A metrology system accreditation management system and method

Abstract

The application provides a metrology system certification management system and method, which constructs a knowledge graph of metrology system certification audit through to-be-audited certification data and associated standard documents; determines constraint conditions for sampling audit of the to-be-audited certification data according to semantic relationship edges in the knowledge graph; determines audit risks of different data categories in an initial audit sample of metrology system certification of a target enterprise; determines risk weights of each data category in the knowledge graph in the to-be-audited certification data according to the association relationship between each data review element in the knowledge graph and the associated standard documents; determines sample utility values of each data category in the initial audit sample during the sampling audit based on all risk weights and all audit risks; and dynamically updates the initial audit sample according to all sample utility values. The scheme can guide the sampling audit direction based on the semantic relationship and risk difference between the certification data to improve the coverage accuracy of the sampling audit of the metrology system certification data.

Description

Technical Field

[0001] This application relates to the field of certification management technology, and more specifically, to a metrology system certification management system and method. Background Technology

[0002] Metrological system certification management refers to the process by which a qualified third-party certification body reviews and evaluates the metrological management system documentation submitted by a company, based on officially published metrological system standards, and issues a certification conclusion. Metrological system certification aims to ensure that a company's metrological management work complies with relevant standards, thereby improving the accuracy and consistency of metrology in product testing, production control, and equipment management, enhancing product quality assurance capabilities, and promoting continuous improvement of the company's management system.

[0003] In practice, enterprises need to submit various types of metrology system certification materials, including the configuration of measuring instruments, traceability records, periodic verification plans, and management systems. Certification bodies typically review and conduct compliance assessments of these materials according to standard requirements. During the metrology system certification management process, due to the large number of documents, records, and business processes involved in an enterprise's metrology system, a comprehensive audit is often impractical in terms of time and resources. Sampling audits can provide reasonable assurance of the overall operation of the metrology system within limited time and resources. A sampling audit refers to selecting a portion of the documents, records, processes, activities, or departments submitted by the enterprise within the audit scope for review to infer the overall conformity and effectiveness of the metrology system. Sampling audits of metrology system certification materials are a common and important method in metrology system certification audits. While sampling audits play a crucial role in improving audit efficiency and focusing on key issues, they also require auditors to fully consider the representativeness and validity of the samples during the sampling audit process to ensure the accuracy and reliability of the audit results. However, in existing technologies, sampling audits are inherently risky (the risk of sampling is that the sample drawn from the population may not be representative), which can easily lead to the omission of high-risk data or excessive attention to low-risk data in the certification materials submitted by enterprises. This can result in deviations in the auditors' metrological system certification conclusions, with significant differences from the results of a comprehensive audit of all certification materials. Consequently, this affects the coverage and accuracy of the metrological system certification audit results. Therefore, how to guide the sampling audit direction based on the semantic relationships and risk differences between certification materials to improve the coverage accuracy of sampling audits of metrological system certification materials has become a challenge for the industry. Summary of the Invention

[0004] This application provides a metrology system certification management system and method, which can guide the sampling and auditing direction based on the semantic relationships and risk differences between certification materials to improve the coverage accuracy when sampling and auditing metrology system certification materials.

[0005] Firstly, this application provides a knowledge graph-based method for sampling and reviewing certification materials submitted by enterprises, which is used by the metrology system certification management system to sample and review the metrology system certification materials submitted by enterprises. The method includes the following steps:

[0006] Obtain the relevant standard documents for the target company's pending certification materials;

[0007] A knowledge graph for metrological system certification audit is constructed based on the certification terms in the certification materials to be audited and the associated standard documents.

[0008] Based on the semantic relationship edges in the knowledge graph and the category distribution characteristics of the certification materials to be audited, the constraints for sampling and auditing the certification materials to be audited are determined. Based on the constraints, the initial audit sample for the metrological system certification of the target enterprise is extracted from the certification materials to be audited, and then the audit risk of different categories of materials in the initial audit sample is determined.

[0009] The risk weight of each document category in the knowledge graph is determined by the relationship between the graph nodes in the knowledge graph and the document review elements in the associated standard documents.

[0010] Based on the non-uniform distribution characteristics of all risk weights and all audit risks, determine the sample utility value of each data category in the initial audit sample during the sampling audit;

[0011] The initial audit sample is dynamically updated based on all sample utility values ​​when sampling and auditing the certification materials to be audited, and the final audit sample list of the target company is output.

[0012] In some embodiments, constructing a knowledge graph for metrological system certification auditing based on the certification documents to be audited and the certification clauses in the associated standard documents specifically includes:

[0013] Extract the certification data elements of the target company from the certification documents to be audited;

[0014] Extract audit clause elements based on the certification clause content in the associated standard document;

[0015] A knowledge graph for metrological system certification audit is constructed based on the semantic relationships between the certification data elements and the audit clause elements.

[0016] In some embodiments, determining the constraints for sampling and reviewing the documents to be reviewed based on the semantic relationship edges in the knowledge graph and the category distribution characteristics of the documents to be reviewed specifically includes:

[0017] Determine the category distribution characteristics of the documents to be audited and certified;

[0018] The semantic core degree of each category of materials in the knowledge graph to be reviewed and certified is determined based on the semantic relation edges in the knowledge graph.

[0019] The constraints for sampling and reviewing the certification materials to be reviewed are determined based on all semantic core degrees and the category distribution characteristics.

[0020] In some embodiments, extracting the initial audit sample for the target enterprise's metrological system certification from the certification documents to be audited based on the constraints specifically includes:

[0021] The sampling quantity for each document category in the documents to be audited and certified is allocated based on the aforementioned constraints;

[0022] Based on the sampling quantity, the initial audit sample for the target company's metrological system certification is selected from the certification documents to be audited.

[0023] In some embodiments, determining the audit risk of different document categories in the initial audit sample specifically includes:

[0024] Determine the risk assessment indicators for each document category in the initial audit sample;

[0025] Based on the risk assessment indicators, each document category in the initial audit sample is evaluated separately to obtain a multi-dimensional score vector for each document category;

[0026] The review risk of different document categories in the initial review sample is determined by various multidimensional scoring vectors.

[0027] In some embodiments, determining the risk weight of each document category in the knowledge graph of the document to be reviewed and certified based on the association between the graph nodes in the knowledge graph and the document review elements in the associated standard documents specifically includes:

[0028] The semantic distribution features of each data category in the data to be reviewed and certified are determined by the graph nodes in the knowledge graph.

[0029] The risk weight of each document category in the knowledge graph is determined based on the semantic distribution characteristics and the correlation between the document review elements in the associated standard documents.

[0030] In some embodiments, determining the sample utility value of each data category in the initial audit sample during sampling audit, based on the non-uniform distribution characteristics of all risk weights and all audit risks, specifically includes:

[0031] Determine the non-uniform distribution characteristics of all risk weights;

[0032] Based on the non-uniform distribution characteristics and all audit risks, the audit value of each data category in the initial audit sample is evaluated to obtain the sample utility value of each data category in the initial audit sample during the sampling audit.

[0033] Secondly, this application provides a metrology system certification management system, including a sampling and review unit. The sampling and review unit is used to sample and review the metrology system certification materials submitted by enterprises. The sampling and review unit includes:

[0034] The acquisition module is used to acquire the associated standard documents of the target company's certification materials to be audited;

[0035] The processing module is used to construct a knowledge graph for metrological system certification audit based on the certification materials to be audited and the certification clauses in the associated standard documents;

[0036] The processing module is further configured to determine the constraints for sampling and auditing the certification materials based on the semantic relationship edges in the knowledge graph and the category distribution characteristics of the certification materials to be audited, extract the initial audit sample for the metrological system certification of the target enterprise from the certification materials to be audited based on the constraints, and then determine the audit risk of different categories of materials in the initial audit sample.

[0037] The processing module is also used to determine the risk weight of each data category in the knowledge graph in the knowledge graph based on the relationship between the graph nodes in the knowledge graph and the data review elements in the associated standard documents;

[0038] The processing module is also used to determine the sample utility value of each data category in the initial audit sample during the sampling audit based on the non-uniform distribution characteristics of all risk weights and all audit risks;

[0039] The execution module is used to dynamically update the initial audit samples when sampling and auditing the certification materials to be audited based on all sample utility values, and output the final audit sample list of the target enterprise.

[0040] Thirdly, this application provides a computer device, which includes a memory and a processor. The memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the computer device performs the above-described knowledge graph-based authentication data sampling and verification method.

[0041] Fourthly, this application provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the aforementioned knowledge graph-based authentication data sampling and verification method.

[0042] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0043] In this application, the process involves: acquiring the associated standard documents of the target company's certification materials to be audited; constructing a knowledge graph for metrological system certification audit based on the certification clauses in the certification materials to be audited and the associated standard documents; determining the constraints for sampling audit of the certification materials to be audited based on the semantic relationship edges in the knowledge graph and the category distribution characteristics of the certification materials to be audited; extracting initial audit samples for the target company's metrological system certification from the certification materials to be audited based on the constraints; determining the audit risk of different material categories in the initial audit samples; determining the risk weight of each material category in the certification materials to be audited in the knowledge graph based on the relationship between the graph nodes in the knowledge graph and the material review elements in the associated standard documents; determining the sample utility value of each material category in the initial audit samples during sampling audit based on the non-uniform distribution characteristics of all risk weights and all audit risks; dynamically updating the initial audit samples during sampling audit of the certification materials to be audited based on all sample utility values; and outputting the final audit sample list of the target company.

[0044] Therefore, in this application, firstly, determining the audit risk of different data categories in the initial audit sample ensures that the audit focus is on data categories with high potential risks, avoiding over-examination of low-risk areas, thereby maximizing the sampling audit effect; secondly, determining the risk weight of each data category in the knowledge graph in the knowledge graph based on the correlation between the graph nodes and the data review elements in the associated standard documents accurately identifies and quantifies the risk uncertainty of different data categories, revealing which data categories are more complex and potentially more difficult to audit, thus giving them higher audit attention; then, determining the sample utility value of each data category in the initial audit sample during sampling audit based on the non-uniform distribution characteristics of all risk weights and all audit risks allows for understanding the representativeness of the data categories in the sampling audit, facilitating subsequent dynamic updates. This approach effectively reduces the amount of review resources consumed by low-risk metrology certification data during sample audits, ensuring more accurate coverage and improving the effectiveness of the final sampling audit. Finally, by dynamically updating the initial audit samples based on the utility values ​​of all samples, the audit coverage of high-risk categories can be strengthened in a timely manner. This ensures continuous optimization of the sampling audit throughout the metrology certification audit process. This dynamic update mechanism makes the sampling audit not merely a static initial allocation, but rather an ongoing adaptation to the actual situation during the audit, ensuring more comprehensive and accurate coverage of potentially problematic certification data and avoiding the omission of key metrology certification data, thereby improving the accuracy and effectiveness of the metrology certification audit. In summary, this solution can guide the sampling audit direction based on the semantic relationships and risk differences between certification data to improve the coverage accuracy of the sampling audit of metrology certification data. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is an exemplary flowchart of a knowledge graph-based authentication data sampling and verification method according to some embodiments of this application;

[0047] Figure 2 This is an exemplary flowchart of constructing a knowledge graph according to some embodiments of this application;

[0048] Figure 3 This is an exemplary flowchart illustrating dynamic updates according to some embodiments of this application;

[0049] Figure 4 This is a schematic diagram of the structure of a sampling review unit according to some embodiments of this application;

[0050] Figure 5 This is a schematic diagram of the structure of a computer device implementing a knowledge graph-based authentication data sampling and verification method according to some embodiments of this application. Detailed Implementation

[0051] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0052] refer to Figure 1 The figure is an exemplary flowchart of a knowledge graph-based authentication data sampling and verification method according to some embodiments of this application. The knowledge graph-based authentication data sampling and verification method 100 mainly includes the following steps:

[0053] In step 101, obtain the associated standard documents of the target company's certification materials to be audited.

[0054] In specific implementation, obtaining the associated standard documents of the target company's certification materials to be audited can be achieved in the following way: receiving the metadata of the certification materials submitted by the target company (such as certification material name, number, submission category, etc.) through the interface of the metrology system certification management system, and obtaining the certification materials of the target company to be audited. The certification materials to be audited include, but are not limited to, standard implementation status descriptions, test reports, management records, process flows, etc. Then, calling the standard clause indexing system of metrology system certification, using keyword matching and semantic matching algorithms (such as TF-IDF similarity or BERT embedding) to search the standard document library for the standard certification clause content (such as the specific clause content in ISO 9001) corresponding to the certification materials to be audited, and outputting the search results as the associated standard documents of the target company's certification materials to be audited, so as to provide standard basis for subsequent knowledge graph construction and semantic audit. Other methods can also be used in other embodiments, which are not specifically limited here.

[0055] It should be noted that the certification documents to be audited in this application refer to various management documents submitted by the target company for the assessment of the metrology system; the related standard documents in this application refer to the review clauses in national or industry standards in the field of quality management that are related to the content of the metrology system certification documents.

[0056] In step 102, a knowledge graph for metrological system certification audit is constructed based on the certification materials to be audited and the certification clauses in the associated standard documents.

[0057] In some embodiments, reference Figure 2 As shown, this diagram is an exemplary flowchart of constructing a knowledge graph in some embodiments of this application. In this embodiment, the knowledge graph for metrological system certification auditing based on the certification materials to be audited and the certification clauses in the associated standard documents can be constructed using the following steps:

[0058] First, in step 1021, the certification data elements of the target enterprise are extracted from the certification data to be audited;

[0059] Secondly, in step 1022, the audit clause elements are extracted based on the certification clause content in the associated standard document;

[0060] Finally, in step 1023, a knowledge graph for metrological system certification audit is constructed based on the semantic relationship between the certification data elements and the audit clause elements.

[0061] In specific implementation, extracting the certification data elements of the target enterprise from the certification data to be audited can be achieved in the following ways: Entity recognition methods in natural language processing, such as BiLSTM-CRF or BERT named entity recognition models, can be used to process the text content of the certification data to be audited, extracting information units with audit significance, such as key elements like "management responsibilities," "continuous improvement plans," and "customer satisfaction data." The set of all key elements is then used as the certification data elements of the target enterprise, transforming unstructured data content into structured semantic entities, providing a data foundation for the establishment of graph nodes. Extracting audit clause elements based on the certification clause content in the associated standard documents can be achieved in the following ways: Clause parsing technology can be used, combined with rule engines and dependency parsing analysis to identify the key review requirements in the certification clause content of the associated standard documents, such as "documented procedures should be established" and "regular management reviews should be conducted." The set of all identified key review requirements is then used as audit clause elements for subsequent graph node construction to reflect the standard audit objectives of the metrology system certification audit. Other methods can also be used in other embodiments, which are not limited here.

[0062] In specific implementation, the knowledge graph for metrological system certification audit can be constructed based on the semantic relationship between the certification data elements and the audit clause elements in the following way: Semantic similarity calculation (such as Word2Vec, S-BERT) can be used to determine whether there are semantic connections (i.e., semantic relationships) such as "satisfy," "prove," or "correspond" between the certification data elements and the audit clause elements. If they exist, the corresponding nodes are connected by relational edges in the graph. Finally, all nodes and relational edges are stored in a graph database (such as Neo4j) to form a queryable audit knowledge graph. This audit knowledge graph is then used as the knowledge graph for metrological system certification audit to support subsequent metrological system certification data sampling and audit path analysis. The relational edges in this knowledge graph are used as semantic relational edges, and the nodes within them are used as graph nodes. Other methods can also be used in other embodiments for determination, which are not limited here.

[0063] It should be noted that, in this application, the certification document element refers to a semantic unit in the certification document to be audited that reflects the operational status of the target enterprise's quality management system. It is used to provide semantic support for the actual implementation of the enterprise's metrology system and to provide a structured data foundation for evidence tracking and standard comparison during the audit activities. The audit clause element in this application refers to a structured content unit in the associated standard document that puts forward specific requirements for the construction and operation of the enterprise's system. Through it, the basis nodes for audit judgment can be clearly identified in the knowledge graph, so that subsequent sampling audits can conduct targeted risk identification and clause coverage around compliance points. The semantic relationship in this application refers to the matching relationship between the certification document element and the audit clause element in logical expression, which can reveal the coverage, correspondence, or missing relationship between the enterprise's data and the standard requirements.

[0064] It should also be noted that the knowledge graph for metrological system certification audit in this application refers to the semantic structure network between the certification materials to be audited by the target enterprise and the associated standard documents. Its function is to provide a mapping structure between standard clauses and actual enterprise data in the metrological system certification audit, which is used to support data sampling, audit path positioning and risk association judgment, so that the audit process of metrological system certification has semantic traceability, structural visibility and decision support capabilities, thereby making the subsequent sampling basis more knowledge-based and context-aware.

[0065] In step 103, constraints for sampling and auditing the certification materials to be audited are determined based on the semantic relationship edges in the knowledge graph and the category distribution characteristics of the certification materials to be audited. Based on the constraints, an initial audit sample for the metrological system certification of the target enterprise is extracted from the certification materials to be audited, and then the audit risk of different categories of materials in the initial audit sample is determined.

[0066] In some embodiments, determining the constraints for sampling and reviewing the data to be reviewed based on the semantic relationship edges in the knowledge graph and the category distribution characteristics of the data to be reviewed and certified can be achieved through the following steps:

[0067] Determine the category distribution characteristics of the documents to be audited and certified;

[0068] The semantic core degree of each category of materials in the knowledge graph to be reviewed and certified is determined based on the semantic relation edges in the knowledge graph.

[0069] The constraints for sampling and reviewing the certification materials to be reviewed are determined based on all semantic core degrees and the category distribution characteristics.

[0070] In practice, determining the category distribution characteristics of the documents to be audited and certified can be achieved in the following way: Semantic analysis of the documents to be audited and certified is performed using natural language processing and text classification algorithms (such as the Fine-tuned BERT classifier), and they are labeled into different document categories according to a standard classification system (such as ISO 9001 document classification), for example, "Management Responsibility," "Resource Management," and "Product Realization." Then, the proportion distribution of each document category in all submitted documents to be audited and certified is statistically analyzed, and this proportion distribution is used as the category distribution characteristics of the documents to be audited and certified, to understand the coverage of each document category in each audit dimension, providing a structural basis for subsequent sampling; based on the semantic relationship edges in the knowledge graph, the position of each document category in the knowledge graph is determined. Semantic core degree can be achieved in the following way: based on the semantic relationship edges in the knowledge graph, the number of connections of each data category in the knowledge graph corresponding to the graph node in the data to be audited and certified is used as the semantic core degree of each data category in the knowledge graph, in order to quantify the closeness between each data category and the audit clause elements. Among them, the data category with a high semantic core degree indicates that its influence in the metrological system certification process is large. Therefore, by determining the semantic core degree, the importance of the data category in the sampling audit can be explored.

[0071] In specific implementation, the constraints for sampling and reviewing the materials to be certified, based on all semantic core degrees and the category distribution characteristics, can be achieved in the following way: Semantic core degree (which reflects the importance of the material category) and category distribution characteristics (which ensure the representativeness of the sample distribution across each material category) can be used as two evaluation dimensions when selecting samples for sampling and reviewing the materials to be certified. A weighted scoring model (e.g., linear weighted or fuzzy comprehensive evaluation) can be constructed, where the weighted scoring model can be constructed by combining expert scoring and historical data verification. The weights of semantic core degree and category distribution characteristics can be determined based on their influence on the sampling and review results. Then, the sampling score for each material category (which can be set to 1-10 points) can be calculated using this weighted scoring model. Finally, based on this... The scoring system sets a minimum sampling ratio for each document category. This minimum sampling ratio is determined based on statistical principles and actual business needs. For example, for document categories with high sampling scores (e.g., greater than 8 points), a higher minimum sampling ratio (e.g., 15%) is set to ensure sufficient review of key document categories. For document categories with low sampling scores (e.g., less than 3 points), a lower minimum sampling ratio (e.g., 5%) is set to avoid wasting review resources. Ultimately, the minimum sampling ratio for each document category is used as a constraint for sampling and reviewing the documents to be certified. This constraint serves as a control standard for subsequent sample selection and allocation, ensuring that the sampling strategy balances semantic focus and structural comprehensiveness, thereby improving the rationality of sampling and the efficiency of review coverage. Other methods can also be used to determine this ratio in other embodiments, and are not limited here.

[0072] It should be noted that the category distribution characteristics in this application represent the proportional structure of the number of different categories of data in the data to be audited and certified; the semantic relationship edge in this application refers to the semantic connection between two graph nodes in the knowledge graph, which is used to reflect the logical relationship between the two graph nodes in the metrological system certification audit scenario; the semantic core degree in this application represents the semantic influence of the data category in the knowledge graph, which can guide the focus on key data categories in the data to be audited and certified during sampling audit; the constraint conditions in this application represent the constraint rules when sampling auditing the data to be audited and certified.

[0073] In some embodiments, the initial audit sample for the metrological system certification of the target enterprise can be extracted from the certification data to be audited based on the constraints using the following steps:

[0074] The sampling quantity for each document category in the documents to be audited and certified is allocated based on the aforementioned constraints;

[0075] Based on the sampling quantity, the initial audit sample for the target company's metrological system certification is selected from the certification documents to be audited.

[0076] In specific implementation, the allocation of the sampling quantity for each data category in the data to be audited and certified based on the constraints can be achieved in the following way: the product of the minimum sampling ratio of each data category in the constraints and the total number of data to be audited and certified can be used as the sampling quantity for each data category in the data to be audited and certified. The initial audit sample for the target enterprise's metrological system certification can be selected from the data to be audited and certified based on each sampling quantity in the data to be audited and certified using the following method: the specific data for the target enterprise's metrological system certification can be selected from each data category in the data to be audited and certified using the stratified sampling method according to the sampling quantity allocated to each data category, and the set of selected specific data can be used as the initial audit sample for the target enterprise's metrological system certification. Other methods can also be used to determine this in other embodiments, and no limitation is made here.

[0077] It should be noted that the sampling quantity in this application refers to the specific number of samples allocated to each category of materials to be audited and certified; the initial audit sample in this application refers to the set of sampled audit and certified materials initially extracted from the materials to be audited and certified.

[0078] In some embodiments, determining the audit risk of different document categories in the initial audit sample can be achieved using the following steps:

[0079] Determine the risk assessment indicators for each document category in the initial audit sample;

[0080] Based on the risk assessment indicators, each document category in the initial audit sample is evaluated separately to obtain a multi-dimensional score vector for each document category;

[0081] The review risk of different document categories in the initial review sample is determined by various multidimensional scoring vectors.

[0082] In specific implementation, the risk assessment indicators for each document category in the initial review sample can be determined in the following way: Risk assessment indicators for each document category in the initial review sample can be constructed based on expert experience (such as "completeness of clause coverage," "accuracy rate of terminology citation," "content consistency," "frequency of historical rectification," "document update cycle," etc.), and cross-validated with reference to existing review record databases to ensure the effectiveness and feasibility of the indicators. This establishes a risk assessment basis with quantifiable characteristics for each document category, facilitating subsequent unified assessment and comparison. The multi-dimensional score vector for each document category in the initial review sample can be obtained by conducting a sub-assessment based on the risk assessment indicators, as follows: Analytic hierarchy process (AHP), fuzzy comprehensive evaluation, or machine learning-based feature scoring models (such as Support Vector Regression (SVR), Random Forest scoring, etc.) can be used to automatically score each document category in the initial review sample under each indicator of the risk assessment indicators, forming a multi-dimensional score vector corresponding to each document category (e.g., document category A is [0.8, 0.6, 0.7, 0.9, 0.5]). This can be achieved through manual review, NLP (Natural Language Processing), etc. Language Processing (LP) quality detection tools or review rule systems automatically assign scores to quantify abstract risk elements, facilitating subsequent risk summarization and comparison. Determining the review risk of different document categories in the initial review sample using various multi-dimensional scoring vectors can be achieved in the following way: First, the weights of the scoring dimensions in each multi-dimensional scoring vector can be set based on expert experience or historical data analysis (e.g., "content accuracy" has a higher weight than "terminology standardization"). Then, aggregation functions (e.g., weighted average, comprehensive scoring, fuzzy normalization) can be used to multiply each score in the multi-dimensional scoring vector of each document category in the initial review sample by its corresponding weight and sum them. The resulting sum is used as the review risk for each document category in the initial review sample, supporting subsequent risk-based dynamic sample optimization. Other methods can also be used in other embodiments, which are not limited here.

[0083] It should be noted that the risk assessment indicators in this application are used to quantify the standardized set of dimensions for evaluating potential audit issues in each type of metrological system certification data. Their role is to provide basic elements for measuring the audit quality risk of certification data. The multidimensional scoring vector in this application represents the performance of the data category in each risk dimension. The audit risk in this application represents the degree of audit risk of the data category in the initial audit sample. Its role is to guide the adjustment of subsequent samples and the direction of key reviews.

[0084] In step 104, the risk weight of each data category in the knowledge graph is determined by the relationship between the graph nodes in the knowledge graph and the data review elements in the associated standard documents.

[0085] In some embodiments, determining the risk weight of each document category in the knowledge graph based on the association between graph nodes in the knowledge graph and document review elements in the associated standard documents can be achieved through the following steps:

[0086] The semantic distribution features of each data category in the data to be reviewed and certified are determined by the graph nodes in the knowledge graph.

[0087] The risk weight of each document category in the knowledge graph is determined based on the semantic distribution characteristics and the correlation between the document review elements in the associated standard documents.

[0088] In specific implementation, determining the semantic distribution characteristics of each document category in the knowledge graph of the documents to be reviewed and certified by the graph nodes in the knowledge graph can be achieved in the following way: Graph analysis techniques (such as graph traversal, depth-first search, shortest path, etc.) can be used to identify the corresponding graph nodes in the knowledge graph of each document category in the documents to be reviewed and certified. Then, by statistically analyzing the association density and coverage between the graph nodes corresponding to each document category and other graph nodes, the semantic distribution characteristics of each document category in the knowledge graph can be measured to understand the importance and spread of each document category in the knowledge graph, providing basic data for subsequent risk weight determination. Determining the risk weight of each document category in the knowledge graph of the documents to be reviewed and certified based on the semantic distribution characteristics and the association between each document review element in the associated standard document can be achieved in the following way: Information entropy calculation methods (such as Shannon entropy) can be used, combined with semantic distribution characteristics... The entropy value of the potential audit risk of each document category in the documents to be audited and certified is determined by the correlation between features (such as association density and coverage) and each document review element. The higher the entropy value, the more dispersed the connection between the corresponding document category and the document review elements in the associated standard document, and the greater the potential audit risk of the corresponding document category. The correlation represents the degree of connection between the document review elements in the associated standard document and the graph nodes in the knowledge graph. Therefore, the semantic distribution features and the correlation of document review elements can be integrated by weighted average or multiple regression models to obtain the risk weight of each document category in the knowledge graph in the documents to be audited and certified. The risk weight reflects the uncertainty of the semantic information of each document category in the knowledge graph, providing a quantitative basis for subsequent risk judgment, thereby understanding the potential audit risk of the document categories in the documents to be audited and certified. Other methods can also be used to determine this in other embodiments, which are not limited here.

[0089] It should be noted that the semantic distribution features in this application represent the connection relationship and strength between the data category and other graph nodes in the knowledge graph; the risk weight in this application represents the uncertainty of risk that may exist in the category data during the review process, which is used to guide the sampling bias and review focus of the data to be reviewed and certified, so as to improve the perception ability of subsequent sampling review.

[0090] In step 105, the sample utility value of each data category in the initial audit sample is determined based on the non-uniform distribution characteristics of all risk weights and all audit risks.

[0091] In some embodiments, determining the sample utility value of each data category in the initial audit sample during sampling audit, based on the non-uniform distribution characteristics of all risk weights and all audit risks, can be achieved through the following steps:

[0092] Determine the non-uniform distribution characteristics of all risk weights;

[0093] Based on the non-uniform distribution characteristics and all audit risks, the audit value of each data category in the initial audit sample is evaluated to obtain the sample utility value of each data category in the initial audit sample during the sampling audit.

[0094] In practice, the non-uniform distribution characteristics of all risk weights can be determined in the following way: statistical methods (such as standard deviation, coefficient of variation, Pearson correlation coefficient, etc.) can be used to measure the dispersion of risk weights in different data categories, and this dispersion can be used as the non-uniform distribution characteristics of all risk weights, so as to determine which data categories are more representative in risk assessment. Other methods can also be used in other embodiments, and there is no limitation here. It should be noted that if the risk weight distribution shows high dispersion, it means that a certain category of data may bear a higher audit risk. Therefore, more attention should be paid to it during sampling audit to determine which categories are more representative in risk assessment.

[0095] In specific implementation, the audit value of each document category in the initial audit sample is evaluated based on the non-uniform distribution characteristics and all audit risks. The sample utility value of each document category in the initial audit sample during sampling audit can be obtained in the following way: A team of experts in the relevant field can combine the non-uniform distribution characteristics and all audit risks, through collective discussion and experience-based judgment, to evaluate the audit value of each document category in the initial audit sample. The evaluation results are then used as the sample utility value of each document category during sampling audit, so as to optimize sample allocation subsequently. For example, in expert group discussions, the expert group can discuss which document categories need more... Pay close attention to data categories with higher risk weights (significantly non-uniform distribution characteristics), as these categories typically exhibit greater uncertainty during sampling and review, and therefore should be given higher review priority. Furthermore, the expert panel can determine which data categories have higher review risks based on their review history, frequency of issues, and corresponding review risks. Generally, data categories with higher review risks will be assigned a higher contribution to ensure they account for a larger share in the sampling and review process. Determining sample utility values ​​serves as a crucial basis for dynamically adjusting the sampling structure, supporting subsequent optimization of sample proportions based on sample utility values, and improving the effectiveness and representativeness of the sampling. Other methods can also be used in other embodiments, and are not limited here.

[0096] It should be noted that the non-uniform distribution feature in this application represents the feature used to measure the dispersion of risk weights. In addition, it can be used to identify which data categories have high uncertainty or potential risks during the review process. The sample utility value in this application represents the representativeness of the data categories in the initial review sample during the sampling review. Therefore, it is a key parameter used to guide the dynamic optimization of the final review sample structure. The sampling strategy can be dynamically adjusted through the sample utility value to improve the review effect.

[0097] In step 106, the initial audit samples for the certification materials to be audited are dynamically updated based on all sample utility values, and the final audit sample list for the target enterprise is output.

[0098] For specific implementation, refer to Figure 3As shown, this diagram is an exemplary flowchart of dynamic updates implemented in some embodiments of this application. Dynamically updating the initial audit samples when sampling and auditing the certification materials to be audited based on all sample utility values, and outputting the final audit sample list for the target enterprise, can be achieved in the following way: Dynamically updating the initial audit samples when sampling and auditing the certification materials to be audited based on all sample utility values ​​involves adjusting the weights or increasing / decreasing the quantity of data categories in the initial audit samples. For example, if a certain data category in the initial audit samples exhibits a high sample utility value (i.e., exceeding a preset utility threshold), it indicates a significant impact on the overall metrological system certification audit results of the target enterprise. Therefore, the sample quantity of that data category can be appropriately increased during dynamic updates. Conversely, if the sample utility value of a certain data category is low (i.e., not exceeding the preset utility threshold), its sample quantity can be appropriately reduced. The sample utility value can be judged by a preset utility threshold. The utility threshold is preset based on the experience of experts in relevant fields and historical experimental data. For example, the utility threshold can be set as the mean of historical sample utility values ​​plus or minus one standard deviation. In other embodiments, the utility threshold can also be dynamically determined by machine learning methods (such as cluster analysis). The proportion of increasing or decreasing the number of samples can be determined based on statistical principles and actual business needs, aiming to ensure the accuracy of certification audit while avoiding excessive adjustment of the number of samples. In other embodiments, other methods can also be used to set this proportion, which is not specifically limited here. After the above dynamic adjustment process, the final audit sample list of the target company will be more scientific and reasonable, and can more accurately reflect the overall certification data of the target company. It will provide a more representative and effective sample basis for subsequent metrological system certification audit work. Then, the output final audit sample list is used by the metrological system certification management system to sample and audit the metrological system certification data submitted by the target company.

[0099] Furthermore, in another aspect of this application, in some embodiments, this application provides a metrological system certification management system, which further includes a sampling and review unit. The sampling and review unit is used to sample and review the metrological system certification materials submitted by the enterprise. (Refer to...) Figure 4 The figure is a schematic diagram of the structure of a sampling review unit 400 according to some embodiments of this application. The sampling review unit 400 includes: an acquisition module 401, a processing module 402, and an execution module 403, which are described below:

[0100] The acquisition module 401 in this application is mainly used to acquire the associated standard documents of the target enterprise's certification materials to be audited;

[0101] Processing module 402, in this application, is mainly used to construct a knowledge graph for metrological system certification audit based on the certification materials to be audited and the certification clauses in the associated standard documents;

[0102] The processing module 402 described in this application is further configured to determine the constraints for sampling and auditing the certification materials to be audited based on the semantic relationship edges in the knowledge graph and the category distribution characteristics of the certification materials to be audited, extract the initial audit sample for the metrological system certification of the target enterprise from the certification materials to be audited based on the constraints, and then determine the audit risk of different categories of materials in the initial audit sample.

[0103] The processing module 402 described in this application is also used to determine the risk weight of each data category in the knowledge graph based on the association relationship between the graph nodes in the knowledge graph and the data review elements in the associated standard documents;

[0104] The processing module 402 described in this application is also used to determine the sample utility value of each data category in the initial audit sample during the sampling audit based on the non-uniform distribution characteristics of all risk weights and all audit risks;

[0105] The execution module 403 in this application is mainly used to dynamically update the initial audit sample when sampling and auditing the certification materials to be audited based on all sample utility values, and output the final audit sample list of the target enterprise.

[0106] The foregoing has detailed examples of the metrological system certification management system and method provided in the embodiments of this application. It is understood that the corresponding apparatus, in order to achieve the above functions, includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed by hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0107] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the above-described knowledge graph-based authentication data sampling and verification method.

[0108] In some embodiments, reference Figure 5The dashed lines in the figure indicate that the unit or module is optional. This figure is a schematic diagram of the structure of a computer device implementing the knowledge graph-based authentication data sampling and verification method of this application. The knowledge graph-based authentication data sampling and verification method in the above embodiments can be... Figure 5 The computer device 500 shown is used to implement this, and the computer device 500 includes at least one processor 501, a memory 502 and at least one communication unit 505. The computer device 500 may be a terminal device, a server or a chip.

[0109] The processor 501 can be a general-purpose processor or a special-purpose processor. For example, the processor 501 can be a central processing unit (CPU). The CPU can be used to control the computer device 500, execute software programs, and process data from the software programs. The computer device 500 may also include a communication unit 505 for inputting (receiving) and outputting (transmitting) signals.

[0110] For example, computer device 500 may be a chip, communication unit 505 may be the input and / or output circuit of the chip, or communication unit 505 may be the communication interface of the chip, and the chip may be a component of terminal device, network device or other device.

[0111] For example, computer device 500 may be a terminal device or a server, and communication unit 505 may be a transceiver of the terminal device or the server, or communication unit 505 may be a transceiver circuit of the terminal device or the server.

[0112] The computer device 500 may include one or more memories 502 storing a program 504. The program 504 can be executed by a processor 501 to generate instructions 503, causing the processor 501 to perform the methods described in the above method embodiments according to the instructions 503. Optionally, the memory 502 may also store data (such as a target audit model). Optionally, the processor 501 may also read data stored in the memory 502, which may be stored at the same storage address as the program 504, or the data may be stored at a different storage address than the program 504.

[0113] The processor 501 and memory 502 can be configured separately or integrated together, for example, integrated on the system-on-chip (SOC) of the terminal device.

[0114] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in the processor 501. The processor 501 can be a CPU, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, such as discrete gate, transistor logic devices, or discrete hardware components.

[0115] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0116] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-described knowledge graph-based authentication data sampling and verification method.

[0117] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0118] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A knowledge graph-based method for sampling and reviewing certification materials, used in a metrology system certification management system to sample and review metrology system certification materials submitted by enterprises, characterized in that... The method includes the following steps: Obtain the relevant standard documents for the target company's pending certification materials; A knowledge graph for metrological system certification audit is constructed based on the certification terms in the certification materials to be audited and the associated standard documents. Based on the semantic relationship edges in the knowledge graph and the category distribution characteristics of the certification materials to be audited, the constraints for sampling and auditing the certification materials to be audited are determined. Based on the constraints, the initial audit sample for the metrological system certification of the target enterprise is extracted from the certification materials to be audited, and then the audit risk of different categories of materials in the initial audit sample is determined. The risk weight of each document category in the knowledge graph is determined by the relationship between the graph nodes in the knowledge graph and the document review elements in the associated standard documents. Based on the non-uniform distribution characteristics of all risk weights and all audit risks, determine the sample utility value of each data category in the initial audit sample during the sampling audit; The initial audit sample is dynamically updated based on all sample utility values ​​when sampling and auditing the certification materials to be audited, and the final audit sample list of the target enterprise is output. Specifically, the constraints for sampling and reviewing the data to be reviewed and certified, determined based on the semantic relationship edges in the knowledge graph and the category distribution characteristics of the data to be reviewed and certified, include: Determine the category distribution characteristics of the documents to be audited and certified; The semantic core degree of each category of materials in the knowledge graph to be reviewed and certified is determined based on the semantic relation edges in the knowledge graph. The constraints for sampling and reviewing the certification materials to be reviewed are determined based on all semantic core degrees and the category distribution characteristics. Specifically, the initial audit sample for the target company's metrological system certification, extracted from the certification materials to be audited based on the aforementioned constraints, includes: The sampling quantity for each document category in the documents to be audited and certified is allocated based on the aforementioned constraints; Based on the sampling quantity, the initial audit sample for the metrological system certification of the target enterprise is selected from the certification materials to be audited; Specifically, determining the sample utility value of each document category in the initial audit sample during sampling audit, based on the non-uniform distribution characteristics of all risk weights and all audit risks, includes: Determine the non-uniform distribution characteristics of all risk weights; Based on the non-uniform distribution characteristics and all audit risks, the audit value of each data category in the initial audit sample is evaluated to obtain the sample utility value of each data category in the initial audit sample during the sampling audit.

2. The method as described in claim 1, characterized in that, The knowledge graph for metrological system certification auditing is constructed based on the certification documents to be audited and the certification clauses in the associated standard documents, specifically including: Extract the certification data elements of the target company from the certification documents to be audited; Extract audit clause elements based on the certification clause content in the aforementioned associated standard documents; A knowledge graph for metrological system certification audit is constructed based on the semantic relationships between the certification data elements and the audit clause elements.

3. The method as described in claim 1, characterized in that, Determining the audit risk of different document categories in the initial audit sample specifically includes: Determine the risk assessment indicators for each document category in the initial audit sample; Based on the risk assessment indicators, each document category in the initial audit sample is evaluated separately to obtain a multi-dimensional score vector for each document category; The review risk of different document categories in the initial review sample is determined by various multidimensional scoring vectors.

4. The method as described in claim 1, characterized in that, The risk weight of each document category in the knowledge graph is determined by the relationship between the graph nodes in the knowledge graph and the document review elements in the associated standard documents. Specifically, this includes: The semantic distribution features of each data category in the data to be reviewed and certified are determined by the graph nodes in the knowledge graph. The risk weight of each document category in the knowledge graph is determined based on the semantic distribution characteristics and the correlation between the document review elements in the associated standard documents.

5. A metrological system certification management system, which employs the method described in any one of claims 1 to 4 for sampling and reviewing certification materials, the metrological system certification management system comprising a sampling and review unit, the sampling and review unit being used to sample and review the metrological system certification materials submitted by enterprises, characterized in that, The sampling review unit includes: The acquisition module is used to acquire the associated standard documents of the target company's certification materials to be audited; The processing module is used to construct a knowledge graph for metrological system certification audit based on the certification materials to be audited and the certification clauses in the associated standard documents; The processing module is further configured to determine the constraints for sampling and auditing the certification materials based on the semantic relationship edges in the knowledge graph and the category distribution characteristics of the certification materials to be audited, extract the initial audit sample for the metrological system certification of the target enterprise from the certification materials to be audited based on the constraints, and then determine the audit risk of different categories of materials in the initial audit sample. The processing module is also used to determine the risk weight of each data category in the knowledge graph in the knowledge graph based on the relationship between the graph nodes in the knowledge graph and the data review elements in the associated standard documents; The processing module is also used to determine the sample utility value of each data category in the initial audit sample during the sampling audit based on the non-uniform distribution characteristics of all risk weights and all audit risks; The execution module is used to dynamically update the initial audit samples when sampling and auditing the certification materials to be audited based on all sample utility values, and output the final audit sample list of the target enterprise.

6. A computer device, characterized in that, The computer device includes a memory and a processor. The memory is used to store computer programs, and the processor is used to call and run the computer programs from the memory, so that the computer device performs the knowledge graph-based authentication data sampling and verification method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions or code that, when executed on a computer, cause the computer to implement the knowledge graph-based authentication data sampling and verification method as described in any one of claims 1 to 4.

Images