description.

The standardized matrix described by the candidate Web service process Q _r graph is denoted as M′, and the value of each element M′(i, j) is calculated as follows:

in:

v _i and v _j respectively represent the i-th node and j-th node in the union set of Web service name sets contained in R and Q _r , and i and j satisfy the conditions 0≤i, j≤n-1;

[a, b] represents the time constraint mark corresponding to the edge (v _i , v _j ) in R;

[a', b'] represents the time constraint mark corresponding to the side (v _i , v _j ) in Q _r ;

[a,b]∩[a′,b′] represents the part of the time constraint mark where [a,b] and [a′,b′] overlap in time, such as [2,5]∩[3,7] =[3,5];

Dur([a,b]) represents the interval time value of the time stamp [a,b], that is, Dur([a,b])=b−a.

If [a,b]∩[a′,b′] has no overlap in time, then

Dur([a,b]∩[a′,b′])=0.

4. Based on the standardized matrix, calculate the difference degree value between the request Web service flowchart description and each candidate Web service flowchart description compatible with its task.

4.1 Set the counter r disposition to 0.

4.2 Calculate the difference value d _r between R and Q _r , the difference value d _r can be calculated by the following formula:

d _r =DiagonalSum((MM′)×(MM′) ^T )

in:

M-M' represents the matrix obtained by subtracting the corresponding element values of the matrix M and the matrix M'; the matrix is multiplied, and the result is still a matrix;

(MM') ^T represents the transposition matrix of matrix MM';

DiagonalSum is used to calculate the sum of the diagonal element values of the matrix, which is the difference between R and Q _r , that is, d _r .

4.3 If r<k-1, then r<r+1, and turn to 4.2, select the next candidate process Q _r and R from the Web service process library to perform task compatibility calculation and difference calculation.

5. Based on the difference value, select the candidate Web service flowchart description with the smallest difference value from the requested Web service flowchart description, and reuse the corresponding Web service process described in the graph in the Web process library.

The differences between all candidate Web service flowchart descriptions compatible with R tasks and R are denoted as d ₀ , d ₁ ,...d _k-1 . The candidate Web service process most similar to R is denoted as Q _x , where the subscript x is determined by d _x , the formula is as follows:

d _x ＝min{d ₀ ，d ₁ ，…d _k-1 }

That is to say, the candidate Web service process corresponding to the smallest value among all the difference values will be the reusable service process most similar to the requested service.

The feature of the time-constrained Web service process mining method proposed by the present invention is that by converting the Web service processes into their graph descriptions, and then comparing the process matrix of the requesting Web service process with the process matrix of the candidate Web service process Finally, the candidate Web service process with the smallest difference from the requesting Web service process is excavated for process reuse. The method proposed by the present invention conforms to the basic human cognition that the greater the degree of similarity, the smaller the degree of difference. The process information extraction and correlation matrix operations involved in this method can be realized by some current mainstream software development tools, and it also has low computational complexity. The complexity in the worst case is O(m*n ³ ), where m is the total number of candidate Web service processes in the process library, and n is the total number of Web services in the process library. The application of this method can achieve our expected goal. It can not only ensure that the mined service process can meet the basic task requirements required by customers, but also ensure that the mined service process can meet the time constraint requirements expected by users.

This embodiment includes two major aspects:

1. Generation of graph descriptions of time-constrained Web service processes

The reusable Web service process is usually saved in the service process library through the Web service process description file. Taking BPEL service description language as an example, a Web service process file is stored in XML format, which contains all the Web service names involved in the process and the mutual activation relationship between Web services. Therefore, a Web service process file described by BPEL can be parsed by an XML parser to generate a graph description of the service process. There are many XML parsers currently available, such as IBM's XML4J, Oracle's XML Parser for Java, and so on.

Specific steps are as follows:

1.1 Analyze each Web service process file in the Web process library in sequence, and extract all Web service names and Web service activation relationships from the Web service process according to the structured activity description of the service process description language. The structured activity description specifies the name of the Web service activity through the <case> tag, and indicates the sequence, OR-split, AND- Conditional relationships such as split and loop. 1.1 Specifically, it can be divided into the following sub-steps:

1.1.1 Extract all Web service (activity) names from the structured activity description through the <case> tag. All extracted service names are stored in a 1-dimensional array, and each element in the array is a service name.

1.1.2 Extract all activation relationships between service activities from the structured activity description through the above conditional relationship tags. The relationships between all service activities are kept in a 4-dimensional array. The first two components of each column vector of this 4-dimensional array represent the activation relationship between two service activities.

1.2 According to the system record logs previously executed by the Web service process, the time constraint relationship between service activations is extracted. The system record log saves the start time and end time of the Web service execution in the process. For a time constraint [a, b] of a service activation relation edge, a is equal to the end time of the service at the beginning of the edge minus the start time, and b is equal to the start time of the service at the end of the edge minus the start time of the service at the beginning. Further, the a and b values are respectively stored in the last two components of the column vector corresponding to the activation relationship in the 4-dimensional array.

1.3 By saving a 1-dimensional array of Web service names and a 4-dimensional array of service activation relations and time constraints, a graph description of a Web service process can be determined. Fig. 2a is a flowchart description of requesting a Web service according to an embodiment of the present invention, denoted as R. Figure 2b and Figure 2c are graph descriptions of two candidate Web service processes, marked as Q ₀ and Q ₁ respectively. They can be determined via a 1-dimensional array and a 4-dimensional array respectively. For example, R is stored like this:

R's web service name 1-dimensional array RNode[]={A2, A3, A4, A5, A6, A7, A9, A10}, R's activation relationship 4-dimensional array RRel[][4]={{A2, A3, 2, 3}, {A2, A4, 2, 5}, {A3, A5, 4, 6}, {A3, A6, 3, 4}, {A5, A9, 3, 4}, {A6, A9, 2, 3}, {A4, A7, 3, 4}, {A7, A10, 4, 6}, {A9, A10, 2, 7}}. Corresponding storage arrays Q0Node, Q0Rel, Q1Node and Q1Rel etc. of Q0 and Q1 may also be established accordingly.

in:

Q0Node[]={A1, A2, A3, A4, A5, A7, A8, A9, A10, A11};

Q0Rel[][4]={{A1, A2, 3, 5}, {A2, A3, 2, 4}, {A2, A4, 3, 5}, {A3, A5, 4, 6}, {A4 , A7, 3, 5}, {A4, A8, 6, 9}, {A5, A9, 3, 5}, {A1, A2, 3, 5}, {A7, A11, 4, 5}, {A8 , A11, 2, 3}, {A9, A10, 3, 6}, {A11, A10, 2, 3}};

Q1Node[]={A1, A2, A3, A4, A6, A7, A8, A9, A10};

Q1Rel[][4]={{A1, A2, 2, 3}, {A2, A3, 2, 4}, {A2, A4, 2, 4}, {A2, A8, 4, 6}, {A3 , A6, 3, 5}, {A4, A7, 3, 4}, {A6, A9, 2, 3}, {A9, A10, 2, 6}, {A8, A10, 5, 7}, {A7 , A10, 6, 8}}.

2. Based on the graph description of the Web service process, the candidate Web service process most similar to the requesting Web service process R is mined from the service process library. In this embodiment, Q ₀ and Q ₁ are used as graph descriptions of candidate Web service processes, and the Web service process most similar to R is found from Q ₀ and Q ₁ . This aspect is specifically divided into the following steps:

2.1 Calculate the task compatibility value taskcomp of R and all candidate Web service processes. Including the following sub-steps:

2.1.1 Add Q ₀ and Q ₁ to the Web flowchart description queue, and m=2 at this time. Set a counter i with an initial value of 0;

2.1.2 Calculate the value of taskcomp(R, Q _i );

2.1.3 If the compatibility value of R and Q _i is not less than 0.6, then add Q _i to the queue described in the task compatible Web service flowchart;

2.1.4 If i<m-1, then set i=i+1, and then return to 2.1.2; otherwise, execute 2.2;

In the present embodiment, through the above sub-steps, the compatibility value of R and _Q0 is:

taskcomp(R,Q ₀ )=7/11=0.636≥0.6

The compatibility values for R and _Q1 are:

taskcomp(R,Q ₁ )=7/10=0.7≥0.6

Therefore, both Q ₀ and Q ₁ are added to the queue described by the task-compatible Web service flow diagram.

2.2 Calculate the difference degree value described by R and all candidate task-compatible Web service flow charts. This step includes the following sub-steps:

2.2.1 For the convenience of counting processing, name the flow chart descriptions in the candidate Web service process queue compatible with R tasks in order: Q ₀ , Q ₁ , ... Q _k-1 , and set the counter r with an initial value of 0 . In the task compatible service process queue in this embodiment, k=2, and includes Q ₀ and Q ₁ at the same time;

2.2.2 Establish the normalization matrix M′ of Q _r ;

2.2.3 Establish a standardized matrix M of R for Q _r ;

2.2.4 Calculate the value of d _r , and add d _r to the difference value queue;

2.2.5 If r<k-1, add 1 to the counter r, and return to 2.2.2 to continue the calculation of the difference degree of the next task-compatible Web service process. Otherwise, go to 2.3;

In this embodiment, according to the aforementioned definition of the normalization matrix for generating the service process, the normalization matrix for calculating the difference between R and Q ₀ is given, as shown in Fig. 3a and Fig. 3b respectively. The standardized matrix for calculating the degree of difference between R and _Q1 is also given, see Figure 4a and Figure 4b, respectively.

In the process of calculating the difference value of R and each Web service process Q _r compatible with its task, two auxiliary matrices will also be generated. One is the matrix obtained by subtracting the corresponding elements of the matrix M and the matrix M′, which is denoted as the matrix M″ for convenience. The other auxiliary matrix is the matrix obtained by multiplying the transposed matrix of M″ and M″, Denoted as M'" for convenience. In this embodiment, the two auxiliary matrices M" and M'" corresponding to Q ₀ are shown in Fig. 3c and Fig. 3d respectively, and the two auxiliary matrices M" and M'" corresponding to Q ₁ are shown in Fig. 4c and Fig. 4d.

Calculating the difference value of d ₀ is to sum the values of the elements on the diagonal in the M′″ matrix of Q ₀ (that is, the elements marked in gray in the matrix), and obtain d ₀ =8.27. Similarly, the value of d ₁ It is to add the values of the elements on the diagonal in the M'" matrix of Q ₁ to obtain d ₁ =5.4;

2.3 From the difference degree value queue d ₀ , d ₁ , ...d _k-1 , select a difference degree value d _x with the smallest difference degree value; then, Q _x is a Web service flow chart compatible with the task most similar to R describe. The specific sub-steps for selecting the minimum difference value are as follows:

2.3.1 Set the counter q, the initial value is 1; set the variable x, its initial value is 0;

2.3.3 If d _q <d _x , then assign the value of q to x;

2.3.4 If q<k-1, then set q=q+1, return to 2.3.3; otherwise, end.

In this example, the difference degree value queue only includes two difference degree values, namely d ₀ and d ₁ . After executing 2.3, the integer variable x in d _x stores the number of the service process with the smallest difference value in the difference value queue. In this embodiment, the value of d ₁ is the smallest, so the value of x is 1. Therefore, in the Web service process library, the candidate Web service process corresponding to the task-compatible Web service flow chart description _Q1 is most similar to the requesting Web service process. Therefore, the candidate Web service process corresponding to the smallest value among all the difference values will be the reusable service process most similar to the requested service.

The above is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present invention can easily think of changes or Replacement should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the protection scope of the claims.

Claims (1)

1. A time-constrained Web service process mining method is characterized in that the method comprises the following steps: Step 1: Establish a graph description of a time-constrained Web service process for all Web service processes and request Web service processes in the Web process library; the graph description of the Web service process can be determined by a collection of nodes and a collection of edges, and the nodes is the name of the Web service in the service process, and the edge is a directed edge marked with a time constraint between two nodes; The process described in the diagram describing the establishment of a time-constrained Web service flow is: Step 101: directly use the graph description to represent the requesting Web service process; Step 102: For all candidate Web service processes described by the process description language in the Web service process library, use an XML parser tool to extract all Web service names and service activation relationships from the Web service process description, and then The system records the log of the process executed before, and extracts the time constraint relationship between service activations, so as to establish and store its graph description for each candidate Web service process; expressed by Q ₀ , Q ₁ ,···Q _m-1 A graph description of all candidate Web service processes, where m is the total number of all candidate Web service processes in the Web process library; Step 2: Calculate the task compatibility value between the request Web service flowchart description and each candidate Web service flowchart description. When the compatibility value is greater than the specified threshold, the two tasks are compatible; the specific process is: Step 201: Set the initial value of the counter i to 0; Step 202: The calculation formula of the task compatibility value taskcomp is: $\mathrm{<span\; class="notranslate">\; taskcomp</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; \&cap;</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{{\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}<span\; class="notranslate">\; |</span>}{<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}}<span\; class="notranslate">\; \&cup;</span>{\mathrm{<span\; class="notranslate">\; V</span>}}_{{\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}<span\; class="notranslate">\; |</span>}$ in: R is the flow chart description of requesting Web services; Q _i is the i-th candidate Web service flowchart description, and 0≤i≤m-1; taskcomp(R,Q _i ) is the task compatibility value of R and Q _i ; V _R and V _Qi represent the node sets contained in R and Qi respectively; Set the task compatibility threshold, if the taskcomp value of R and Q _i is greater than or equal to the set threshold, it means that they are task compatible, go to step 203; otherwise, the task is incompatible, go to step 204; Step 203: Add Q _i to the task-compatible Web service flowchart description queue; Step 204: If i<m-1 at this time, then set i=i+1, and then go to sub-step 202, select the next candidate Web service process Q _i from the Web service process library for task compatibility Compare; otherwise, if i≥m-1, stop the compatibility calculation described in the flow chart and execute step 3; Step 3: On the basis of step 2, establish a standardized matrix for each candidate Web service flowchart description that requests the Web service flowchart description and its tasks; the specific steps include: Step 301: The service flowchart descriptions in the Web service flowchart description queue compatible with the requesting Web service flowchart description R task are denoted as Q ₀ , Q ₁ ,...Q _k-1 , 0≤k≤m; Step 302: Based on the graph description of the Web service process, establish a standardized matrix for requesting the Web service flowchart description R; The flow chart of requesting Web services describes the standardized matrix of R as M, and the value of each element M(i,j) is calculated as follows:

in: v _i and v _j represent the i-th node and the j-th node in the union of the Web service name sets contained in R and Q _r (that is, V _R ∪ V _Qr ), and i and j satisfy the condition 0≤ i≤n-1, 0≤j≤n-1, and n=|V _R ∪V _Qr |, which is the base of the union; Step 303: Based on the graph description of the Web service process, a standardized matrix is established for each candidate Web service flowchart description; The standardized matrix described by the candidate Web service process Q _r graph is denoted as M′, and the value of each element M′(i,j) is calculated as follows:

in: v _i and v _j represent the i-th node in the union of the Web service name sets contained in R and Q _r respectively Point and the jth node, and i and j satisfy the conditions 0≤i≤n-1, 0≤j≤n-1; [a,b] represents the time constraint mark corresponding to the edge (v _i , v _j ) in R; [a′,b′] represents the time constraint mark corresponding to the edge (v _i , v _j ) in Q _r ; [a,b]∩[a′,b′] represents the part of the time constraint where [a,b] and [a′,b′] overlap in time Remember, such as [2,5]∩[3,7]=[3,5]; Dur([a,b]) represents the interval time value of the time stamp [a,b], that is, Dur([a,b])=b-a; if [a,b]∩[a′,b′] in time There is no overlapping part, then Dur([a,b]∩[a′,b′])=0; Step 4: Based on the standardized matrix, calculate the difference value between the request Web service flowchart description and each candidate Web service flowchart description compatible with the task; the specific steps include: Step 401: set the counter r disposition to 0; Step 402: Calculate the degree of difference d _r between R and Q _r , the value of the degree of difference d _r can be calculated by the following formula: d _r =DiagonalSum((MM′)×(MM′) ^T ) in: M-M' represents the matrix M obtained after subtracting the corresponding element values of the matrix M' Matrix; the matrix is multiplied, and the result is still a matrix; (MM') ^T represents the transposition matrix of matrix MM'; DiagonalSum is used to calculate the sum of the diagonal element values of the matrix, which is R and The difference degree value between Q _r , namely d _r ; Step 403: If r<k-1, then r<r+1, go to step 402, select the next candidate process Q _r and R from the Web service process library to perform task compatibility calculation and difference calculation; otherwise, Execute step 5; Step 5: Based on the difference value, select the candidate Web service flowchart description with the smallest difference value from the requested Web service flowchart description, and reuse the corresponding Web service process in the Web process library; the specific steps include: Step 501: Record the differences between all candidate Web service flowchart descriptions compatible with R tasks and R as d ₀ , d ₁ ,...d _k-1 ; record the candidate Web service process most similar to R as Q _x , where the subscript x is determined by d _x , the formula is as follows: d _x =min{d ₀ ,d ₁ ,···d _k-1 } Step 502: Select the candidate Web service process corresponding to the smallest value among all the difference values as the reusable service process most similar to the requested service.

Images