KR20080100931A - Method of semantic web service discovery using process-based ontology - Google Patents

Abstract

An intelligent web service detection method using the process information enhancing the accuracy rate and the recall factor of the service discovery is provided to take the process match making method for being satisfied with the query of user based on the process base ontology expressing the process information. An intelligent web service detection method using the process information includes the steps of: the process signature match making step(s120) which uses the process base ontology, and calculates the syntactic similarity which is the process name, the input name or the output name and discovers the process; the process structure match making step discovering the upper process including the process having the specific process control structure; the process concept match making step finding the process included in the specific concept.

Description

Method of Semantic Web Service Discovery using Process-based Ontology

1 is an example illustrating a similarity calculation method for two input names according to the discovery method of the present invention.

2 is an embodiment of a process signature matchmaking algorithm according to the discovery method of the present invention,

3 is an example of a process having a sequential structure in the form of a complex process;

4 is an example of a process having a parallel structure in the form of a complex process,

5 is an example of a process having a selection structure in the form of a complex process;

6 is an embodiment of a process structure analysis algorithm according to the discovery method of the present invention,

7 is an embodiment of a process concept matchmaking algorithm according to the discovery method of the present invention,

8 is another embodiment of a process concept matchmaking algorithm according to the discovery method of the present invention.

The present invention proposes a process matchmaking method for finding a process satisfying a user's query based on a process-based ontology representing process information. In detail, a method for discovering a web service, the process signature matching process of using a process-based ontology to find a process by calculating syntactic similarity with respect to a process name, an input name, or an output name; A process structure matchmaking step of finding an upper process including a process having a specific process control structure; The present invention proposes a method of discovering a process-based web service, comprising a process concept matchmaking step of finding a process included in a specific concept.

Web services are software systems designed to provide interaction between machines that are commonly available on a network. The most widely used standards of Web services to date are Web Service Description Language (WSDL) for the description of services and Universal Description Discovery and Integration (UDDI) for the discovery of services. The service provider describes the service using WSDL and registers the service in the UDDI registry. The service requester retrieves the desired service through the UDDI registry. However, WSDL only describes the syntactic information of the service, and UDDI provides search and matching of web services based on keywords. A good search engine for the initial discovery of web services meant finding as many web services as possible for a single keyword. However, with the exponential expansion of the volume of Internet information, such keyword-based web service discovery is rather a user-friendly web service. It is a stumbling block to discovery. Users had to find hundreds of pages of web services in order to find the information they wanted, and they had to find their own web services, and service providers had to spend a lot of time and money to build and maintain unautomated services.

In addition, since keyword-based web service discovery is determined and provided by a mechanical operation based on a combination of several words entered by a user, there is a limit in how the user expresses a desired service in detail so as to have a personally accurate meaning. do.

In 2000, the concept of the semantic web appeared. The Semantic Web is a semantic based search technique, not based on probability and regular language. The semantic basis is the expression of knowledge, reasoning and interpretation, which are human intellectual domains, with concepts and languages that can be understood by computers. Therefore, the computer recognizes the meaning of the information connected to the web itself, and selectively obtains and integrates or processes the information according to the meaning so that the accurate information of the individual meaning can be provided to the user individual rather than the accurate information of the objective meaning.

The most important technology of the Semantic Web is ontology. Ontologies define the concepts, the relationships between them, and provide a rule of inference by expressing them in a hierarchical manner.

Semantic web service discovery is retrieved using the service profile of Web Service Modeling Ontology (WSMO) and OWL-based Web Service Ontology (OWL-S). A service profile is an ontology that provides the information needed to discover a service. It consists of three pieces of information: the person or organization that provides the service, the capabilities of the service, and the characteristics of the service. The capabilities of the service represent information changes and state changes. The information change is an input, which is information required for execution, and an output, which is information of execution result. State changes are represented by preconditions, which are conditions for the successful execution of a service, and effects, which are expected results of service execution. The characteristics of a service can be described using a classification or quality of the service.

Recent research on semantic web service discovery includes the matchmaker project, the METOR-S project, and the WSMO discovery specification.

The Semantic Web Service Discovery Study by the Matchmaker Project not only finds the service defined by the service profile of OWL-S using the OWL-S matching engine, but also searches the service using UDDI. The OWL-S matching engine searches for a service by comparing only the degree of matching between the input / output concepts defined in the service profile. This degree of matching is defined using four matching algorithms. As an example, the matching algorithm for output is correct when the user's querying output, outR, and the registered service's output, outA, are the same when outR and outA are the same or outR is a subclass of outA. (exact) matches. It is simply defined as plugIn when outA contains outR, and vice versa as the other, and not otherwise.

The METOR-S project is about discovering services when adding services in an e-workflow. The service is defined using the profile ontology defined in version 0.7 of OWL-S. This research searches for services using service name, description, input / output, and quality of service (QoS) information. These elements perform service matching through three similarity calculations. That is, syntactic similarity, operational similarity, and semantic similarity. Syntax similarity is obtained by using syntax matching for service name and description. Operational similarity is calculated using the QoS of the service, that is, time, cost, and reliability. Semantic similarity is obtained by calculating semantic inclusion and syntactic similarity between input and output. The search results are sorted in order of the largest similarity value based on the similarity method selected by the user and displayed to the user.

Discovery studies using WSMO have not yet been specifically developed, with the exception of the discovery methods defined in the WSMO discovery specification. The discovery methods defined in the WSMO discovery specification provide the purpose discovery, web service discovery and service discovery methods. The purpose discovery is to look up existing defined objectives in the objective repository. In other words, the user can find and modify a purpose similar to his or her own requirements and reuse it. Web service discovery refers to finding a semantic web based service that meets the purpose described by the user. Service discovery refers to the process of checking whether a service found through web service discovery specifically meets the user's requirements.

However, this semantic web service discovery also has its limitations.

The first problem is that semantic web service discovery, which discovers web services using the format of the service's name, input / output, and precondition / effect, has the same output for the same input value in a similar class. If a value occurs, the user recognizes it as desired information and provides a corresponding service. The book sale service is a query that asks a user who wants to find an Internet bookstore that can pay for a book by phone number in the language "find a book sale service that accepts payment by telephone number" (Q1). In class, the input / output is telephone number / payment. The services found under these conditions are the service 1 and the telephone number that are required in the process for confirmation when the ID and password are lost in the login for payment in the book sale service. In the book sale service, service 2 that requires a telephone number is not distinguished in the payment process, and both services are provided to the user in the same way. Therefore, when discovering a web service, you need to search for the service by searching for related processes along with input / output and precondition / effect.

The second problem is that it only retrieves the simple concept of a process. The OWL-based web service (OWL-S) process model does not provide a concept and syntax for process ontology, so it is impossible to search for processes that are semantically related to the user's query. .

The third problem is that detailed information about the performance of the service is not known. Therefore, it is impossible to execute a query that contains internal behavior, which is an internal process that constitutes a service. Deferred payment after inside a book sale service when the query in the language of a user's request to find an Internet bookstore that can pay for a book after receiving a book is "find a book sale service allowing deferred payment after delivery" (Q2) No search for internal behavior such as delivery is possible.

Therefore, such a web service discovery technique of the semantic web causes a problem that the user cannot effectively express the needs of the individual, together with the provision of unwanted information and the provision of restricted information.

In order to solve the above problems, we propose a process matchmaking method that satisfies a user's query based on a process-based ontology expressing process information.

The present invention uses a process-based ontology in the discovery method of a web service to achieve the above technical problem, and process signature matchmaking for discovering a process by calculating syntactic similarity with respect to a process name, an input name, or an output name. step; A process structure matchmaking step of finding an upper process including a process having a specific process control structure; It provides a process-based web service discovery method comprising a process concept matchmaking step of finding a process included in a particular concept.

At this time, if there is no other definition in the technical terms and scientific terms used, it has a meaning commonly understood by those of ordinary skill in the art.

In addition, the following embodiments are provided as examples to sufficiently convey the spirit of the present invention to those skilled in the art. Accordingly, the present invention is not limited to the embodiments described below and may be embodied in other forms.

The process-based ontology (hereinafter referred to as TM-S ontology) uses a topic map, which is an ontology language, and includes a process signature ontology including a definition of a process name or a name for an input / output; A process structure ontology that includes a definition for a process control structure; It includes a process concept ontology that includes definitions of process concepts, definitions of relationships between concepts, definitions of perspectives on concepts, definitions of classifications between concepts, or definitions of views on classifications.

The process signature ontology defines text information for the process and input / output which is request information / result information associated with the process. In other words, the process signature ontology contains the definition of the process name and the input / output name. The process structure ontology is an ontology for describing the internal behavior of a service. It is a sequence, split, split and join, and-order, and choice. ), Including the definition of control structures needed to express process structures such as if-then-else and iterates. The process concept ontology defines process concepts from various viewpoints, and defines process concepts, definitions of relationships between concepts, definitions of perspectives on concepts, definitions of classifications between concepts, or definitions of classification concepts. Include.

In describing the TM-S ontology using XTM, the concept of the process structure, the input / output of the process signature ontology, or the control structure of the process structure ontology is defined as the topic of the topic map. In addition, process signature ontology process names or input / output names are defined as the topic's basename. The relationship between the process of the process signature ontology and the input / output or the control structure between the processes of the process structure ontology is expressed using the relationship of the topic map. Process concept ontology defines concepts using topics in the topic map, and hierarchies between concepts using associations that include superclass-subclass or superproperty-subproperty. Express multiple relations using relations, and express scope of concepts using scope of relations.

In general, matchmaking means finding services that are sufficiently similar to a user's query. In the present invention, a process-based web service is found to find processes satisfying a user's query based on the TM-S ontology. (TM-S matchmaking) method is provided.

The TM-S matchmaking method of the present invention is based on the TM-S ontology, the process signature matchmaking step of finding a process by calculating the syntactic similarity with respect to the process name, input name or output name; A process structure matchmaking step of finding an upper process including a process having a specific process control structure; And a process concept matchmaking step of finding a process included in a specific concept.

Process Signature Matchmaking and Process Structure Matchmaking provides the ability to improve the accuracy of process discovery by discovering the processes the user wants using an ontology that defines internal service information. Process concept matchmaking also has the effect of increasing the recall of process discovery by discovering semantically related processes.

The process signature matchmaking step is to match a process name with an input / output name associated with that process using a process signature ontology. Process signature matchmaking performs matching by calculating syntactic similarities between process names and input / output names. The syntax similarity is calculated by using Lucene's scoring algorithm, and the similarity between two variables is calculated and has a value of 0 to 1 depending on the degree of similarity. When calculating syntax similarity, each has a weight according to the process name and the input / output name. Since the input / output is additional information of the process, the weight for the process name is greater than the weight for the input / output name, and the sum of the weight for the process name and the weight for the input or output name is 1 or less. . The weight indicates the degree of matching desired by the service requester to the process name and the input / output name. For example, the syntax similarity between the process name and the input name is calculated by Equation 1 below.

(Formula 1)

Α is a weight for the process name, and β is a weight for the input name. The R.pn is a process name queried by the service requester, and the S.pn is a process name registered by the service provider. The SynPN () function calculates the syntactic similarity between the process names R.pn and S.pn using the Lucene scoring algorithm. Similarly, the R.in is an input name queried by the service requester, S.in is an input name registered by the service provider, and the SynIN () function is syntactically related to the input names R.in and S.in using Lucene scoring algorithm. Calculate the similarity. Thus, SynPN (R.pn, S.pn) provides syntax similarity for process names, and SynIN (R.in, S.in) provides syntax similarity for input names.

As another example, the syntax similarity between the process name and the output name is calculated by Equation 2 below.

(Formula 2)

The input name and output name have the same value weight β, the R.out is the output name requested by the service requester, the S.out is the output name registered by the service provider, and the SynOUT () function is Lucene The scoring algorithm is used to calculate the syntactic similarity between the output names R.out and S.out.

If there are multiple inputs / outputs in a process, the syntactic similarity for the input name or output name is equivalent to the input name or output name with the maximum syntactic similarity with the service requester's input name or service requester's output name. Select to calculate syntax similarity. For example, when the input name entered by the service requester is A and the input name registered by the service provider for one process is B1, B2, B3, the syntax similarity for the input name is SynIN (A, B1), SynIN The syntax similarity for the input name is obtained by selecting the maximum value among the values of (A, B2) and SynIN (A, B3). Therefore, if the service requester has one input name and the number of input names of a single process is m, the syntax similarity between the process name and the input name is shown in Equation 3.

(Formula 3)

In SynIN _j (R.in, S.in), j is 1 to m, which means the syntax similarity for each input name of the service requester and m input names of a single process provided by the service provider, and MAX (SynIN _j (R.in, S.in)) means that the maximum value among the syntax similarity values for each of the calculated names is selected for m input names.

In addition, when there are multiple service requester input names or multiple service requester output names, the plurality of service requester inputs are selected by selecting the maximum syntactic similarity with respect to each service requester input name or service requester output name. Calculate syntactic similarity by taking an average over the number of names or the number of output names of multiple service requesters. 1 illustrates a method of calculating similarity for two input names. As shown in Fig. 1, the input names required by the service requester are A1 and A2, and the names given to one process are C1, C2, C3, and arrows are calculated for A1 or A2 and C1, C2 or C3, respectively. Means a measure of syntax similarity to names. The syntactic similarity for the names is the maximum of the values of SynIN (A1, C1), SynIN (A1, C2), and SynIN (A1, C3), and SynIN (A2, C1), SynIN (A2, C2), and SynIN ( The maximum of A2 and C3) is added to be the average value divided by two, the number of input names required. Therefore, when the number of input names requested by the service requester is n and the number of input names of one process is m, the syntax similarity is as shown in Equation 4.

(Formula 4)

의 의미는 상기 수식 3에서 설명한 것과 동일한 의미이다. At this time,

Has the same meaning as described in Equation 3 above.

In the above Equations 3 to 4, a method of calculating the syntactic similarity between the process name and the input name is provided. However, the syntactic similarity between the process name and the output name is also used using the SynOUT () function instead of the SynIN () function. R.out and S.out instead of in and S.in may be calculated similarly to Equations 3 to 4 above.

The process signature matchmaking step includes: (1a) calculating process name similarity; (2a) calculating an average of maximum similarity values for the input name and the output name; (3a) using a weighting method to calculate a total similarity value.

MAX(SynIN_j()))/n 함수를 사용하여 입력/출력 이름에 대한 유사도를 계산하는 단계(s140)이전에 프로세스 이름 유사도를 먼저 계산하여(s120) 프로세스 이름 유사도를 기반으로 특정값 이상이면 유효한 프로세스로 판단하고, 특정값 이하이면 프로세스 시크니처 매치메이킹을 종료하는 판단을 내리고(s130), 유효한 프로세스일 경우 입력/출력 이름에 대한 유사도를 계산(s140)한다. 총 구문 유사도는 식 1내지 식 4를 이용하여 계산된다. One embodiment of a detailed process signature matchmaking algorithm is shown in FIG. 2. When the process name, input / output name, and weight requested by the service requester are received (s110), the process name similarity is first calculated using the SynPN () function (s120). This is the same as using the weights α and β and having the weight β for the process name to have a value greater than α, so that matching for process names is more important than matching for input / output names. SynIN (), MAX (SynIN _j ()), or

Before calculating the similarity for input / output names using the MAX (SynIN _j ())) / n function (s140), calculate the process name similarity first (s120), and if above a certain value based on the process name similarity, If it is determined to be a valid process, and if it is less than or equal to a specific value, a determination is made to terminate process signature matchmaking (S130). Total syntax similarity is calculated using Equations 1-4.

The process structure matchmaking step is to find a parent process that includes a process with a particular process control structure, and use process structure ontology to analyze the process structure and match the parent process with the required process structure. The process control structure includes sequence, separation, separation-combination, selection, random-order, condition or repetition, or a combination thereof.

The process structure matchmaking performs process structure matchmaking by grouping and classifying the process control structure into semantically similar structures. Specifically, the process structure matchmaking step includes a sequential structure matchmaking step that finds a sequential or random-order structure, a parallel structure matchmaking step that finds a separate or separate and combined structure, and a selection structure match that finds a selection or condition structure. It consists of a classified matchmaking step including a making step.

In detail, sequential matchmaking searches for higher-level processes whose lower-level processes are processed sequentially. For example, use it to process a query such as "find book sales process with post-delivery (P1) payment (P2)". This query retrieves that among the book selling processes, subordinate processes of delivery process P1 and payment process P2 are sequentially processed. Therefore, the sequential matchmaking analyzes and searches for processes having a sequential or random-order structure among control structures.

Parallel structure matchmaking searches for higher-level processes whose subprocesses branched from the same process are processed in parallel. For example, it is used to process a query such as "find a book sales process in which delivery (P1) and inventory update (P2) are performed in parallel". Accordingly, the parallel structure matchmaking analyzes and searches for processes having a separated or separated and combined structure among control structures.

Selective Structure Matchmaking searches for processes with a structure that selects and processes only one process from among two or more processes. For example, it is used to process a query such as "find a book sales process that can accept credit card payment (P1) or unpaid bank transfer payment (P2)". Accordingly, the selection structure matchmaking analyzes and searches for processes having a selection or condition structure that provides a selection structure among control structures.

An iteration is a structure in which one or more processes are processed repeatedly. In the process structure matching process of the present invention, an iterative structure is semantically treated as a process performing one operation, and the process structure is analyzed and searched.

The process structure is complicated by using various control structures. Therefore, when performing the service requester's query, it takes a lot of time to pass the matchmaking result after analyzing all the structures of the registered processes. Therefore, in order to provide a fast matchmaking processing speed for the process structure, the service provider has a process structure ontology. Analyze the process structure and save the structure information for matchmaking to a database including sequential structure tables, parallel structure tables, or condition structure tables.

Thus, the sequential matchmaking uses a sequential structure table to search for a process having a sequential or random-order control structure, and the parallel structure matchmaking uses a parallel structure table to separate or separate and combine control structures. Searching for a process to have, and the selection structure matchmaking can increase the time and efficiency of the search by searching for a process having a control structure of the selection or condition using the selection structure table.

The sequential structure table includes information of an upper process, a preceding process, a following process, a preceding process form, or a following process form including a sequential structure, and the parallel structure table includes an upper process, processes 1 and 2 having a parallel structure, Information in the form of process 1 or in the form of process 2, wherein the selection structure table includes information in the form of a higher process, a selectable process, and a selectable process.

One embodiment of a process having a sequential structure is shown in FIG. 3.

When there is a top-level process A 300 having a sequential structure in the form of a complex process as shown in FIG. 3, the process A 300 includes B 310, a complex process, and c 321 and d 322, an atomic process. Is a subprocess, and is executed in the order of B → c → d. B310 is composed of b1 311 and b2 312, which are atomic processes, and are executed in the order b1-> b2.

The sequential table stores sequential structure information for each process. In this table, RootProcess specifies the parent process with a sequential structure. In the sequential structure for the two processes, the preceding process is stored in Process and the subsequent process in NextProcess. Process type and Next Process type define the types of processes, and have a value of 1 for a compound process and 0 for an atomic process. Table 1 stores the sequence structure information of FIG. 3 in a sequence structure table.

(Table 1) Sequential Structure Table

One embodiment of a process having a parallel structure is shown in FIG. 4.

When there is a process A 400 having a parallel structure in the form of a complex process, the A 400 has a parallel structure of B 410 in the form of a complex process and e 420 in the form of an atomic process. In addition, B 410 has a sub process of c 411 and d 412 having an sequential structure in the form of an atomic process. Since B 410 and e 420 have a parallel structure, it can be seen that c 411 and d 412, which are sub-processes of B 410, also have a parallel structure with e 420.

The parallel structure table stores the parallel structure information for each process. Parallel processes are stored in RootProcess, and parallel processes are stored in Process 1 and Process 2. Process 1 type and Process 2 type define the types of processes. For processes with more than one parallel structure, the Root Process can look for the same ones to find the processes running in parallel. Table 2 stores the parallel structure information of FIG. 4 in a parallel structure table.

(Table 2) Parallel Structure Table

One embodiment of a process having a selection structure is shown in FIG. 5.

5 shows a composite process A 500 with a selection structure. This process consists of subprocesses b (510), c (520), and d (530) configured in the form of atomic processes, and selects one of these subprocesses and executes it.

The selection structure table stores selection structure information about the process. In the selection structure table, RootProcess is the parent process that has the selection structure as in other tables. The process you select is stored in Process 1 and Process 2. If there are two or more selection processes, they can be found by performing an AND operation by selecting RootProcesses that satisfy at least two selection conditions. Table 3 stores the selection structure information of FIG. 5 in the selection structure table.

(Table 3) Selection structure table

As shown in FIG. 6, the process structure analysis algorithm includes: (1b) a top-level process search step S210; (1b) a first determining step (S220) for determining whether the process is an atomic process or a complex process; (2b) a second determination step of determining whether the control structure is an existing structure or a new structure if the determination result of the first determination step is an atomic process, and if the determination result of the first determination step is a compound process ( S230); (3b) If the determination result of the second determination step (S230) is an existing structure, the structure information is stored in the structure information table by searching whether the structure is a sequential structure, a parallel structure, or a selection structure (S250). If the determination result is a new structure, simultaneously storing new structure information in the structure information table and the new structure information table (S240); (4b) a third determining step (S260) of determining whether the process having the stored structure information of step (3b) is an atomic process or a composite process; (5b) If the determination result of the third determination step is an atomic process, the search is terminated (S270). If the determination result of the third determination step is a complex process, the second determination step S230 of step (2b) is performed. It is carried out including the step.

As shown in FIG. 6, the process structure analysis algorithm first reads a TM-S based ontology defined by a service provider for process structure analysis, searches for the top level process in the ontology, and then checks the type of the process. do. If this process has a complex process type, the subprocesses are checked; otherwise, the process structure analysis is terminated. If the top-level process is a composite process, check and save the control structure of the child processes. Process structure information is stored in the three types of structure tables described above. Each table stores all the process structure information that can occur in process structure matchmaking. If a process having a complex process type exists among the stored processes, the control structure of subprocesses for the process is again checked and the structure information is stored in a table. If the control structure of the lower process is different from the control structure of the previously stored upper process, the lower process structure information is extended and stored in the old structure table and the new structure table. This process is repeated until all of the subprocesses are in atomic form.

Process concept ontology is defined by several separate concepts such as 'sell', 'buy', 'pay', etc., and has a classification ontology for each of these concepts as the top level concept. Process concept matchmaking looks for processes that are included in the concept or subconcept selected by the service requester in the process concept ontology.

As described above, process concept matchmaking has the effect of increasing the recall of process discovery by discovering semantically related processes. At this time, the process concept ontology defines the scope of classification by using several viewpoints on concepts such as 'sell', 'buy', and 'pay' and divides them into several sub-concepts according to this perspective. Therefore, if the top concepts of the selected concepts are the same, the operation is applied differently depending on whether the viewpoint is the same or different.

Process concept matchmaking is characterized by the fact that the relationships between concepts are processed through OR operations when the top level concepts of processes required by the service requester are the same and have the same point of view.

Concepts within the same view are defined separately from each other based on that view. Therefore, concepts within the same view can apply OR operations to handle relationships between them. For example, if the user has selected the concept of selling books ("Sell Book") and selling computers ("sell computer") in terms of sales type ("Sell What"), in this example, the book sales concept and the computer sales concept are sales. It is a distinct concept from the point of view of kind Therefore, it can be interpreted in the same way as the query "find processes to sell books or computers".

Process concept matchmaking is characterized by the fact that relationships between concepts are processed through AND operations when the top level concepts of processes required by the service requester are the same and have different points of view.

Concepts within different views are categorized based on different views of the same concept. This can be handled using AND operations between concepts that are in different perspectives. For example, the user selects the concept of "Sell via Electronic Store" from the point of sale ("Sell How"), and selects the concept of "Sell Book" from the point of sale ("Sell What"). If so, this is a different classification of the "Sell" concept in terms of sales method and type of sale. Therefore, it can be interpreted in the same manner as the query "find the process of selling a book in an e-shop".

The process concept matchmaking has a feature in that relationships between the concepts are processed through an OR operation when the top-level concepts of processes requested by the service requester are different.

The different top-level concepts of the selected concepts mean that they are defined in separate concept ontology. It can use OR operations to handle relationships between different concepts. For example, suppose there are separate conceptual ontology for 'buy' and 'sell'. If the user selects the concept of "buy book" in "buy" ontology and the concept of "sell book" in "sell" ontology, it is included in 'buy' and 'sell' ontology respectively. It means to find out which processes are in place. Therefore, it can be interpreted in the same way as the query "find a process that makes it possible to sell or buy a book."

The process concept matchmaking step includes (1c) process concept selection step (s310); (2c) top level concept analysis step (s320) of the selected concepts; (3c) a first determining step (s330) for determining whether the most significant concept of the step (2c) is the same or not the same; (4c) If the determination result of the first determination step is the same, the concept matchmaking is performed when the highest concept is the same (s350). If the determination result of the first determination step is not the same, the concept selected for each top concept is selected. And (S340) searching for processes included in subordinate concepts and OR processing processes included in different top-level concepts (s360).

Concept matchmaking in the case where the top-level concepts of step (4c) are the same may include (1d) analyzing a viewpoint of the selected concepts (s351); (2d) a first determining step (s352) of determining whether the selected concepts have the same viewpoint or different viewpoints; (3d) If the determination result of the first determination step is the same point of view, the processes included in each of the selected concepts and sub-concepts are searched (s353), and OR processes the processes included in each concept (s354), and the first If the determination result of the determination step is a different viewpoint, the process performed by searching for processes included in the selected concept and sub-concept for each viewpoint (s355) and ANDing the processes included in the different viewpoints (s356) Have

FIG. 7 illustrates a processing algorithm for a case where a user selects two or more concepts and the selected top concepts differ from each other. First, the top level concepts of the concepts input from the user are first examined. If the top level concept is different, the selected concept and the processes included in the subordinate concept are searched, and the OR operation is processed for each result and the result is passed. Otherwise, if the top-level concepts are the same, the result is processed using the algorithm of FIG.

8 shows a processing algorithm for the case where the highest concept of the concepts selected by the user are the same. First, examine the perspectives on the concepts chosen by the user. If the points of view are the same, the selected concepts and the processes included in the subordinate concepts are each searched, and the OR operation is processed for each of the found results and the results are passed. If the point of view is different, the AND operation is performed on the searched results.

As described above, in order to solve problems of existing semantic web service discovery and provide intelligent web service discovery, process information is represented by ontology and a matchmaking mechanism using the same is proposed. Process information enables intelligent web service discovery by describing and discovering services in more detail than existing discoveries, and search using process information increases the accuracy and reproducibility of service discovery required for intelligent web service discovery. do. It can also be used for discovery systems for the automation of services, as well as systems for the combination, integration and interoperability of services.

Claims (17)

How to find a web service Use process-based ontology, A process signature matchmaking step of discovering a process by calculating syntactic similarities for the process name, input name, or output name; A process structure matchmaking step of finding an upper process including a process having a specific process control structure; Process concept matchmaking steps to find processes that are included in a particular concept Process-based web service discovery method comprising the. The method of claim 1, The process-based ontology uses a topic map, which is an ontology language, A process signature ontology containing definitions for process names, input names, or output names; A process structure ontology that includes a definition for a process control structure; Process concept ontology, which includes definitions of process concepts, definitions of relationships between concepts, definitions of perspectives on concepts, definitions of classifications between concepts, or definitions of views on classifications. Process-based web service discovery method comprising the. The method of claim 1, The syntactic similarity has a weight, and the weight for the process name is greater than the weight for the input name or the output name. The method of claim 1, When multiple input names or multiple output names exist in one process, The syntactic similarity for input name or output name is a method of discovery of a process-based web service, characterized by selecting an input name or output name having the maximum syntactic similarity with respect to the service requester's input name or service requester's output name. . The method of claim 4, wherein When there are multiple service requester input names or multiple service requester output names, For each service requester's input name or service requester's output name, the maximum syntactic similarity is selected to average the syntactic similarity by averaging over the number of input names of multiple service requesters or the output name of multiple service requesters. A method of finding a process-based web service characterized by calculating. The method of claim 1, Wherein said process control structure comprises sequence, separation, separation and combination, random-order, selection, condition or iteration, or a combination thereof. The method of claim 1, The process structure matchmaking step includes a sequential structure matchmaking step that finds a sequential or random-order structure, a parallel structure matchmaking step that finds a separate or separate and combined structure, and a selection structure matchmaking step that finds a selection or conditional structure. Method of finding a process-based web service, characterized in that it includes. The method of claim 7, wherein The sequential structure matchmaking process finds a specific structure by using a sequential structure table including information of an upper process, a preceding process, a following process, a preceding process form, or a following process form including the sequential structure. How to find the service. The method of claim 7, wherein The parallel structure matchmaking may be performed by using a parallel structure table including information on a higher level process including a parallel structure, a process 1 and a process 2 having a parallel structure, a form of a process 1, or a form of a process 2. Method of finding a process-based web service, characterized in that. The method of claim 7, wherein In the selection structure matchmaking, a method for discovering a process-based web service using a selection structure table including information on a form of a superior process, a selectable process, and a selectable process including the selection structure is selected. . The method of claim 1, The process concept matchmaking is a process-based web service discovery method characterized in that the relationship between the concept is processed through the OR operation when the highest concept of the processes required by the service requester is the same and has the same view. The method of claim 1, The process concept matchmaking is a process-based web service discovery method characterized in that the relationship between the concept is processed by the AND operation when the top level concepts of processes required by the service requester are the same and have different points of view. The method of claim 1, The process concept matchmaking is a process-based web service discovery method characterized in that the relationship between the concept is processed through the OR operation when the top level concepts of the processes requested by the service requester is different. The method of claim 1, The process signature matchmaking step (1a) calculating process name similarity; (2a) calculating an average of maximum similarity values for the input name and the output name; And (3a) calculating a total similarity value using the weights; Method of finding a process-based web service, characterized in that it comprises a. The method of claim 1, In the process structure matchmaking step, Process structure analysis and storage steps (1b) top level process searching; (1b) a first determining step of determining whether the process is an atomic process or a complex process; (2b) if the determination result of the first determination step is an atomic process, the search ends; A second determination step of determining whether a control structure is an existing structure or a new structure when the determination result of the first determination step is a composite process; (3b) If the determination result of the second determination step is an existing structure, the structure information is stored in the structure information table by searching whether it is a sequential structure, a parallel structure or a selection structure, and if the determination result of the second determination step is a new structure. Simultaneously storing new structure information in the structure information table and the new structure information table; (4b) a third determining step of determining whether the process having the stored structure information of step (3b) is an atomic process or a composite process; And (5b) if the determination result of the third determination step is an atomic process, the search ends; Performing a second determination step of step (2b) if the determination result of the third determination step is a composite process; Method of finding a process-based web service, characterized in that it comprises a. The method of claim 1, The process concept matchmaking step (1c) selecting a process concept; (2c) analyzing a top level concept of the selected concepts; And (3c) a first judging step of judging whether or not the most significant concept of step (2c) is the same or not; (4c) if the determination result of the first determination step is the same, concept matchmaking when the highest concept is the same, and ORing the processes included in the different top-level concepts by searching for the processes included in the selected concept and the lower-level concept for each top-level concept if the determination result of the first determination step is not the same; Method of finding a process-based web service, characterized in that it comprises a. The method of claim 16, If the highest concept in step (4c) is the same, concept matchmaking (1d) analyzing the viewpoints of the selected concepts; (2d) a first discriminating step of determining whether the selected concepts have the same viewpoint or different perspectives; And (3d) if the determination result of the first determination step is the same point of view, OR processes the processes included in each concept by searching for processes included in each of the selected concepts and sub-concepts, If the determination result of the first determination step is a different view, searching for processes included in the selected concept and subordinate concepts for each view and ANDing the processes included in the different views; Process-based web service discovery method comprising the. .

Images