A design-time-oriented service monitoring modeling and analysis method

Abstract

The application discloses a kind of design time-oriented service supervision modeling and analysis method, comprising: collecting and analyzing service supervision source document;Extract the regulation of service process design time in source document suitable for business flow and formal modeling and analysis are carried out according to regulation analysis specification, wherein formalization follows pi calculus specification;Build regulation library;According to the input format requirement, input business process BPMN file;According to the result of semantic similarity calculation, determine whether the business process triggers corresponding regulation;If no trigger rule, it is marked as not in the scope of this service supervision, and end, if trigger a rule and more, then next step semantic implication reasoning is carried out;Determine whether the business process exists with regulation contradiction place.The present application can help business analysts and regulatory personnel to quickly analyze the design time of business process analysis and supervision, so as to improve the efficiency of supervision, effectively alleviate the problem of supervision lag.

Description

Technical Field

[0001] This invention relates to the fields of software engineering technology and natural language processing technology, particularly to the fields of service computing in business process management and the application of large-scale pre-trained language models, specifically to a service supervision modeling and analysis method oriented towards design temporality. Background Technology

[0002] Thanks to the maturity of new-generation information technologies such as mobile internet, cloud computing, and big data, the modern service industry has entered a stage of vigorous development and has become an important engine for economic development in various countries. However, at the same time, the modern service industry also suffers from serious lags in governance and regulation, leading to frequent occurrences of some violations.

[0003] Existing research on service regulation modeling and analysis theories largely focuses on public administration and the economic sphere. From a market regulation perspective, service regulation can be categorized into three main types: economic regulation, social regulation, and administrative regulation. Economic regulation refers to "various taxes and subsidies, as well as explicit legislative and administrative control over tax rates, market access, and other aspects of economic activity"; social regulation is characterized by safeguarding public interests and protecting basic human rights; and administrative regulation centers on the supervision of regulators, primarily focusing on their internal operational mechanisms. The development of these three regulatory concepts has given rise to many classic regulatory theories, including interest group theory, asymmetric information game theory, and institutionalist regulatory theory. However, these theories primarily analyze from a qualitative perspective and are insufficient to directly support the need for rapid intelligent regulatory analysis for specific services or scenarios in modern service industries.

[0004] In the field of software engineering, existing research on service regulatory modeling and analysis techniques primarily focuses on business process consistency verification. Business process consistency verification compares a business process instance with a given business process model to identify deviations between the actual behavior of the business process instance and its modeled behavior. However, this is mainly a runtime-oriented regulatory technique, relying on data collected from the information technology systems upon which service delivery depends to identify and analyze inconsistencies. Research on service design-oriented regulatory techniques is relatively limited, mainly focusing on formal verification to achieve requirement consistency and logical consistency verification. Due to the high technical barriers and application costs of formal verification, it is difficult to popularize in specific service application scenarios. Furthermore, the rapid evolution of services makes it difficult for formal verification to keep pace with service iteration, rendering it unsuitable for addressing the regulatory lag problem in modern service industries.

[0005] As modern service industries gradually transform into new business models operating in a network environment, service implementation and delivery methods are shifting from offline to innovative online-offline hybrid models. Traditional offline regulatory models are inefficient and costly, making them ill-suited to the regulatory needs of new online digital service industries. Therefore, exploring intelligent and efficient service regulation theories and models, and researching related key technologies, is of great significance for ensuring the standardized, healthy, and efficient development of modern service industries. Summary of the Invention

[0006] To address the shortcomings of existing regulations in this field, such as regulatory lag and low efficiency, this invention provides a service regulatory modeling and analysis method oriented towards the design phase. This method can help business analysts or regulators quickly identify and locate violations in business process design. This invention enables business analysts and regulators to rapidly conduct regulatory analysis on business processes at the design phase, thereby improving regulatory efficiency and effectively mitigating the problem of regulatory lag.

[0007] A service supervision modeling and analysis method oriented towards design time includes the following steps:

[0008] (1) Collect and analyze service supervision source documents for business process design;

[0009] (2) Extract the regulatory rules applicable to the business process design time from the service regulatory source document in natural language text, and perform formal modeling and analysis according to the analysis specifications in the regulatory rules to obtain the natural language text expression and formal formula corresponding to the regulatory rules.

[0010] (3) Based on the natural language text and formal formulas corresponding to the regulatory rules obtained in step 2, according to the regulatory rule storage specifications, assign values ​​to the ID, scope of application of the regulatory rule, trigger words of the regulatory rule, subject of application of the regulatory rule, text content of the regulatory rule, and formal formula of the regulatory rule for each rule, so as to obtain a regulatory rule library that conforms to the JSON specification;

[0011] (4) Input a business process BPMN file that conforms to the XML specification, and extract all task names in the BPMN file;

[0012] (5) Based on all the task names obtained in step 4, calculate the cosine semantic similarity between each task name and all regulatory rule trigger words. If the calculated cosine semantic similarity is greater than the predefined threshold, then determine that the business process triggers the corresponding regulatory rule and proceed to step 6; if all cosine semantic similarities are less than the predefined threshold, then mark them as not being within the scope of this service regulation.

[0013] (6) Based on the business process and corresponding regulatory rules obtained in step 5, generate natural language text describing the business process using a sentence generation algorithm for all task names included in the business process. Use the natural language text describing the business process as the hypothetical input to the semantic implications reasoning algorithm, and use the content of the regulatory rules as the premise input to the semantic implications reasoning algorithm. The output is either an implicit or non-implicit label. If the output is implicit, it indicates that the business process is compliant; if it is non-implicit, it indicates a non-compliant process.

[0014] (7) Based on the violation process and corresponding regulatory rules obtained in step 6, firstly match the service participants in the violation process according to the subject of application of the regulatory rules in the regulatory rules, and then calculate the cosine semantic similarity between the regulatory rule trigger word and the names of all tasks executed by the service participant. The task with the highest cosine semantic similarity is output as a potentially high violation risk event.

[0015] In step (1), the service supervision source documents for the business process design temporality include relevant laws and standards.

[0016] In step (2), formal modeling and analysis are performed according to the analytical specifications in the regulatory rules, specifically including:

[0017] 2.1) The regulatory rule will be selected from the following rules:

[0018] Data object requirement rules, data object value rules, special task supervision rules, prohibited task supervision rules, critical task supervision rules, and task execution supervision rules.

[0019] 2.2) Following the rules corresponding to step 2.1), formal modeling and analysis are performed in accordance with the π calculus syntax specification.

[0020] The regulatory rule storage specification in step (3) follows the JSON data format. Each rule storage must include the following field names: rule ID, rule scope, rule type, rule trigger word, rule subject, rule text content, and rule formal formula.

[0021] Compared with the prior art, the present invention has the following advantages:

[0022] 1) A regulatory rule analysis and storage specification was constructed, which decouples regulatory rule modeling from business processes. This allows only the changed parts to be modified when business or regulatory policies iterate and evolve, thereby improving regulatory efficiency and effectively alleviating the current problem of regulatory lag.

[0023] 2) The service supervision modeling and analysis method proposed in this invention provides business analysts or supervisors with the ability to locate intelligent violation events, thereby further improving the efficiency of violation analysis;

[0024] 3) The service supervision modeling and analysis method oriented towards design temporality proposed in this invention is a general service supervision method oriented towards design temporality, and has a wide range of applications. Attached Figure Description

[0025] Figure 1 Flowchart for service regulation modeling and analysis method oriented towards design time;

[0026] Figure 2 This is a schematic diagram of the BPMN service process used in the example.

[0027] Figure 3 This is an example diagram of the regulatory rule base built during the implementation process;

[0028] Figure 4 This is a map showing the location of violations during the implementation process. Detailed Implementation

[0029] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Operating methods not specifically specified in the following embodiments are generally performed under conventional conditions or as recommended by the manufacturer.

[0030] The flowchart of the service supervision modeling and analysis method oriented to design time of this invention is as follows: Figure 1 As shown, the specific steps include:

[0031] (1) Collect and analyze service supervision source documents for business process design;

[0032] (2) Extract the regulatory rules applicable to the business process design time from the service regulatory source document in natural language text, and perform formal modeling and analysis according to the analysis specifications in the regulatory rules to obtain the natural language text expression and formal formula corresponding to the regulatory rules.

[0033] (3) Based on the natural language text and formal formulas corresponding to the regulatory rules obtained in step 2, according to the regulatory rule storage specifications, assign values ​​to the ID, scope of application of the regulatory rule, trigger words of the regulatory rule, subject of application of the regulatory rule, text content of the regulatory rule, and formal formula of the regulatory rule for each rule, so as to obtain a regulatory rule library that conforms to the JSON specification;

[0034] (4) Input a business process BPMN file that conforms to the XML specification, and extract all task names in the BPMN file;

[0035] (5) Based on all the task names obtained in step 4, calculate the cosine semantic similarity between each task name and all regulatory rule trigger words. If the calculated cosine semantic similarity is greater than the predefined threshold, then determine that the business process triggers the corresponding regulatory rule and proceed to step 6; if all cosine semantic similarities are less than the predefined threshold, then mark them as not being within the scope of this service regulation.

[0036] (6) Based on the business process and corresponding regulatory rules obtained in step 5, generate natural language text describing the business process using a sentence generation algorithm for all task names included in the business process. Use the natural language text describing the business process as the hypothetical input to the semantic implications reasoning algorithm, and use the content of the regulatory rules as the premise input to the semantic implications reasoning algorithm. The output is either an implicit or non-implicit label. If the output is implicit, it indicates that the business process is compliant; if it is non-implicit, it indicates a non-compliant process.

[0037] (7) Based on the violation process and corresponding regulatory rules obtained in step 6, firstly match the service participants in the violation process according to the subject of application of the regulatory rules in the regulatory rules, and then calculate the cosine semantic similarity between the regulatory rule trigger word and the names of all tasks executed by the service participant. The task with the highest cosine semantic similarity is output as a potentially high violation risk event.

[0038] The regulatory rule analysis specification mentioned in step (2) is as follows: Regulatory rules are mainly divided into six categories, including data object requirement rules, data object value rules, special task regulatory rules, prohibited task regulatory rules, key task regulatory rules, and task execution regulatory rules. Data object requirement rules require that the business process must include all data objects required by the regulatory rules; data object value rules require that the values ​​of specified data objects appearing in the business process must meet specified thresholds; prohibited task regulatory rules stipulate that prohibited tasks cannot appear in the business process; key task regulatory rules stipulate that the business process must include specified tasks; task execution regulatory rules regulate the order of task execution; and special task regulatory rules are supplementary rules for special scenarios. The modeling specification mainly conforms to the π calculus syntax requirements, specifically:

[0039] 1) Data object requirement rules:

[0040] d required =(d1,d2,...,d n );

[0041] f(m).(if m==d required then P c else P v )

[0042] Where drequired It is the required collection of data objects, d1, d2, ..., d n These are the corresponding data objects.

[0043] "=" is the assignment operator. The "=" sign determines whether the left and right sides are equal. f(m) refers to the input information received, where m is the specific information content, and P... c and P v Then it executes the corresponding actions of marking compliance and marking violations. "if then" refers to the judgment logic.

[0044] 2) Data object value retrieval rules:

[0045] d standard = (m1, m2);

[0046] f(m).(if m∈d standard then P c else P v )

[0047] Where d standard This refers to the required threshold values ​​for the data object, where m1 and m2 are the corresponding thresholds, f(m) refers to the received input information, and m is the specific information content. "if m∈d..." standard "P" refers to determining whether the specific information content received is within the specified threshold range. c and P v This involves executing the corresponding actions for marking compliance and violation. Based on the "if then" conditional logic, if the input information meets the threshold requirements, then P is enabled. c The system is deemed compliant and does not meet the requirements for enabling P. v Then it is determined to be a violation.

[0048] 3) Special Task Supervision Rules:

[0049] m specific =(m task1 ,m task2 ,...,m taskn );

[0050] f(m).(if m∈m violation and m∈m specific then P c )

[0051] Where m specific It is a collection of tasks in a specific scenario, m task1 ,m task2 ,...,m taskn That is the specific task, m violationIt is the set of violation tasks, where f(m) refers to the received input information, and m is the specific information content. P c and P v Then, the corresponding actions of marking compliance and violation will be executed. If the input information m belongs to a violation task but also belongs to a task set under a special scenario, then the input is judged to be compliant.

[0052] 4) Rules for supervising prohibited tasks:

[0053] m violation =(m task1 ,m task2 ,...,m taskn );

[0054] f(m).(if m∈m violation then P v else P c )

[0055] m violation It is a collection of illegal tasks, m task1 ,m task2 ,...,m taskn The specific task is f(m), where f(m) refers to receiving input information, and m is the specific information content. P c and P v This involves executing the corresponding actions to mark compliance and violation. This rule states that if the input information is in the set of violation tasks, the process is deemed a violation.

[0056] 5) Key Task Oversight Rules:

[0057] m required =(m task1 ,m task2 ,...,m taskn );

[0058]

[0059] Where m required It is a set of tasks that must be executed, m task1 ,m task2 ,...,m taskn The specific task is f(m), where f(m) refers to receiving input information and m is the specific information content. It means it does not belong to. (P) c and P v This involves executing the corresponding actions to mark processes as compliant or non-compliant. This type of rule defines a process as non-compliant if the input information does not fully cover the set of critical tasks.

[0060] 6) Task execution supervision rules:

[0061] m finish =(m task1 ,m task2 ,...,m taskn );

[0062] f(m).((if m∈m finished then Task).P c then P v )

[0063] Where m finish It refers to a set of tasks that need to be completed in advance, m task1 ,m task2 ,...,m taskn The specific task is f(m), where f(m) refers to receiving input information, m is the specific information content, and P... c and P v This involves executing the corresponding actions to mark compliance and violation, where "∈" indicates belonging to. For a specific task, if its execution before m... finish If all steps have been completed, then it is considered a compliant process; otherwise, it is considered a non-compliant process.

[0064] The rule storage specification described in step (3) is in JSON data format. Each rule needs to be stored with the following key-value pairs: rule ID, rule scope, rule type, rule trigger word, rule subject, rule text content, and rule formal formula.

[0065] The semantic similarity cosine(v,w) calculation in step (5) adopts the cosine similarity calculation method, and the specific calculation formula is as follows:

[0066]

[0067] Where v and w are the word vectors of the trigger word and the task in the business process, N is the word vector dimension, i refers to the i-th dimension, and |v| and |w| are the word vector lengths, respectively.

[0068] The semantic implication reasoning in step (7) specifically involves: generating corresponding sentences based on business process elements as hypotheses for semantic implication reasoning, taking regulatory rules as the premise for semantic implication reasoning, and using a large-scale pre-trained language model to determine whether it is an implication relationship or a contradiction relationship.

[0069] like Figure 1 As shown, the service supervision modeling and analysis method oriented towards design time in this embodiment includes:

[0070] (1) Collect and analyze service regulatory source documents, including relevant laws, regulations and industry standards and other normative documents;

[0071] (2) Extract regulatory rules applicable to the business process design phase from the source documents, and perform formal modeling and analysis according to the regulatory rule analysis specifications;

[0072] (3) Construct a regulatory rule base, and store the rules obtained from the analysis and modeling in step 2 according to the storage specifications to construct the corresponding regulatory rule base;

[0073] (4) Enter the business process BPMN file according to the input format requirements;

[0074] (5) Determine whether the business process triggers the corresponding regulatory rules based on the semantic similarity calculation results;

[0075] (6) If no triggering rule is triggered, mark it as not within the scope of this service supervision and end. If one or more supervision rules are triggered, proceed to the next semantic implication reasoning.

[0076] (7) For processes that trigger regulatory rules, perform semantic implication reasoning, that is, determine whether the business process contradicts the regulatory rules. If so, mark it as a non-compliant process; otherwise, mark it as a compliant process and end it.

[0077] (8) Locate the violation incidents. Based on the regulatory rules triggered by the violation process, locate the corresponding violation objects in the business process files, including violation tasks and data objects.

[0078] The following example, using the business process of internet-based medical and health services, illustrates the technical solution of this invention. The specific implementation process is as follows:

[0079] (1) Collect relevant management regulations and documents for Internet medical and health services.

[0080] (2) Seven regulatory rules applicable to design-time business processes were extracted from the regulatory documents collected in step 1. The following is an example of regulatory rule modeling:

[0081] Qua medicalStaff =(PQC,PPC);

[0082] f(qua doctor ).(if qua doctor ==Qua medicalStaff then P c else P v )

[0083] Qua medicalStaff This refers to the collection of materials required for the qualification certification of medical personnel, including two specific qualification certification documents: PQC and PPC. doctorThese are the certification materials submitted by doctors during the actual qualification certification process. This rule refers to the requirement that if a doctor submits materials during the qualification certification process... doctor and the required Qua medicalStaff If the agreement is consistent, then it is considered compliant; otherwise, it is considered non-compliant.

[0084] (3) Construct a regulatory rule base. The regulatory rule base is stored in JSON format. Each rule needs to contain seven items: rule id, category, scope, subject, trigger words, formal style, and content, corresponding to the seven field names. See the example of the regulatory rule base in this embodiment. Figure 3 .

[0085] (4) Input the business process BPMN file according to the input format requirements. In this embodiment, the BPMN file storage format is required to be XML. The input process in this embodiment is shown in [link to relevant documentation]. Figure 2 This process describes online diagnosis and treatment services in internet-based healthcare services based on the BPMN standard.

[0086] (5) Determine whether the business process triggers the corresponding regulatory rule based on the semantic similarity calculation result. In this embodiment, the semantic similarity threshold is set at 0.9. If the semantic similarity between the task name in the process and the trigger word of the regulatory rule exceeds 0.9, then the process is determined to trigger this regulatory rule. Through calculation, the input online diagnosis and treatment business process and Figure 3 The regulatory rule with ID 7 has a semantic similarity of more than 0.9, therefore the online diagnosis and treatment business process will be bound to this regulatory rule.

[0087] (6) Determine whether the process triggers regulatory rules. In this embodiment, the online diagnosis and treatment business process is bound to the regulatory rule with ID 7, so proceed to the next step of semantic implication reasoning; if no corresponding rule is triggered, end directly.

[0088] (7) Semantic Implication Reasoning: All task names in the process are extracted, and natural language sentences are generated using a sentence generation algorithm as the hypothetical input for semantic implication reasoning. The content of the regulatory rule judgment serves as the premise input for semantic implication reasoning. The semantic implication reasoning algorithm uses a pre-trained large language model. In this embodiment, the model output is a contradiction label. Since this regulatory rule determines compliance if the requirements are met, a contradiction indicates that the business process does not meet the regulatory rule requirements, and therefore is judged as a violation.

[0089] (8) Violation Event Location: Since the regulatory rules clearly define the regulatory body and the trigger words include the regulatory event, the specific event in the business process can be located based on the subject and trigger words. In this embodiment, the regulatory body is the internet healthcare service provider, and the trigger event is online diagnosis and treatment; therefore, the violation event can be quickly located in the process file. The violation event location result in this embodiment is as follows: Figure 4 As shown.

[0090] Finally, it should be noted that the above examples are only used to illustrate and not limit the technical solutions of the present invention. Although the present invention has been described in detail with reference to the above examples, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention without departing from the spirit and scope of the present invention. Any modifications or partial substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A service supervision modeling and analysis method oriented towards design time, characterized in that, Includes the following steps: (1) Collect and analyze service supervision source documents for business process design; (2) Extract the regulatory rules applicable to the business process design time from the service regulatory source document in natural language text, and perform formal modeling and analysis according to the analysis specifications in the regulatory rules to obtain the natural language text expression and formal formula corresponding to the regulatory rules. (3) Based on the natural language text expression and formal formula corresponding to the regulatory rules obtained in step 2, according to the regulatory rule storage specification, assign values ​​to the ID, scope of application, trigger words, subject of application, text content, and formal formula of each regulatory rule to obtain a regulatory rule library that conforms to the JSON specification. (4) Input a business process BPMN file that conforms to the XML specification, and extract all task names in the business process BPMN file; (5) Based on all the task names obtained in step 4, calculate the cosine semantic similarity between each task name and all regulatory rule trigger words after being assigned in step (3). If the calculated cosine semantic similarity is greater than the predefined threshold, then determine that the business process triggers the corresponding regulatory rule and proceed to step (6). If all cosine semantic similarities are less than or equal to a predefined threshold, then the service is marked as not being within the scope of this service supervision. (6) For the regulatory rules triggered by the business process in step (5), generate natural language text describing the business process by using a sentence generation algorithm for all task names included in the business process. Use the natural language text describing the business process as the hypothetical input of the semantic implication reasoning algorithm. Use the regulatory rule text content after the assignment in step (3) as the premise input of the semantic implication reasoning algorithm. The output is two types of labels: implication or non-implication. If the output is implication, it means that the business process is a compliant process. If it is non-implication, it means that the business process is a non-compliant process. Proceed to step (7). (7) Based on the violation process and the regulatory rules corresponding to the violation process obtained in step (7), firstly match the service participants corresponding to the violation process according to the subject of application of the regulatory rules in the regulatory rules. Then, calculate the cosine semantic similarity between the regulatory rule trigger words assigned in step (3) and all task names executed by the service participants. The task with the highest cosine semantic similarity is the potential high violation risk event.

2. The service supervision modeling and analysis method oriented towards design temporality as described in claim 1, characterized in that, In step (1), the service supervision source documents for the business process design temporality include relevant laws and standards.

3. The service supervision modeling and analysis method oriented towards design temporality as described in claim 1, characterized in that, In step (2), formal modeling and analysis are performed according to the analytical specifications in the regulatory rules, specifically including: 2.1) The regulatory rule will be selected from the following rules: Data object requirement rules, data object value rules, special task supervision rules, prohibited task supervision rules, critical task supervision rules, and task execution supervision rules; 2.2) Following the rules corresponding to step 2.1), formal modeling and analysis are performed in accordance with the π calculus syntax specification.

4. The service supervision modeling and analysis method oriented towards design temporality as described in claim 1, characterized in that, In step (3), after the regulatory rules and corresponding content are assigned values, the data is stored in JSON format.

Images