A method, system and medium for evidence chain quantification verification and closed-loop re-inspection scheduling

Abstract

The application discloses a kind of evidence chain quantization verification and closed loop reexamination scheduling method, system and medium, belong to the technical field of detection compliance supervision, comprising the following steps: S1, obtaining standardized data package;S2, generating evidence mapping set;S3, constructing four consistent atlas;S4, define verification score result;S5, reexamination and arrange path;S6, execute difference attribution;S7, output audit report and update organization risk image.By the above-mentioned mode, the present application obtains report text, original detection data, sample flow log, organization qualification range and standard clause library and carries out standardization processing, further extracts detection conclusion item, generates multidimensional evidence node, and constructs four consistent atlas containing report conclusion, original detection data, standard clause and organization qualification information, so that the verification object of detection report is converted into evidence chain level consistency verification from single result comparison.

Description

Technical Field

[0001] This invention relates to the field of testing compliance and regulatory technology, specifically to a method, system, and medium for quantitative verification and closed-loop re-inspection of evidence chains. Background Technology

[0002] As testing and inspection activities are increasingly used in product quality evaluation, compliance supervision, and risk traceability, third-party testing reports have gradually evolved from simple paper or electronic result files into comprehensive data objects that include report text, original testing data, sample transfer records, institutional qualification information, and standard clauses. Correspondingly, testing report management systems have begun to be developed with information technology covering report receipt, data archiving, workflow, basic review, and regulatory record keeping, in order to record and manage the formation and use of testing reports.

[0003] Existing technical solutions receive report texts, raw test data, and related circulation information through a test report system. They then perform basic review or result comparison of the test results and store, query, and archive information such as reports, test records, sample link logs, qualification scope, and standard clauses. This type of solution provides basic data support for the generation, circulation, and storage of test reports, and allows for further manual judgment based on report conclusions, raw data, or retest results when needed.

[0004] However, when verifying test reports, the aforementioned existing technologies usually rely primarily on the report results or partial data comparisons. It is still difficult to integrate the report conclusions, original test data, standard clauses, and institutional qualification scope into the same evidence chain structure for consistency verification. At the same time, for situations such as missing key evidence, broken links, and mismatched qualifications, there is a lack of quantifiable risk assessment criteria for missing certificates, which makes the triggering of re-inspection and the arrangement of re-inspection paths heavily reliant on human experience. Summary of the Invention

[0005] To address the aforementioned shortcomings of existing technologies, this invention provides a method, system, and medium for quantitative verification and closed-loop re-examination of evidence chains.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for quantitative verification and closed-loop re-examination of evidence chains includes the following steps:

[0008] S1. Obtain the original detection dataset, merge it, and perform standardization processing to obtain a standardized data package;

[0009] S2. Obtain the set of detection conclusion items through standardized data packets, generate multi-dimensional evidence nodes, calculate the evidence mapping credibility of multi-dimensional evidence nodes, and summarize the multi-dimensional evidence nodes according to the detection conclusions to generate the final evidence mapping set.

[0010] S3. Construct a quaternion consistency graph, which includes the following steps:

[0011] S301. Construct a set of four-element consistency nodes based on the report conclusions, original test data, standard clauses and institutional qualification information in the evidence mapping set, and form a set of graph connection edges based on the corresponding sample flow records in the evidence mapping set.

[0012] S302. Calculate the graph association completeness by using the graph connection edge set and the quaternary consistency node set, and calculate the graph coverage by using the quaternary consistency node set and the detection conclusion item set.

[0013] S303. Generate a quaternion-consistent graph based on the set of quaternion-consistent nodes, the set of graph connection edges, the graph association completeness, and the graph coverage.

[0014] S4. Based on the quaternary consistency map, calculate the overall consistency score and the overall missing certificate score, and define the verification score result;

[0015] S5. Based on the evidence mapping set, obtain the re-inspection index, trigger threshold judgment through the re-inspection index, generate the corresponding re-inspection path, calculate the re-inspection priority index, sort the re-inspection paths in descending order according to the re-inspection priority index, and form the re-inspection arrangement result.

[0016] S6. Based on the re-examination arrangement results, calculate the re-examination difference quantity and the source of difference, and calculate the attribution confidence. Obtain the difference attribution results based on the attribution confidence, multidimensional evidence nodes and the source of difference.

[0017] S7. Based on the attribution confidence level and the overall missing certificate score corresponding to the attribution results of the differences, calculate the risk level of each attribution result of the differences, generate a set of institutional risk profiles, and generate a testing and audit report based on all the attributions of differences.

[0018] Furthermore, in step S1, the raw test data set includes report text, raw test data, sample transfer log, institutional qualification scope, and standard clause library.

[0019] Furthermore, step S301 specifically involves constructing a corresponding quaternary consistency node based on the report conclusions, original test data, standard clauses, and institutional qualification information in each evidence node of the evidence mapping set;

[0020] Based on the sample flow records corresponding to each evidence node, and combined with the corresponding detection object identifier, detection item number and sample flow order, the association relationship between the quaternary consistency nodes is established to form a graph connection edge set.

[0021] Furthermore, the expression for the quadruple consistency node is:

[0022]

[0023] in, For the i-th quaternion consensus node, As the report concludes, This is the original test data. For institutional qualification information, For standard terms;

[0024] The expression for the set of quaternion-consistent nodes is:

[0025]

[0026] in, For a set of quaternion consistent nodes, This is the i-th quaternion consensus node;

[0027] The expression for the set of graph connection edges is:

[0028]

[0029] in, For the set of graph connection edges, Let y be the j-th graph connection edge, and y be the number of graph connection edges. This is a record of sample transfer.

[0030] Furthermore, in step S302, the formula for calculating the map association completeness is:

[0031]

[0032] in, For the completeness of the map association, This is the set of necessary associated edges that are actually established in the graph's connection edge set. The set of necessary associated edges established in the set of graph connection edges;

[0033]

[0034] in, For map coverage, Let be the total number of elements in the set of quaternion-consistent nodes. This represents the total number of elements in the set of test conclusion items.

[0035] Furthermore, in step S4, the formula for calculating the overall consistency score is:

[0036]

[0037] in, Score for overall consistency. Let be the consistency coefficient of the i-th node. This represents the number of quaternion-consistent nodes. For the completeness of the map association, For map coverage;

[0038] The overall formula for calculating the missing certificate score is as follows:

[0039]

[0040] in, The overall score is based on the number of missing certificates. Let x be the missing certificate coefficient of the i-th quaternary consensus node, x be the number of quaternary consensus nodes, and K be the graph coverage.

[0041] Furthermore, step S5 specifically includes:

[0042] S501. Based on each evidence node in the evidence mapping set, calculate the re-inspection index of the evidence node according to the credibility of the evidence mapping and the verification score.

[0043] S502. Determine the re-inspection trigger threshold for the re-inspection index and generate the corresponding re-inspection path;

[0044] S503. Based on all evidence nodes to be re-examined, calculate the re-examination priority index according to the map association completeness, map coverage and evidence mapping credibility.

[0045] S504. After obtaining the re-inspection paths corresponding to all evidence nodes to be re-inspected, sort the re-inspection paths from largest to smallest according to the re-inspection priority index to form the re-inspection arrangement result.

[0046] Furthermore, in step S6, based on the re-inspection arrangement results, each re-inspection path is executed sequentially according to the re-inspection priority, and the obtained re-inspection results are compared with the original detection data in the corresponding evidence node to calculate the re-inspection difference.

[0047] Based on the difference in the re-examination and the credibility of the evidence mapping, the attribution confidence of the source of the difference is calculated.

[0048] A system for quantifying and verifying evidence chains and arranging closed-loop re-examinations includes:

[0049] Data access module: Acquires the raw detection dataset, merges it, and performs standardization processing to obtain a standardized data packet;

[0050] Conclusion extraction and mapping module: Obtains the set of detection conclusion items through standardized data packets, generates multi-dimensional evidence nodes, calculates the evidence mapping credibility of multi-dimensional evidence nodes, and summarizes the multi-dimensional evidence nodes according to the detection conclusions to generate the final evidence mapping set;

[0051] The graph construction module constructs a set of quaternary consistent nodes and a set of graph connecting edges based on the evidence mapping set. It calculates the graph association completeness through the set of graph connecting edges and the set of quaternary consistent nodes, and calculates the graph coverage through the set of quaternary consistent nodes and the set of detection conclusion items. Finally, it generates a quaternary consistent graph.

[0052] Scoring module: Calculates the overall consistency score and overall missing certificate score based on the quaternary consistency map, and finally defines the verification score result based on the overall consistency score and missing certificate score;

[0053] Re-inspection orchestration module: Based on the evidence mapping set, a re-inspection index is obtained. The re-inspection index triggers a threshold judgment, generates a corresponding re-inspection path, calculates the re-inspection priority index, sorts the re-inspection paths in descending order according to the re-inspection priority index, and forms the re-inspection orchestration result.

[0054] Attribution module: Based on the re-examination arrangement results, calculate the re-examination difference quantity and the source of difference, and calculate the attribution confidence. Based on the attribution confidence, multidimensional evidence nodes and the source of difference, obtain the difference attribution results.

[0055] Audit module: Based on the attribution confidence level and the overall missing certificate score corresponding to the attribution results of the discrepancies, calculate the risk level of each discrepancy attribution result, generate a set of institutional risk profiles, and generate an audit report based on all discrepancy attributions.

[0056] A computer storage medium for storing program data, which, when executed by a computer, is used to implement a method for quantitative verification of the chain of evidence and closed-loop re-examination arrangement.

[0057] Compared with the prior art, the beneficial effects of this invention are as follows:

[0058] 1. This invention acquires and standardizes report text, raw test data, sample flow logs, institutional qualification scope, and standard clause library. It then extracts test conclusion items, generates multi-dimensional evidence nodes, and constructs a quaternary consistency graph containing report conclusions, raw test data, standard clauses, and institutional qualification information. This transforms the verification of test reports from a single result comparison to a chain-of-evidence consistency verification. Simultaneously, it quantifies the sufficiency of correspondence between test conclusions and supporting evidence, the degree of evidence deficiency, and overall consistency through evidence mapping credibility, graph association completeness, graph coverage, overall consistency score, and overall lack of evidence score. This effectively solves the problems in existing technologies, such as the difficulty in unifying verification report conclusions, raw data, standard clauses, and institutional qualification scope, and the lack of quantitative basis for missing evidence risk.

[0059] 2. This invention also calculates a re-inspection index based on evidence mapping credibility, overall consistency score, and overall missing evidence score. When a trigger threshold is reached, a re-inspection path is automatically generated. Then, a re-inspection scheduling result is formed based on the re-inspection priority index, enabling re-inspection tasks to be arranged in an orderly manner according to the weakness of the evidence chain and verification risks. After the re-inspection results are fed back, this invention further calculates the re-inspection discrepancy, generates the source of the discrepancy, calculates the attribution confidence, and updates the institution's risk profile and outputs a testing audit report in conjunction with the missing evidence score. This connects re-inspection triggering, path scheduling, discrepancy attribution, risk profile updating, and audit documentation into a closed-loop process, effectively solving the problems of reliance on manual experience for re-inspection, non-standard discrepancy attribution after re-inspection, and insufficient support for regulatory review in existing technologies. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0061] Figure 1 This is a flowchart illustrating the method structure of the present invention.

[0062] Figure 2 This is a structural block diagram of the quaternary consistency map of the present invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0064] Example 1: A method for quantitative verification and closed-loop re-examination of the chain of evidence, comprising the following steps:

[0065] S1. Obtain standardized data packets

[0066] The report text R, raw test data D, sample transfer log L, institutional qualification scope Q, and standard clause library S are obtained as the raw test data set. The raw test data is then standardized to form a unified data package. A data integrity coefficient is calculated, and based on this coefficient, a standardized data package is obtained. Specifically:

[0067] S101. Calculate the unified data packet based on the original detection data set:

[0068]

[0069] in, To unify data packets, The set of data standardization functions includes at least text field extraction functions, unit normalization functions, timestamp format conversion functions, sample number normalization functions, institutional qualification code mapping functions, standard clause version matching functions, and abnormal field marking functions. Each function transforms data from different sources according to a preset processing order, and retains the source identifier and processing log for each transformation result to ensure that the standardization process is verifiable. A collection of report texts, This is the original set of detection data. This is a collection of sample flow logs. This refers to a collection of institutional qualifications. A collection of standard clauses;

[0070] By unifying and formatting multi-source heterogeneous detection texts, logs, and qualification standards, a standardized and structured underlying data foundation is built for subsequent evidence chain verification, effectively improving the parsing compatibility and processing efficiency of cross-source data.

[0071] S102. Construct a data integrity coefficient for determination and obtain standardized data packets;

[0072] To evaluate whether the unified data packet meets the requirements for subsequent evidence chain construction, a data integrity coefficient is calculated for the unified data packet. The formula for calculating the data integrity coefficient is as follows:

[0073]

[0074] in, The data integrity coefficient. To ensure consistency in the number of data categories actually retrieved in the data packets, To preset the total number of data categories to be acquired ( =5, corresponding to: Report Text (R), Raw Test Data (D), Sample Transfer Log (L), Institutional Qualification Scope (Q), and Standard Clause Library (S), respectively.

[0075] When the data integrity coefficient satisfies:

[0076]

[0077] in, The data integrity coefficient. The threshold for determining data integrity;

[0078] If the unified data packet meets the requirements for constructing the subsequent chain of evidence, then the missing data categories will be supplemented until the integrity requirements are met.

[0079] Once the unified data packet meets the integrity requirements, it is output to the evidence mapping stage and used as the sole input data for subsequent steps to obtain the standardized data packet.

[0080]

[0081] in, To standardize data packets, To standardize data packets;

[0082] By unifying and formatting multi-source heterogeneous detection texts, logs, and qualification standards, a standardized and structured underlying data foundation is built for subsequent evidence chain verification, effectively improving the parsing compatibility and processing efficiency of cross-source data.

[0083] S2, Generate a set of evidence maps

[0084] The detection conclusion item set is obtained by standardizing data packets, multi-dimensional evidence nodes are generated, the evidence mapping credibility of the multi-dimensional evidence nodes is calculated, and the multi-dimensional evidence nodes are summarized according to the detection conclusions to generate the final evidence mapping set.

[0085] S201. Perform multidimensional semantic association and feature calculation based on standardized data packets to obtain a set of detection conclusion items, and generate multidimensional evidence nodes based on the set of detection conclusion items;

[0086] Obtain the standardized data package, and parse the following sets of components: report text, raw test data, sample transfer logs, institutional qualification information, and standard clauses. The expression is:

[0087]

[0088] in, To standardize data packets, A standardized collection of report texts. The standardized set of raw test data The standardized sample flow log set, This is a standardized collection of institutional qualification information. It is a standardized set of standard clauses;

[0089] Multidimensional semantic association and feature calculation are performed on standardized data packets to identify detection items, detection results, and detection conclusions in the report. A set of detection conclusion items is generated according to the detection items, and the calculation formula is as follows:

[0090]

[0091] in, For the set of detection conclusion items, For the i-th detection conclusion, This represents the total number of test conclusion items;

[0092] Furthermore, multidimensional semantic association and feature calculation include matching of detection project name, sample number, detection time, standard clause keyword, institutional qualification scope, and detection result numerical range. The unique identifier of the detection conclusion item is generated by combining the report number, sample number, detection project number, and conclusion sequence number. The system establishes the mapping relationship between the detection conclusion and the corresponding evidence item only when the comprehensive matching degree reaches the preset association threshold.

[0093] Based on the set of detection conclusion items The i-th detection conclusion in ,by The unique identifier serves as the core anchor point, and a corresponding chain of evidence is derived in reverse. This involves targeted tracing and fragment extraction within the report conclusion set, the original test data set, the sample flow log set, the institutional qualification information set, and the standard clause set, respectively, to calculate the corresponding five local evidence items and establish the corresponding evidence nodes. The multidimensional evidence node expression is as follows:

[0094]

[0095] in, For the i-th multidimensional evidence node, For the i-th detection conclusion, To and The corresponding report conclusions, To and The corresponding raw test data, To and The corresponding sample transfer records, To and Corresponding institutional qualification information To and Corresponding standard terms;

[0096] By accurately anchoring the core detection conclusion through multi-dimensional semantic association, and extracting five related local evidence fragments in reverse targeting based on the identifier, a deep interlock and logical binding between the upper-level business conclusion and the lower-level original data is achieved, providing a cornerstone structure for building a robust and tamper-proof network of evidence;

[0097] S202. Calculate the credibility of the evidence mapping for each evidence node in the multidimensional evidence nodes, and summarize all evidence nodes containing credibility in the natural order of the detection conclusions to generate the final evidence mapping set.

[0098] To evaluate the reliability of the correlation between the test results and the corresponding evidence, the evidence mapping credibility is calculated for each evidence node, and the calculation formula is as follows:

[0099]

[0100] in, Let the credibility of the evidence mapping for the i-th evidence node be denoted as . The number of pieces of evidence that successfully establish a correspondence in the i-th evidence node (evidence that can participate in establishing a correspondence includes: report text, original test data, sample transfer log, institutional qualification information, standard clauses, therefore...) );

[0101] Furthermore, the credibility of evidence mapping introduces evidence quality weights on the basis of quantity completeness. Weight coefficients are assigned to the matching degree of report text, consistency of original data, continuity of sample circulation time, coverage of institutional qualifications, and adaptation of standard clauses. When any evidence item exists but the matching degree is lower than the corresponding threshold, the evidence item is treated as low credibility or invalid mapping, thereby avoiding artificially high credibility simply because the quantity of evidence is complete.

[0102] When all corresponding evidence exists:

[0103]

[0104] At this point, it is clear that the test result is supported by complete evidence.

[0105] When some evidence is missing, the credibility of the evidence mapping decreases as the number of corresponding successes decreases;

[0106] Finally, all multidimensional evidence nodes are summarized in the order of detection conclusions to generate an evidence mapping set:

[0107]

[0108] in, For evidence mapping set, For the i-th evidence node, This represents the total number of test conclusion items;

[0109] In the evidence mapping set, each evidence node saves the report conclusions and original test data, sample transfer logs, institutional qualification information and standard clauses, and synchronously records the corresponding evidence mapping credibility.

[0110] By configuring the credibility of evidence mapping for each evidence node, the sufficiency of the correspondence between the detection conclusion and the supporting evidence can be quantified, providing a calculable basis for subsequent consistency scoring, missing evidence scoring, and re-inspection triggering;

[0111] S3. Construct a quaternion consistency graph

[0112] Based on the evidence mapping set, a set of quaternary consistent nodes and a set of graph connection edges are constructed respectively, and the graph association completeness and graph coverage are calculated and integrated into a quaternary consistent graph.

[0113] S301. Based on the report conclusions, original test data, standard clauses and institutional qualification information in each evidence node of the evidence mapping set, construct the corresponding quaternary consistency node and generate the corresponding quaternary consistency node set. At the same time, establish the association relationship between the quaternary consistency nodes based on the sample flow records corresponding to each evidence node to form a graph connection edge set.

[0114] Based on the report conclusions, original test data, standard clauses, and institutional qualification information in each evidence node of the evidence mapping set, a corresponding quaternary consistency node is constructed (sample transfer records are not included in the composition of quaternary consistency nodes, but are retained as the basis for establishing the subsequent association between quaternary consistency nodes), and sample transfer records are retained as the basis for node association. The expression of the quaternary consistency node is:

[0115]

[0116] in, For the i-th quadruple consensus node, As the report concludes, This is the original test data. For institutional qualification information, For standard terms;

[0117] And generate the corresponding set of quaternion consistent nodes:

[0118]

[0119] in, For a set of quaternion-consistent nodes, This is the i-th quaternion consensus node;

[0120] Based on the sample transfer records corresponding to each evidence node By combining the corresponding detection object identifier, detection item number, and sample flow sequence, the association relationship between the quaternary consistency nodes is established, forming a set of graph connection edges:

[0121]

[0122] in, For the set of graph connection edges, Let y be the j-th graph connection edge, and y be the number of graph connection edges. For sample transfer records;

[0123] A connection edge is established when any two quaternary consistency nodes meet any of the following conditions: they correspond to the same sample transfer record, the same test object, the same test item, the same standard clause, or the same institutional qualification. The sample transfer record is used to determine the transfer relationship and association order between nodes, so that the quaternary consistency nodes form a complete chain of evidence.

[0124] S302. Calculate the graph association completeness using the graph connection edge set and the quaternion-consistent node set, and simultaneously calculate the graph coverage using the quaternion-consistent node set and the detection conclusion item set. Specifically:

[0125] To evaluate the completeness of the relationships between nodes in a quaternion consistency graph, the graph's association completeness is calculated:

[0126]

[0127] in, For the completeness of the map association, This is the set of necessary associated edges that are actually established in the graph's connected edge set. The set of necessary associated edges established in the set of graph connection edges;

[0128] The higher the graph association completeness, the more complete the association relationship established between the quaternion consistent nodes;

[0129] To evaluate the coverage of the constructed quaternary consistency nodes based on the detection results, the map coverage rate is calculated:

[0130]

[0131] in, For map coverage, Let be the total number of elements in the set of quaternion-consistent nodes. The total number of elements in the set of test conclusion items;

[0132] when =1 indicates that all detection results have established corresponding quaternary consistency nodes. <1 indicates that some detection results failed to establish a complete quaternary consistency node;

[0133] S303. Generate a quaternion-consistent graph based on the set of quaternion-consistent nodes, the set of graph connection edges, the graph association completeness, and the graph coverage.

[0134] The expression for a quaternion consistency graph is:

[0135]

[0136] in, This is a quaternary consistency map. For a set of quaternion consistent nodes, For the set of graph connection edges, For the completeness of the map association, For map coverage;

[0137] In the quaternary consistency map, each quaternary consistency node stores the correspondence between the report conclusion, the original test data, the standard clauses and the institution's qualification information. The connection edges of each map store the association relationship between different quaternary consistency nodes based on the sample flow record. At the same time, the overall map records the map association completeness and map coverage.

[0138] S4. Define the verification scoring results

[0139] The overall consistency score and overall missing document score are calculated based on the quaternary consistency map. Finally, the verification score result is defined based on the overall consistency score and missing document score. The specific steps are as follows:

[0140] S401. Obtain each quaternary consistency node from the set of quaternary consistency nodes in the quaternary consistency graph. The consistency between the report conclusions, raw test data, standard clauses, and institutional qualifications was verified to obtain the corresponding node consistency coefficients:

[0141]

[0142] in, Let be the consistency coefficient of the i-th node. Let be the number of elements that satisfy the consistency requirement in the i-th element-consistent node;

[0143] Furthermore, the node consistency verification should take the correspondence between elements as the judgment object, including whether the report conclusion is consistent with the original test data, whether the original test data falls within the scope of the standard clause, whether the test items are included in the scope of the institution's qualifications, and whether the conclusions stated in the report text are consistent with the judgment results of the clause;

[0144] therefore:

[0145]

[0146] when 1. This indicates that the quaternary consensus nodes are completely consistent;

[0147] Based on the node consistency coefficient, and combined with the graph association completeness and graph coverage, the overall consistency score is calculated:

[0148]

[0149] in, Score for overall consistency. Let be the consistency coefficient of the i-th node. This represents the number of quaternion-consistent nodes. For the completeness of the map association, For map coverage;

[0150] The higher the overall consistency of the quaternary consistency nodes, and the more complete and comprehensive the graph associations, the higher the consistency score.

[0151] S402. To evaluate the evidence gaps at each quaternary consensus node, calculate the evidence gap coefficient for each quaternary consensus node based on the evidence mapping credibility and the graph association completeness:

[0152]

[0153] in, The missing certificate coefficient for the i-th quaternion consensus node ( ), U represents the credibility of the evidence mapping, and U represents the completeness of the graph association.

[0154] Furthermore, when calculating the missing evidence coefficient, the missing evidence of a single node is taken as the dominant factor, and the completeness of the graph association is taken as the correction factor. A minimum missing evidence retention item is set. When the credibility of the evidence mapping is lower than the preset credibility threshold, even if the completeness of the graph association is high, the risk of missing evidence of the corresponding node is still retained to avoid the overall graph integrity from covering up the missing evidence of local evidence.

[0155] Based on the missing certificate coefficients of all quaternary consistency nodes and combined with the map coverage, the overall missing certificate score is calculated:

[0156]

[0157] in, The overall score is based on the number of missing certificates. Let x be the missing certificate coefficient of the i-th quaternary consensus node, x be the number of quaternary consensus nodes, and K be the graph coverage.

[0158] S403. Define the verification score result based on the consistency score and the missing document score:

[0159]

[0160] in, To verify the scoring results, Score for overall consistency. The overall score is based on the number of missing certificates.

[0161] By analyzing the quaternary consistency graph, the consistency coefficient and missing evidence coefficient of each quaternary consistency node are calculated. The consistency score and missing evidence score are formed by combining the graph association completeness and graph coverage. This allows for a quantitative evaluation of the completeness of the evidence chain and the risk of missing evidence in the test report. This provides a unified and objective scoring basis for subsequent automatic re-inspection triggering, re-inspection path arrangement and risk level determination.

[0162] S5. Automatically trigger a re-inspection and re-arrange the path when the threshold is reached.

[0163] Based on the evidence mapping set, and according to the credibility of the evidence mapping and the verification score, a re-inspection index is obtained. The re-inspection index triggers a threshold judgment, generates a corresponding re-inspection path, calculates the re-inspection priority index, and sorts the re-inspection paths in descending order according to the re-inspection priority index to form a re-inspection arrangement result.

[0164] S501. Based on each evidence node in the evidence mapping set, calculate the re-inspection index of the evidence node according to the credibility of the evidence mapping and the verification score.

[0165] For each evidence node in the evidence mapping set, the re-examination index of the evidence node is calculated based on the evidence mapping credibility, the overall consistency score in the verification scoring results, and the overall missing evidence score:

[0166]

[0167] in, Let be the re-examination index of the i-th evidence node. The overall score is based on the number of missing certificates. Score for overall consistency. To map the credibility of evidence;

[0168] Before calculation, the re-inspection index normalizes the overall consistency score, overall missing certificate score, and evidence mapping credibility to the range of 0 to 1, and sets weight coefficients. When the calculation result is less than 0, it is taken as 0, and when it is greater than 1, it is taken as 1, so that the re-inspection index between different test reports, different test items, and different institutions is comparable.

[0169] When the consistency score decreases, the missing evidence score increases, or the credibility of the evidence mapping decreases, the re-examination index of the corresponding evidence node gradually increases.

[0170] S502. Determine the re-inspection trigger threshold for the re-inspection index and generate the corresponding re-inspection path;

[0171] Set the re-inspection trigger threshold to ,when: If so, the i-th evidence node is determined to enter the state pending re-examination, and the corresponding re-examination path is generated:

[0172]

[0173] in, This is the re-examination path for the i-th item. For the corresponding evidence nodes, The corresponding re-inspection methods (including at least: re-inspection by the original testing institution, cross-inspection by a third party, and re-inspection by regulatory sampling).

[0174] when If so, the current verification result of the evidence node will be maintained, and a re-examination will not be performed;

[0175] S503. Based on all evidence nodes to be re-examined, calculate the re-examination priority index according to the map association completeness, map coverage and evidence mapping credibility.

[0176] For all evidence nodes to be re-examined, a priority index for the re-examination task is calculated based on the completeness of the graph association, the graph coverage, and the credibility of the evidence mapping. The formula for calculating the priority index is as follows:

[0177]

[0178] in, Let be the priority index of the i-th re-inspection task. To map the credibility of evidence, The completeness of the graph association of evidence nodes. The priority index represents the graph coverage of evidence nodes. The higher the priority index, the lower the integrity of the evidence chain for the corresponding evidence node, and the more likely it is to be re-examined.

[0179] By calculating the re-examination priority index, the priorities among multiple re-examination tasks can be distinguished, so that tasks with weak evidence chains, insufficient map coverage, or low mapping credibility can be processed first.

[0180] S504. After obtaining the re-inspection paths corresponding to all evidence nodes to be re-inspected, sort the re-inspection paths from largest to smallest according to the re-inspection priority index to form the re-inspection arrangement result.

[0181] The definition of the re-inspection and arrangement result is:

[0182]

[0183] in, The results of the re-examination are arranged (including the corresponding evidence nodes, the corresponding re-examination methods, and the corresponding execution priorities). Sort by re-inspection priority index from largest to smallest. For the re-examination path of the i-th item, Let be the priority index of the i-th re-inspection task. The number of tasks awaiting re-inspection;

[0184] By generating re-inspection scheduling results according to re-inspection priorities, the re-inspection targets, re-inspection methods, and execution priorities can be output in a structured manner, which facilitates the assignment of re-inspection tasks, execution tracking, and supervision record keeping.

[0185] S6. Perform difference attribution after the retest results are returned.

[0186] Based on the re-examination arrangement results, the re-examination differences and sources of differences are calculated. At the same time, based on the re-examination difference amount and the evidence mapping credibility, the attribution confidence is calculated, and the difference attribution results are calculated based on the attribution confidence, evidence nodes and sources of differences.

[0187] S601. Based on the re-inspection arrangement results, execute each re-inspection path sequentially according to the re-inspection priority, compare the obtained re-inspection results with the original detection data in the corresponding evidence nodes, and calculate the re-inspection difference:

[0188]

[0189] in, Let be the re-examination difference of the i-th task. Let be the number of pieces of evidence indicating a discrepancy between the test results and the original test data during the re-examination of the i-th task. Let be the total number of evidence items corresponding to the i-th evidence node;

[0190] when 0 indicates that the re-examination result is completely consistent with the original evidence. →1 indicates that the differences in evidence are gradually increasing;

[0191] Based on the difference in re-inspection The changes in various pieces of evidence during the re-examination process are analyzed, and evidence with discrepancies is categorized to generate a set of sources of discrepancies, represented as follows:

[0192]

[0193] in, For the set of sources of difference, For the i-th difference source;

[0194] By comparing the differences between the re-examination results and the original evidence, the magnitude of the changes in evidence before and after the re-examination can be quantified, and the source of the differences can be preliminarily located, providing basic data for subsequent attribution of anomalies.

[0195] S602. Based on the re-examination difference and the confidence level of the evidence mapping, calculate the attribution confidence level of the source of the difference:

[0196]

[0197] in, Let i be the attribution confidence level for the source of the difference in the i-th term. Let be the re-examination difference of the i-th task. To map the credibility of evidence;

[0198] S603. Obtain the differential attribution results based on attribution confidence, evidence nodes, and sources of difference;

[0199] When the variance is large and the confidence level of the evidence mapping is low, the attribution confidence gradually increases, indicating that the corresponding source of variance has a high probability of being abnormal. The expression for the variance attribution result is as follows:

[0200]

[0201] The attribution results for the differences are expressed as follows:

[0202]

[0203] in, For the attribution results of the differences, For the i-th attribution result, For the corresponding evidence nodes, For the corresponding sources of difference, The corresponding attribution confidence level;

[0204] Furthermore, the attribution confidence is calculated by combining the weight of the source of difference, the severity of the difference, and the credibility of the evidence mapping. When there are multiple sources of difference for the same evidence node, the attribution confidence of each source of difference is calculated separately, and multi-source combined attribution results are allowed to reflect the situation that the actual anomaly may be caused by multiple links. By calculating the attribution confidence of the source of difference, the re-examination difference can be combined with the credibility of the evidence to identify the evidence links that are more likely to cause the anomaly and improve the interpretability of the attribution results.

[0205] S7. Output the audit report and update the institutional risk profile.

[0206] Based on the attribution confidence level and overall missing evidence score corresponding to the attribution results of the discrepancies, the risk level of each attribution result is calculated, and an institutional risk profile set is generated. Finally, an audit report is generated based on all discrepancies, specifically including the following steps:

[0207] S701. Based on the attribution confidence and missing certificate score corresponding to each difference attribution result, calculate the risk level of each difference attribution result, generate an institutional risk profile set, and update it synchronously.

[0208] The formula for calculating the risk level corresponding to the attribution result is:

[0209]

[0210] in, Let i be the risk level corresponding to the i-th attribution result. For the corresponding attribution confidence level, Scoring for missing certificates;

[0211] The higher the risk level, the higher the probability that the corresponding evidence node is abnormal;

[0212] Furthermore, the risk level calculation incorporates factors such as the institution's historical frequency of anomalies, the validity period of the institution's qualifications, the credibility of the re-inspection method, the weight of regulatory attention, and the concentration of anomalies in similar projects.

[0213] Based on the risk levels corresponding to all attribution results, an institutional risk profile is generated:

[0214]

[0215] in, Create a risk profile for the i-th institution. For institutional qualification information, Risk level;

[0216] The risk profiles of all institutions are aggregated to form an institutional risk profile set, the expression of which is:

[0217]

[0218] in, A collection of institutional risk profiles. Create a risk profile for the i-th institution;

[0219] When the same institution corresponds to multiple attribution results, the corresponding risk levels will be summarized and statistically analyzed to generate a risk profile for that institution.

[0220] The newly generated institutional risk profile is linked with the institution's historical risk profiles. The risk level, abnormal evidence type, attribution results of discrepancies, and verification time of the corresponding institution are updated to form the latest institutional risk profile. The institutional risk profile set is also updated simultaneously.

[0221] S702. Based on all the attribution results and the updated institutional risk profile set, generate the audit report:

[0222]

[0223] in, To examine the audit report, For the attribution results of the differences, A collection of institutional risk profiles;

[0224] Output test audit report This serves as the final output of the automatic verification process for the test report of this invention.

[0225] The testing audit report preserves the abnormal evidence, sources of discrepancies, risk levels, and institutional risk profiles corresponding to the testing report, achieving a complete record of the testing report verification process and providing unified data support for subsequent supervision and inspection by regulatory authorities, quality evaluation of testing institutions, and historical risk tracing.

[0226] A computer storage medium for storing program data, which, when executed by a computer, is used to implement a method for quantitative verification of the chain of evidence and closed-loop re-examination.

[0227] Example 2: A system for quantitative verification and closed-loop re-examination of the chain of evidence, comprising:

[0228] Data access module: Acquires the raw detection dataset, merges it, and performs standardization processing to obtain a standardized data packet;

[0229] Conclusion extraction and mapping module: Obtains the set of detection conclusion items through standardized data packets, generates multi-dimensional evidence nodes, calculates the evidence mapping credibility of multi-dimensional evidence nodes, and summarizes the multi-dimensional evidence nodes according to the detection conclusions to generate the final evidence mapping set;

[0230] The graph construction module constructs a set of quaternary consistent nodes and a set of graph connecting edges based on the evidence mapping set. It calculates the graph association completeness through the set of graph connecting edges and the set of quaternary consistent nodes, and calculates the graph coverage through the set of quaternary consistent nodes and the set of detection conclusion items. Finally, it generates a quaternary consistent graph.

[0231] Scoring module: Calculates the overall consistency score and overall missing certificate score based on the quaternary consistency map, and finally defines the verification score result based on the overall consistency score and missing certificate score;

[0232] Re-inspection orchestration module: Based on the evidence mapping set, a re-inspection index is obtained. The re-inspection index triggers a threshold judgment, generates a corresponding re-inspection path, calculates the re-inspection priority index, sorts the re-inspection paths in descending order according to the re-inspection priority index, and forms the re-inspection orchestration result.

[0233] Attribution module: Based on the re-examination arrangement results, calculate the re-examination difference quantity and the source of difference, and calculate the attribution confidence. Based on the attribution confidence, multidimensional evidence nodes and the source of difference, calculate the difference attribution results, and define the difference attribution results.

[0234] Audit module: Based on the attribution confidence level and the overall missing certificate score corresponding to the attribution results of the discrepancies, calculate the risk level of each discrepancy attribution result, generate a set of institutional risk profiles, and generate an audit report based on all discrepancy attributions.

[0235] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for quantitative verification and closed-loop re-examination of the chain of evidence, characterized in that, Includes the following steps: S1. Obtain the original detection dataset, merge it, and perform standardization processing to obtain a standardized data package; S2. Obtain the set of detection conclusion items through standardized data packets, generate multi-dimensional evidence nodes, calculate the evidence mapping credibility of multi-dimensional evidence nodes, and summarize the multi-dimensional evidence nodes according to the detection conclusions to generate the final evidence mapping set. S3. Construct a quaternion consistency graph, which includes the following steps: S301. Construct a set of four-element consistency nodes based on the report conclusions, original test data, standard clauses and institutional qualification information in the evidence mapping set, and form a set of graph connection edges based on the corresponding sample flow records in the evidence mapping set. S302. Calculate the graph association completeness by using the graph connection edge set and the quaternary consistency node set, and calculate the graph coverage by using the quaternary consistency node set and the detection conclusion item set. S303. Generate a quaternion-consistent graph based on the set of quaternion-consistent nodes, the set of graph connection edges, the graph association completeness, and the graph coverage. S4. Based on the quaternary consistency map, calculate the overall consistency score and the overall missing certificate score, and define the verification score result; S5. Based on the evidence mapping set, obtain the re-inspection index, trigger threshold judgment through the re-inspection index, generate the corresponding re-inspection path, calculate the re-inspection priority index, sort the re-inspection paths in descending order according to the re-inspection priority index, and form the re-inspection arrangement result. S6. Based on the re-examination arrangement results, calculate the re-examination difference quantity and the source of difference, and calculate the attribution confidence. Obtain the difference attribution results based on the attribution confidence, multidimensional evidence nodes and the source of difference. S7. Based on the attribution confidence level and the overall missing certificate score corresponding to the attribution results of the differences, calculate the risk level of each attribution result of the differences, generate a set of institutional risk profiles, and generate a testing and audit report based on all the attributions of differences.

2. The evidence chain quantification verification and closed-loop re-examination arrangement method according to claim 1, characterized in that, In step S1, the raw test data set includes report text, raw test data, sample transfer log, institutional qualification scope, and standard clause library.

3. The evidence chain quantification verification and closed-loop re-examination arrangement method according to claim 1, characterized in that, Specifically, step S301 involves constructing a corresponding quaternary consistency node based on the report conclusions, original test data, standard clauses, and institutional qualification information of each evidence node in the evidence mapping set. Based on the sample flow records corresponding to each evidence node, and combined with the corresponding detection object identifier, detection item number and sample flow order, the association relationship between the quaternary consistency nodes is established to form a graph connection edge set.

4. The evidence chain quantification verification and closed-loop re-examination arrangement method according to claim 3, characterized in that, The expression for the quadruple consistency node is: in, For the i-th quadruple consensus node, As the report concludes, This is the original test data. For institutional qualification information, For standard terms; The expression for the set of quaternion-consistent nodes is: in, For a set of quaternion-consistent nodes, This is the i-th quaternion consensus node; The expression for the set of graph connection edges is: in, For the set of graph connection edges, Let y be the j-th graph connection edge, and y be the number of graph connection edges. This is a record of sample transfer.

5. The evidence chain quantification verification and closed-loop re-examination arrangement method according to claim 4, characterized in that, In step S302, the formula for calculating the map association completeness is as follows: in, For the completeness of the map association, This is the set of necessary associated edges that are actually established in the graph's connected edge set. The set of necessary associated edges established in the set of graph connection edges; in, For map coverage, Let be the total number of elements in the set of quaternion-consistent nodes. This represents the total number of elements in the set of test conclusion items.

6. The evidence chain quantification verification and closed-loop re-examination arrangement method according to claim 1, characterized in that, In step S4, the formula for calculating the overall consistency score is as follows: in, Score for overall consistency. Let be the consistency coefficient of the i-th node. This represents the number of quaternion-consistent nodes. For the completeness of the map association, For map coverage; The overall formula for calculating the missing certificate score is as follows: in, The overall score is based on the number of missing certificates. Let x be the missing certificate coefficient of the i-th quaternary consensus node, x be the number of quaternary consensus nodes, and K be the graph coverage.

7. The evidence chain quantification verification and closed-loop re-examination arrangement method according to claim 1, characterized in that, Step S5 specifically involves: S501. Based on each evidence node in the evidence mapping set, calculate the re-inspection index of the evidence node according to the credibility of the evidence mapping and the verification score. S502. Determine the re-inspection trigger threshold for the re-inspection index and generate the corresponding re-inspection path; S503. Based on all evidence nodes to be re-examined, calculate the re-examination priority index according to the map association completeness, map coverage and evidence mapping credibility. S504. After obtaining the re-inspection paths corresponding to all evidence nodes to be re-inspected, sort the re-inspection paths from largest to smallest according to the re-inspection priority index to form the re-inspection arrangement result.

8. The evidence chain quantification verification and closed-loop re-examination arrangement method according to claim 1, characterized in that, In step S6, based on the re-inspection arrangement results, each re-inspection path is executed sequentially according to the re-inspection priority. The obtained re-inspection results are compared with the original detection data in the corresponding evidence nodes, and the re-inspection difference is calculated. Based on the difference in the re-examination and the credibility of the evidence mapping, the attribution confidence of the source of the difference is calculated.

9. A system for quantitative verification and closed-loop re-examination of evidence chains, characterized in that, The system is used in the evidence chain quantification verification and closed-loop re-examination arrangement method according to any one of claims 1-8, the system comprising: Data access module: Acquires the raw detection dataset, merges it, and performs standardization processing to obtain a standardized data packet; Conclusion extraction and mapping module: Obtains the set of detection conclusion items through standardized data packets, generates multi-dimensional evidence nodes, calculates the evidence mapping credibility of multi-dimensional evidence nodes, and summarizes the multi-dimensional evidence nodes according to the detection conclusions to generate the final evidence mapping set; The graph construction module constructs a set of quaternary consistent nodes and a set of graph connecting edges based on the evidence mapping set. It calculates the graph association completeness through the set of graph connecting edges and the set of quaternary consistent nodes, and calculates the graph coverage through the set of quaternary consistent nodes and the set of detection conclusion items. Finally, it generates a quaternary consistent graph. Scoring module: Calculates the overall consistency score and overall missing certificate score based on the quaternary consistency map, and finally defines the verification score result based on the overall consistency score and missing certificate score; Re-inspection orchestration module: Based on the evidence mapping set, a re-inspection index is obtained. The re-inspection index triggers a threshold judgment, generates a corresponding re-inspection path, calculates the re-inspection priority index, sorts the re-inspection paths in descending order according to the re-inspection priority index, and forms the re-inspection orchestration result. Attribution module: Based on the re-examination arrangement results, calculate the re-examination difference quantity and the source of difference, and calculate the attribution confidence. Based on the attribution confidence, multidimensional evidence nodes and the source of difference, obtain the difference attribution results. Audit module: Based on the attribution confidence level and the overall missing certificate score corresponding to the attribution results of the discrepancies, calculate the risk level of each discrepancy attribution result, generate a set of institutional risk profiles, and generate an audit report based on all discrepancy attributions.

10. A computer storage medium, characterized in that, The computer storage medium is used to store program data, which, when executed by the computer, is used to implement the evidence chain quantification verification and closed-loop re-examination arrangement method as described in any one of claims 1 to 8.

Images