A business scenario-oriented data asset integrity assessment and risk prompting method

Abstract

The present application relates to the technical field of data asset evaluation, and discloses a data asset integrity evaluation and risk prompt method for business scenarios. A data query request of a purchaser is analyzed to obtain a target data asset category, an application scenario identifier, and a scenario requirement parameter set, and a candidate data asset is obtained. Based on the application scenario, integrity evaluation indexes, requirement thresholds, minimum tolerance thresholds, and weights are matched. A standardized feature statistical summary generated based on local original data on the supplier side is received, and in the case that the original data does not leave the domain, segmentation scoring and weighting are performed to obtain an overall integrity score. The risk level is determined in combination with the core index compliance state, and the quantitative business negative influence and risk mitigation scheme are determined by using a missing influence knowledge graph to form a structured risk report and display. The present application can improve the data asset scene adaptation evaluation and risk prompt capability, and reduce the risk of original data leaving the domain.

Description

Technical Field

[0001] This invention relates to the field of data asset assessment technology, and more specifically, to a method for assessing the integrity of data assets and providing risk warnings for business scenarios. Background Technology

[0002] With the development of data element circulation and data asset trading, data asset registration systems have gradually become an important infrastructure for data providers to publish data products and for data demanders to query, screen, and purchase data assets. Data purchasers typically need to judge whether candidate data assets can meet their business needs based on the data categories, field descriptions, data uses, time ranges, and sample information disclosed in the data asset registration system.

[0003] However, the requirements for data asset integrity often differ significantly across different business scenarios. For example, when the same type of data asset is used in different business scenarios, the data fields, data coverage, temporal continuity, sample size, and degree of data matching may vary. Therefore, evaluating data assets solely based on their category or using uniform data quality indicators is insufficient to accurately reflect their suitability for specific business scenarios. Existing data asset registration and evaluation methods typically lack mechanisms to link specific business scenarios with corresponding integrity assessment indicators, evaluation thresholds, and indicator weights, making it easy for the evaluation results of different data assets to deviate from the actual business needs of the procuring entity.

[0004] On the other hand, to improve the accuracy of data asset integrity assessments, it is usually necessary to obtain more detailed data content or directly access the raw data held by the data provider. However, the raw data may contain commercially valuable or security-critical data. Directly transmitting the raw data to data trading platforms or third-party assessment equipment for processing increases the risks of data leakage, unauthorized access, and data leaving the domain. Relying solely on the limited information proactively disclosed by the data provider during the registration process may result in insufficient statistical information or the omission of certain indicators, making it difficult to arrive at accurate and consistent integrity assessment results.

[0005] Furthermore, existing data evaluation results are typically expressed using technical indicators such as field missing rate, data coverage, and data continuity, or simple data quality grades. These results primarily reflect the technical state of the data itself, but are insufficient to explain the potential impact of corresponding data defects on specific business applications. For example, when a key indicator fails to meet business requirements, the resulting decline in model performance, insufficient business coverage, or data usage risks usually still require the purchasing party to rely on manual experience to make judgments. This not only increases the data purchasing party's analysis costs but also makes it difficult to provide direct evidence for subsequent data supplementation, supplier negotiation, and other risk management.

[0006] Therefore, how to automatically establish data asset integrity evaluation standards that are adapted to the specific application scenarios of the purchaser, obtain data characteristics that can be used for quantitative evaluation without avoiding the external transmission of the supplier's original data, and further transform the integrity defects of data assets into quantitative risk information and risk mitigation solutions related to the business scenario, has become a technical problem that needs to be solved in the process of data asset trading and evaluation. Summary of the Invention

[0007] This application proposes a data asset integrity assessment and risk warning method for business scenarios to solve one of the aforementioned technical problems.

[0008] The technical solution adopted in this application is as follows: In a first aspect, embodiments of this application provide a method for data asset integrity assessment and risk warning in a business scenario, applicable to a data asset transaction environment including a purchaser terminal, a data asset registration system, an integrity assessment engine, and a local summary computing node on the supplier side. The method includes: S101, the integrity assessment engine receives a data query request message sent by the purchaser's terminal, parses the target data asset category, application scenario identifier, and scenario requirement parameter set from the data query request message, and obtains a candidate data asset list from the data asset registration system according to the target data asset category; S102, using the application scenario identifier as an index, determine the corresponding integrity assessment index set and the requirement threshold, minimum tolerance threshold and weight corresponding to each integrity assessment index from the preset scenario-integrity index mapping library, wherein the integrity assessment index set includes core indicators and auxiliary indicators. S103, for each candidate data asset in the candidate data asset list, a statistical summary collection request is sent to the corresponding local summary calculation node on the supplier side, and a standardized feature statistical summary generated by the local summary calculation node on the supplier side after statistical calculation based on the supplier's local original data is received. The supplier's local original data is not sent to the integrity assessment engine. The integrity assessment engine marks the integrity assessment indicators of the standardized feature statistical summary that have not reported statistical values ​​as undisclosed. S104. Based on the statistical measured values, demand thresholds, and minimum tolerance thresholds of each integrity assessment indicator expressed in the form of positive indicators in the standardized feature statistical summary, the individual indicator scores corresponding to each integrity assessment indicator are determined according to the segmented scoring rules, and the individual indicator scores are weighted and summed according to the weights corresponding to each integrity assessment indicator to obtain the overall integrity score of the candidate data asset. S105. Based on the overall integrity score and the compliance status of the core indicators, determine the risk level corresponding to the candidate data assets. Based on the integrity assessment indicators whose statistical measured values ​​do not reach the corresponding demand threshold or are in an undisclosed state, determine the indicator defect type. Based on the indicator defect type, search in the preset knowledge graph of missing impacts to determine the corresponding quantitative business negative impact, and generate a risk mitigation plan based on the indicator defect type. S106, the overall integrity score, risk level, quantified negative business impact and risk mitigation plan are assembled into a structured risk report, and the structured risk report is sent to the purchaser's terminal for display.

[0009] Preferably, in step S101, the set of scenario requirement parameters includes filtering condition parameters and threshold constraint parameters; The filtering condition parameters are sent to the data asset registration system as the initial screening conditions for candidate data assets. The data asset registration system then removes asset records that do not meet the filtering condition parameters from the asset records that match the data asset category, thus obtaining the candidate data asset list. In step S102, based on the mapping relationship between the threshold constraint parameter and the integrity evaluation index, the integrity evaluation index corresponding to the threshold constraint parameter is determined, and the required threshold of the corresponding integrity evaluation index is determined based on the threshold constraint parameter. The determined required threshold is written into the corresponding required threshold field to replace the default required threshold preset in the scenario-integrity index mapping library for the current application scenario.

[0010] Preferably, determining the weights corresponding to each integrity assessment index in step S102 includes: Obtain the default weights corresponding to each integrity assessment indicator from the scenario-integrity indicator mapping library; When the data query request message does not contain weight fine-tuning parameters, the default weight is determined as the weight of the corresponding integrity assessment index; When the data query request message contains a weight fine-tuning parameter, the default weight of the corresponding integrity assessment index is adjusted according to the weight fine-tuning parameter to form a weight to be normalized, and the sum of the weights to be normalized of all integrity assessment indices is calculated. The ratio of the unnormalized weight of each integrity assessment indicator to the sum of the unnormalized weights is determined as the normalized weight of the corresponding integrity assessment indicator. Calculate the sum of the normalized weights of the core indicators, and determine whether the sum of the normalized weights of the core indicators is lower than the preset minimum weight ratio threshold of the core indicators. At the same time, determine whether the normalized weight of any auxiliary indicator is higher than the lowest value among the normalized weights of the core indicators. When the sum of the normalized weights of the core indicators is not lower than the minimum weight percentage threshold of the core indicators, and the normalized weights of each auxiliary indicator are not higher than the lowest value among the normalized weights of each core indicator, the normalized weights are determined as the weights of the corresponding integrity assessment indicators. Otherwise, discard the weight fine-tuning parameters, use the corresponding default weights in the scenario-integrity index mapping library to perform subsequent overall integrity score calculations, and return a weight parameter constraint exception prompt to the purchasing terminal.

[0011] Preferably, in step S103, the standardized feature statistical summary received by the integrity assessment engine is generated by the local summary calculation node on the supplier side through the following steps: When data assets are put on the shelf, statistical calculations are performed on the full original dataset of the supplier's local machine according to a preset unified template to generate an initial statistical summary corresponding to the general statistical dimension, and the initial statistical summary is stored as baseline data. Upon receiving the statistical summary collection request, the system obtains the filtering conditional parameters determined by the integrity assessment engine based on the scenario requirement parameter set from the statistical summary collection request. For the conditional statistical dimensions related to the filtering conditional parameters, the system performs conditional filtering and statistical recalculation on the original data locally based on the filtering conditional parameters to obtain conditional adaptive statistical values. The condition-adaptive statistical values ​​are merged with the baseline data to form a standardized feature statistical summary corresponding to the current candidate data asset; For a negative indicator where a decrease in value indicates improved integrity, a positive indicator value corresponding to the negative indicator is calculated when generating the standardized feature statistical summary, and the positive indicator value replaces the negative indicator value in the standardized feature statistical summary.

[0012] Preferably, in step S103, marking the integrity assessment indicators of unreported statistical values ​​as undisclosed includes: The local summary calculation node on the supplier side constructs index slots that correspond one-to-one with the integrity assessment index set determined in step S102 according to a preset fixed template. Each index slot includes at least an index unique identifier and an index statistical value field. For integrity assessment indicators that have not been calculated or reported, retain the corresponding indicator slot and the indicator's unique identifier, and write a preset undisclosed null value marker into the corresponding indicator statistical value field; The integrity assessment engine parses the standardized feature statistical summary, and when it identifies the undisclosed null value marker, it determines the integrity assessment indicator based on the corresponding indicator unique identifier and assigns an undisclosed status label to the integrity assessment indicator. When cleaning the standardized feature statistical summary, abnormal data in the reported indicator statistical values ​​are cleaned, and the undisclosed null value markers and corresponding undisclosed status labels with undisclosed status semantics are retained.

[0013] Preferably, in step S104, if the demand threshold corresponding to each integrity assessment indicator is greater than the corresponding minimum tolerance threshold, the individual indicator score corresponding to each integrity assessment indicator is determined according to the segmented scoring rules, including: When the statistical measured value of the integrity assessment indicator is greater than or equal to the corresponding demand threshold, the single indicator score of the integrity assessment indicator is determined as the full score of the single indicator. When the statistical measured value of the integrity assessment indicator is less than the corresponding minimum tolerance threshold, or when the integrity assessment indicator is in an undisclosed state, the single indicator score of the integrity assessment indicator shall be determined as zero. When the statistical measured value of the integrity assessment indicator is greater than or equal to the corresponding minimum tolerance threshold and less than the corresponding demand threshold, calculate the first difference between the statistical measured value and the minimum tolerance threshold and the second difference between the demand threshold and the minimum tolerance threshold, and multiply the ratio of the first difference to the second difference by the full score of the single indicator to obtain the single indicator score of the integrity assessment indicator. The overall integrity score is determined by summing the products of the weights corresponding to each integrity assessment indicator and the scores of the corresponding individual indicators.

[0014] Preferably, step S105, which determines the risk level and quantifies the negative impact on business, includes: The basic risk level is determined based on the preset scoring range in which the overall integrity score falls, and the compliance status of each core indicator is detected. When at least one core indicator has a statistically measured value that is lower than the corresponding demand threshold or is not disclosed, and the basic risk level is low risk, the basic risk level will be adjusted to medium risk; when the basic risk level is medium risk or high risk, the basic risk level will remain unchanged. From the knowledge graph of missing impacts, which is established according to five dimensions of indicator defects, business scenarios, risk types, quantitative impacts and mitigation strategies, risk information matching the current indicator defect is retrieved based on the indicator defect type and the application scenario identifier, and the corresponding quantitative impact benchmark interval is read. For an integrity assessment index that has a statistically measured value and the statistically measured value is lower than the corresponding demand threshold, the deviation between the statistically measured value and the demand threshold is calculated, and a quantitative impact sub-interval is determined from the corresponding quantitative impact benchmark interval based on the deviation. The larger the deviation, the closer the quantitative impact sub-interval is to the upper limit of the quantitative impact benchmark interval. The quantitative negative impact on business is generated based on the retrieved risk type and the corresponding quantitative impact benchmark range or quantitative impact sub-range.

[0015] Preferably, in step S105, the risk mitigation plan includes data supplementation recommendations, and the generation of the data supplementation recommendations includes: Extract unique identifiers from integrity assessment indicators whose statistically measured values ​​are below the corresponding demand thresholds or are in an undisclosed state to form a set of defect indicators; Based on the target data asset category and the set of defect indicators, a search is initiated with the data asset registration system to obtain candidate supplementary assets that are the same as the target data asset category and whose statistical measured value in at least one dimension of the defect indicator is greater than or equal to the corresponding demand threshold. For each candidate supplementary asset, a basic fit score is determined based on the ratio of the number of defect indicators that the candidate supplementary asset can compensate for to the total number of defect indicators in the defect indicator set. When at least one defect indicator in the candidate supplementary asset has a statistically measured value that exceeds the corresponding demand threshold by a preset excess range, a preset bonus is added to the basic fit score to obtain the fit score. When there are no defect indicators whose statistically measured values ​​exceed the corresponding demand threshold by the preset excess range, the basic fit score is determined as the fit score. The candidate supplementary assets are sorted according to their suitability scores, and a data supplement recommendation list is generated by selecting the candidate supplementary assets that meet the preset ranking conditions.

[0016] Preferably, in step S105, the risk mitigation plan includes data supplementation recommendations and supplier negotiation suggestions; in step S106, the structured risk report is displayed, including: Based on the completeness assessment indicators that are not disclosed or whose statistical measured values ​​are lower than the corresponding demand threshold, the statistical indicators and verification parameters that need to be supplemented or verified by the supplier are determined, and the supplier negotiation suggestions are formed. The supplier negotiation suggestions include supplementary information requirements and re-evaluation of interface information. The structured risk report is presented in three layers: overview, details, and solutions. The overview layer displays the overall integrity score and risk level. The details layer displays the statistical measured values, individual index scores, weight contributions, and compliance status of each integrity assessment indicator. The solutions layer displays the quantified negative impact on business and risk mitigation solutions. The system provides differentiated prompts for core indicators that do not meet the demand threshold, marks undisclosed integrity assessment indicators that are not yet disclosed, and provides data supplement recommendation jumps corresponding to the data supplement recommendations, as well as supplier information supplement interaction entries corresponding to the supplier negotiation suggestions.

[0017] Secondly, embodiments of this application provide a data asset integrity assessment and risk warning system for business scenarios, deployed on a data trading platform, including: The requirement parsing and retrieval module is used to receive data query request messages sent by the purchasing terminal, parse the target data asset category, application scenario identifier and scenario requirement parameter set from the data query request message, and obtain a candidate data asset list from the data asset registration system according to the target data asset category. The indicator matching and parameter initialization module is used to determine the corresponding integrity assessment indicator set and the requirement threshold, minimum tolerance threshold and weight of each integrity assessment indicator from the preset scenario-integrity indicator mapping library, using the application scenario identifier as an index. The integrity assessment indicator set includes core indicators and auxiliary indicators. The summary acquisition and status marking module is used to send a statistical summary acquisition request to the corresponding local summary calculation node on the supplier side for each candidate data asset in the candidate data asset list, and to receive a standardized feature statistical summary generated by the local summary calculation node on the supplier side after statistical calculation based on the supplier's local original data, wherein the supplier's local original data is not sent to the system; it is also used to mark the integrity assessment indicators of unreported statistical values ​​in the standardized feature statistical summary as undisclosed status; The quantitative scoring module is used to determine the individual index score corresponding to each integrity assessment index according to the statistical measured value, demand threshold and minimum tolerance threshold of each integrity assessment index expressed in the form of positive indicators in the standardized feature statistical summary, and to perform weighted summation of the individual index scores according to the weights corresponding to each integrity assessment index to obtain the overall integrity score of the candidate data asset. The risk reasoning module is used to determine the risk level of candidate data assets based on the overall integrity score and the compliance status of core indicators, determine the indicator defect type based on the integrity assessment indicators whose statistical measured values ​​have not reached the corresponding demand threshold or are in an undisclosed state, search in the pre-set knowledge graph of missing impact based on the indicator defect type, determine the corresponding quantitative business negative impact, and generate a risk mitigation plan based on the indicator defect type. The visualization output module is used to assemble the overall integrity score, risk level, quantified negative business impact and risk mitigation plan into a structured risk report, and send the structured risk report to the purchaser's terminal for display.

[0018] This invention analyzes the data query requests from the purchaser and determines the corresponding integrity assessment indicators, requirement thresholds, minimum tolerance thresholds, and weights from the scenario-integrity indicator mapping library based on application scenario identifiers. This allows the data asset integrity evaluation standards to match specific business scenarios, improving the scenario adaptability of the evaluation results and the comparability between different candidate data assets. Simultaneously, by generating standardized feature statistical summaries based on the original data through the supplier's local summary calculation node, and sending only the statistical summaries to the integrity assessment engine, the integrity evaluation can be completed without the original data leaving the supplier's side. This ensures the validity of the evaluation basis while reducing the security risks associated with cross-domain transmission of original data. Furthermore, by combining statistically measured values, requirement thresholds, and minimum tolerance thresholds for segmented scoring, and calculating the overall integrity score according to indicator weights, a quantitative evaluation of the data asset integrity status can be achieved. Combining the overall integrity score and the compliance status of core indicators to determine the risk level can reduce the possibility that defects in key indicators are masked by the overall score. Furthermore, by identifying the types of defects in non-compliant or undisclosed indicators, and using the knowledge graph of missing impacts to determine the corresponding quantitative negative business impacts and risk mitigation solutions, data integrity defects can be transformed into risk warnings and handling suggestions for specific business scenarios, thereby improving the automation and effectiveness of data asset procurement evaluation and risk decision-making. Attached Figure Description

[0019] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart illustrating a data asset integrity assessment and risk warning method for business scenarios, provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a data asset integrity assessment and risk warning system for business scenarios provided in an embodiment of the present invention. Detailed Implementation

[0020] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] like Figure 1 As shown in some embodiments of this application, this embodiment provides a data asset integrity assessment and risk warning method for business scenarios, applied in a data asset transaction environment including a purchaser terminal, a data asset registration system, an integrity assessment engine, and a supplier-side local summary computing node. Specifically, the method includes the following steps: Step S101: The integrity assessment engine receives a data query request message sent by the purchaser's terminal, parses the target data asset category, application scenario identifier, and scenario requirement parameter set from the data query request message, and obtains a candidate data asset list from the data asset registration system according to the target data asset category.

[0022] Specifically, step S101 is mainly used to receive the query request submitted by the purchaser regarding data assets, convert the query content submitted by the purchaser into structured parameters that can be identified and processed by the subsequent integrity assessment process, and determine the candidate data assets to be assessed based on the type of data assets the purchaser wishes to acquire. Specifically, the purchaser's terminal can be a computer terminal connected to the data trading platform, a business system terminal, or other electronic device capable of submitting data asset query requests. When the purchaser needs to query or purchase data assets, it sends a data query request message to the integrity assessment engine through its purchaser terminal. The data query request message carries the basic requirement information needed for this data asset query and assessment, and it includes at least the target data asset category, application scenario identifier, and scenario requirement parameter set.

[0023] The target data asset category indicates the category of the data asset the purchaser wishes to query, such as consumer profile data, financial risk control data, and autonomous driving road test data. The application scenario identifier uniquely identifies the business scenario in which the purchaser intends to apply the data asset, enabling subsequent processing to determine the appropriate completeness assessment criteria for that business scenario. The scenario requirement parameter set describes the specific data requirements proposed by the purchaser for the current application scenario, and may include parameters related to business needs such as data scope, time range, sample size, and data matching requirements.

[0024] After receiving a data query request message, the integrity assessment engine parses the message according to a preset message structure or field definition, extracting the target data asset category, application scenario identifier, and scenario requirement parameter set, and forming structured requirement information corresponding to the current assessment task. By structuring the purchaser's query requirements, parameter ambiguity caused by natural language descriptions or non-standardized input can be reduced, providing a unified data input foundation for subsequent scenario indicator matching, statistical summary collection, and integrity scoring.

[0025] After parsing the query request, the integrity assessment engine initiates a query to the data asset registration system based on the target data asset category. The data asset registration system pre-stores registration information for registered data assets, including asset category, asset identifier, asset name, and supplier information. The data asset registration system searches for matching data asset records based on the target data asset category and returns the matched data assets as candidate data assets to the integrity assessment engine, which then uses this list to generate a candidate data asset list.

[0026] For example, suppose a purchaser plans to acquire consumer profile data and use it for targeted marketing of financial products. The purchaser submits a data query request to the integrity assessment engine through a terminal. The target data asset category is "consumer profile data," the application scenario identifier corresponds to "targeted marketing of financial products," and the scenario requirement parameter set may include business requirement information such as the target population's age range, data time range, device compatibility requirements, and sample size requirements.

[0027] After receiving the data query request, the integrity assessment engine parses the information carried in the request to obtain the target data asset category of "consumer profile data", the application scenario identifier of "precision marketing of financial products" and the corresponding scenario requirement parameter set, and associates the above information with the current assessment task.

[0028] Subsequently, the integrity assessment engine accesses the data asset registration system using "consumer profile data" as the query condition. The data asset registration system searches for data asset records whose categories match "consumer profile data" from the already registered data assets, and returns the corresponding data asset identifier, asset name, and supplier information to the integrity assessment engine.

[0029] Assuming the data asset registration system contains multiple consumer profile-type data assets, the integrity assessment engine can create a candidate data asset list from these data assets that meet the target data asset category. Then, for each candidate data asset in the candidate data asset list, steps S102 to S106 are executed to determine the integrity score and risk information according to the assessment requirements corresponding to the current "precision marketing of financial products" scenario, thereby providing the purchaser with data asset assessment results that can be used for comparison and selection.

[0030] Step S102: Using the application scenario identifier as an index, determine the corresponding integrity assessment index set and the corresponding requirement threshold, minimum tolerance threshold and weight of each integrity assessment index from the preset scenario-integrity index mapping library. The integrity assessment index set includes core indicators and auxiliary indicators.

[0031] Specifically, step S102 is mainly used to determine the integrity assessment indicator system that matches the current business scenario from a pre-built scenario-integrity indicator mapping library based on the application scenario specified by the purchaser, and to initialize the corresponding evaluation parameters for subsequent integrity scoring. Specifically, the scenario-integrity indicator mapping library pre-establishes the correspondence between different application scenarios and integrity assessment indicators. For each application scenario, the mapping library can associate multiple integrity assessment indicators, and for each integrity assessment indicator, it records parameters such as the corresponding requirement threshold, minimum tolerance threshold, weight, and indicator type. The application scenario identifier can serve as an index for the corresponding indicator set for different scenarios, enabling the integrity assessment engine to quickly locate the assessment indicators and evaluation parameters corresponding to the current business scenario from the mapping library.

[0032] After parsing the application scenario identifier in step S101, the integrity assessment engine uses this identifier as a search condition to read the set of integrity assessment indicators corresponding to the current application scenario from the scenario-integrity indicator mapping library. Different application scenarios can be configured with different combinations of indicators. Therefore, even if the data assets to be assessed are of the same category, different integrity assessment standards can be used when the business purposes of the purchaser are different, so that the subsequent evaluation results can reflect the degree of adaptability of the data assets to the specific business scenario.

[0033] Each integrity assessment indicator has a corresponding demand threshold and a minimum tolerance threshold. The demand threshold represents the target standard for the indicator to meet expected business requirements in the current application scenario; the minimum tolerance threshold represents the lowest standard that the indicator can be accepted by the current business scenario. For integrity assessment indicators evaluated using a segmented scoring rule, the demand threshold is higher than the corresponding minimum tolerance threshold, forming a linear conversion interval between the demand threshold and the minimum tolerance threshold corresponding to the partially compliant state. By setting two different levels of thresholds for the same indicator, different states such as fully meeting business requirements, partially meeting business requirements, and falling below the acceptable business level can be distinguished, providing a parameter basis for subsequent segmented quantitative scoring.

[0034] Weights are used to characterize the relative importance of each integrity assessment metric in the current application scenario. Metrics with a significant impact on business performance can be assigned higher weights, while metrics that primarily play a supporting optimization role can be assigned relatively lower weights. When calculating the overall integrity score, the integrity assessment engine can use the weights of each metric to weight the scores of the corresponding individual metrics, ensuring that the overall integrity score reflects the differences in the importance of different metrics to the specific application scenario.

[0035] Furthermore, the integrity assessment indicator set includes core indicators and auxiliary indicators. Core indicators mainly represent evaluation metrics that have a significant impact on whether the current business scenario can be implemented normally or whether the main business objectives can be achieved; auxiliary indicators are mainly used to reflect the performance of data assets in terms of granularity, coverage, or other auxiliary quality dimensions. By distinguishing between core and auxiliary indicators, not only can the importance of different indicators be reflected in the overall scoring, but it can also provide a basis for subsequent risk level determination based on the compliance status of core indicators.

[0036] For example, suppose the data query request submitted by the purchaser in step S101 indicates that the target data asset category is "consumer profile data" and the application scenario is "precision marketing of financial products." After obtaining the corresponding application scenario identifier, the integrity assessment engine uses this application scenario identifier as an index to access the scenario-integrity indicator mapping library. Suppose the mapping library pre-configures five integrity assessment indicators for the "precision marketing of financial products" scenario, among which the monthly income field completion rate, the consumption category tag coverage completion rate, and the continuous completion rate of transaction records in the past 90 days are classified as core indicators; the device ID stability rate and the target population sample coverage multiple are classified as auxiliary indicators.

[0037] For each metric, the mapping library also pre-configures corresponding demand thresholds, minimum tolerance thresholds, and weights. For example, for the monthly income field completion rate, a corresponding demand threshold can be set according to the requirements of the financial product precision marketing business for the completeness of income information, and a minimum tolerance threshold below this demand threshold can be further set; for the target population sample coverage multiple, the corresponding demand threshold and minimum tolerance threshold can be set according to the current business requirements for sample size.

[0038] After the integrity assessment engine completes the matching, it obtains the five assessment indicators and their evaluation parameters corresponding to the current "precision marketing of financial products" scenario. When collecting statistical summaries for candidate data assets, the engine can obtain corresponding statistical measured values ​​based on the above indicators. In step S104, the statistical measured values ​​are compared with the demand threshold and the minimum tolerance threshold, and then combined with the corresponding weights to form an overall integrity score.

[0039] Step S103: For each candidate data asset in the candidate data asset list, send a statistical summary collection request to the corresponding local summary calculation node on the supplier side, and receive the standardized feature statistical summary generated by the local summary calculation node on the supplier side after performing statistical calculations based on the supplier's local original data. The supplier's local original data is not sent to the integrity assessment engine. The integrity assessment engine marks the integrity assessment indicators of the standardized feature statistical summary that have not reported statistical values ​​as undisclosed.

[0040] Specifically, step S103 is mainly used to acquire statistical characteristics that reflect the integrity status of the candidate data assets determined in step S101, and to provide unified data input for subsequent quantitative scoring when the original data from the supplier is not transmitted to the integrity assessment engine. Specifically, the candidate data asset list may contain one or more data assets provided by different data suppliers. For each candidate data asset in the list, the integrity assessment engine determines the corresponding local summary calculation node on the supplier side based on the supplier information corresponding to the candidate data asset, and sends a statistical summary collection request to that local summary calculation node. The statistical summary collection request may carry the candidate data asset identifier corresponding to the current assessment task and the unique identifier of the indicator corresponding to the integrity assessment indicator determined in step S102, enabling the local summary calculation node on the supplier side to determine the indicator that needs to be statistically analyzed and to construct indicator slots corresponding one-to-one with the corresponding integrity assessment indicators according to a preset template.

[0041] The local summary calculation node on the supplier side is deployed on the data supplier's side. It can access the original data corresponding to the candidate data assets locally and perform statistical calculations based on the original data. The calculated results are organized into a standardized feature statistical summary according to a predetermined data organization method, and then the statistical summary is sent to the integrity assessment engine. The standardized feature statistical summary contains statistical results corresponding to the current integrity assessment indicators, which are used to reflect the actual status of the candidate data assets in terms of field completeness, temporal continuity, data coverage, or other evaluation dimensions.

[0042] Throughout the process described above, the original data remains on the data provider's side. The local summary computing node on the provider's side provides the integrity assessment engine with statistical characteristics calculated based on the original data, rather than the specific original data records that constitute the candidate data assets. Therefore, the integrity assessment engine does not need to directly acquire or centrally store the provider's original data; it can use the statistical results corresponding to the actual data to perform subsequent integrity evaluations, thereby reducing the risk of data leakage during the transmission of original data across security domains.

[0043] After receiving the standardized feature statistical summary, the integrity assessment engine matches the statistical results with the integrity assessment indicators determined in step S102. For integrity assessment indicators for which statistical values ​​can be obtained, the corresponding statistical values ​​are used as the measured statistical values ​​of the indicators; for integrity assessment indicators for which the supplier has not provided statistical values, the indicators are marked as undisclosed.

[0044] The "undisclosed" status indicates that the current integrity assessment task cannot obtain valid statistical results for this indicator from the supplier, rather than ignoring the indicator altogether. By retaining undisclosed indicators and their status information, they can continue to participate in subsequent scoring and risk assessment, preventing the data integrity risk from being overlooked during the assessment process due to the supplier's failure to provide certain statistical information.

[0045] For example, suppose the candidate data asset list determined in step S101 includes "Consumer Profile Dataset V3" provided by a data provider. The completeness evaluation indicators determined in step S102 for the "precision marketing of financial products" scenario include the monthly income field filling completeness rate, the consumption category label coverage completeness rate, the continuous completeness rate of transaction records in the past 90 days, the device ID stability rate, and the target population sample coverage multiple.

[0046] Based on the supplier information corresponding to the "Consumer Profile Dataset V3," the integrity assessment engine sends a statistical summary collection request to the local summary calculation node deployed on the supplier's side. The local summary calculation node reads the raw data corresponding to the data asset locally on the supplier's side and performs statistical calculations according to the integrity indicators to be evaluated. After the calculation is completed, only the obtained statistical results are organized into a standardized feature statistical summary and sent to the integrity assessment engine, while the original data such as specific consumer records and transaction details remain on the supplier's local side and are not sent with the statistical summary.

[0047] For example, the statistical results returned by the local summary calculation node may include: a monthly income field completion rate of 94%, a continuous completion rate of transaction records in the past 90 days of 88%, a device ID stability rate of 82%, and a target population sample coverage multiple of 1.5; the supplier did not provide corresponding statistical values ​​for the consumer category tag coverage completion rate.

[0048] After receiving the above statistical summary, the integrity assessment engine associates each statistical value with its corresponding integrity assessment metric, and uses the values ​​corresponding to the monthly income field completion rate, the continuous completeness rate of transaction records in the past 90 days, the device ID stability rate, and the target population sample coverage multiple as their respective measured statistical values. Since no statistical value was obtained for the consumer category tag coverage completeness rate, this metric is marked as undisclosed.

[0049] At this point, a completeness assessment input for the "Consumer Profile Dataset V3" has been formed for subsequent processing. This input includes both the statistically measured values ​​that have been obtained and the status of indicators for which no statistical results have been provided. Subsequently, in step S104, a quantitative completeness score can be performed on the candidate data asset based on the aforementioned statistically measured values, undisclosed statuses, and the demand threshold, minimum tolerance threshold, and weights determined in step S102.

[0050] Step S104: Based on the statistical measured values, demand thresholds, and minimum tolerance thresholds of each integrity assessment indicator expressed in the form of positive indicators in the standardized feature statistical summary, determine the individual indicator scores corresponding to each integrity assessment indicator according to the segmented scoring rules, and perform weighted summation of the individual indicator scores according to the weights corresponding to each integrity assessment indicator to obtain the overall integrity score of the candidate data asset.

[0051] Specifically, step S104 is mainly used to convert the statistical measured values ​​obtained in step S103 into comparable single-indicator scores, and further combine the importance of each indicator to form an overall integrity score for the candidate data assets. Specifically, all integrity assessment indicators entering this step are represented in a positive indicator form, meaning that the larger the statistical measured value of the indicator, the higher the integrity level of the corresponding dimension. For inverse indicators that originally indicated improved data quality by a decrease in value, they can be pre-converted into corresponding positive indicators when generating standardized feature statistical summaries. Therefore, the integrity assessment engine does not need to set different scoring logics for different indicator directions when executing this step.

[0052] For each integrity assessment metric, the integrity assessment engine obtains its statistical measured value and reads the demand threshold, minimum tolerance threshold, and weight determined in step S102. The demand threshold represents the target standard corresponding to the metric meeting the expected requirements of the current business scenario, and the minimum tolerance threshold represents the lowest acceptable standard for the current business scenario. The integrity assessment engine compares the statistical measured value with the two thresholds to divide the metric into corresponding scoring intervals and determines the score for each metric according to the corresponding segmented scoring rules.

[0053] Therefore, when a certain indicator has met the target requirements of the current business scenario, a high score can be obtained for that single indicator. When an indicator has not yet met the target requirements but is still within the acceptable range for the business, the corresponding score is determined based on the degree of difference between it and the target standard. When an indicator is below the minimum acceptable level or is in an undisclosed state, the integrity deficiency of that indicator is reflected in the scoring results. Through the above processing, the binary evaluation method of "meeting or not meeting the standard" can be avoided, and data states that partially meet business needs can also be expressed in a differentiated quantitative manner.

[0054] After obtaining the individual scores for each integrity assessment metric, the integrity assessment engine calculates the contribution of each metric to the overall evaluation result based on its corresponding weight, and sums the weighted results of each metric to obtain the overall integrity score of the current candidate data asset. The weights reflect the importance of different metrics in the current application scenario; therefore, metrics with a greater impact on business results can have a larger impact on the overall score, while relatively less important metrics have a smaller impact.

[0055] Through the above scoring process, statistical characteristics of different dimensions such as field completeness, temporal continuity, and data coverage can be uniformly converted into comparable quantitative results, and further form an overall score that can comprehensively reflect the completeness of the candidate data assets' adaptation to the current application scenario, providing quantitative input for risk level determination and business risk analysis in step S105.

[0056] For example, assuming the aforementioned "precision marketing of financial products" scenario, for the candidate data asset "consumer profile dataset V3", step S103 obtains the statistical results corresponding to the integrity assessment indicators such as the monthly income field filling completeness rate, the consumption category label coverage completeness rate, the continuous completeness rate of transaction records in the past 90 days, the device ID stability rate, and the target population sample coverage multiple.

[0057] The integrity assessment engine reads the required threshold, minimum tolerance threshold, and weight configured for the above indicators in step S102, and compares the measured statistical values ​​of each indicator with the corresponding thresholds. For indicators that have reached the required threshold, a higher score can be obtained according to the corresponding scoring rules; for indicators that are below the required threshold but still above the minimum tolerance threshold, a corresponding converted score is obtained based on the actual degree of achievement; for indicators whose statistical values ​​are not disclosed, they are processed according to the scoring rules corresponding to the undisclosed status.

[0058] For example, the statistically measured completion rate of the monthly income field is 94%, the demand threshold is 95%, and the minimum tolerance threshold is 80%. Therefore, this indicator falls between the demand threshold and the minimum tolerance threshold. The continuous completeness rate of transaction records in the past 90 days and the stability rate of device IDs both reach their corresponding demand thresholds. The target population sample coverage multiple is 1.5 times, which is lower than the demand threshold of 2.0 times but higher than the minimum tolerance threshold of 1.0 times. The completeness rate of consumer category tag coverage is not disclosed because it has not been reported. The completeness assessment engine obtains the corresponding single indicator score based on the interval in which each indicator is located, then calculates the weight contribution by combining the weights of each indicator, and finally adds up the weight contributions.

[0059] Step S105: Determine the risk level of the candidate data asset based on the overall integrity score and the compliance status of the core indicators; determine the indicator defect type based on the integrity assessment indicators whose statistical measured values ​​have not reached the corresponding demand threshold or are in an undisclosed state; search the pre-set knowledge graph of missing impacts based on the indicator defect type to determine the corresponding quantitative negative business impact; and generate a risk mitigation plan based on the indicator defect type.

[0060] Specifically, step S105 is mainly used to further analyze the risk of the integrity evaluation results obtained in step S104. It transforms the technical defects in the integrity of candidate data assets into business risk information related to the current application scenario and generates corresponding mitigation solutions for the identified risks. Specifically, the integrity assessment engine first determines the corresponding risk level based on the overall integrity score of the candidate data assets and the compliance status of core indicators. The overall integrity score reflects the comprehensive performance of the candidate data assets across multiple integrity assessment dimensions, while the compliance status of core indicators reflects whether key data conditions that have a significant impact on the current application scenario are met. By considering both types of information simultaneously, it avoids situations where high scores in other indicators mask obvious defects in individual key indicators when risk judgment is based solely on the overall score.

[0061] For example, when the overall integrity score is high, but a certain core indicator fails to meet the corresponding threshold or is not disclosed, the risk assessment result can be revised based on the status of the core indicator. This ensures that the final risk level more accurately reflects the actual usage risk of the candidate data asset in the current business scenario. The risk level can be used to represent the degree of risk that may exist when the candidate data asset meets the current business requirements, and serves as the basis for subsequent risk analysis and the generation of risk mitigation solutions.

[0062] While determining the risk level, the integrity assessment engine further identifies deficient indicators from each integrity assessment metric. For indicators whose measured statistical values ​​are below the corresponding requirement threshold, the corresponding indicator defect type can be determined based on the evaluation dimension of that indicator; for indicators that are not disclosed, the corresponding information missing type defect can be determined based on the missing statistical information. This forms a set of indicator defects corresponding to the current candidate data assets.

[0063] Subsequently, the integrity assessment engine uses the type of indicator defect as the primary search condition and, combined with the current application scenario, performs a matching search within a pre-built knowledge graph of missing data impacts. This knowledge graph establishes the correlation between data integrity defects and business risks, enabling different types of data defects to be mapped to their potential negative business impacts in specific business scenarios. For example, insufficient field completion may affect the quality of subsequent model inputs, insufficient temporal continuity may affect time series analysis results, and the lack of disclosure of key statistical information may prevent the purchasing party from effectively verifying the corresponding data quality.

[0064] The above search can yield quantitative negative business impacts corresponding to the current indicator deficiencies. These negative impacts can be represented by factors such as the range of model performance changes, the probability of business failure, the range of cost losses, or other information that can quantify the business consequences. This transforms what were originally just technical data issues like "indicators not meeting standards" or "indicators not being disclosed" into business risk information that the purchaser can directly understand.

[0065] Furthermore, the integrity assessment engine determines risk mitigation plans based on the identified types of indicator defects and corresponding risk information. These risk mitigation plans provide the purchaser with directions to reduce current data procurement risks. For example, they may offer directions for data supplementation in cases of insufficient integrity dimensions, or provide suggestions for supplementary disclosure or further verification in cases where important indicators are not disclosed.

[0066] Therefore, step S105 does not simply output an abstract risk level, but forms a processing chain of "overall integrity evaluation - defect indicator identification - business risk mapping - mitigation solution generation", so that the integrity evaluation results of data assets can be further transformed into risk decision information with business implications.

[0067] For example, assuming a "precision marketing of financial products" application scenario, for the candidate data asset "Consumer Profile Dataset V3", step S104 calculates an overall completeness score of 72. Among the various completeness assessment indicators, the continuous completeness rate of transaction records in the past 90 days and the stability rate of device IDs meet the corresponding requirement thresholds; the completeness rate of the monthly income field and the target population sample coverage multiple are lower than the corresponding requirement thresholds; and the completeness rate of consumer category tag coverage is not disclosed.

[0068] The integrity assessment engine first determines the basic risk level based on the overall integrity score. Since the overall integrity score is 72, falling within the medium-risk range of 50-80, the basic risk level is medium. Further testing of the core indicators' compliance status reveals that while some candidate data assets have core indicators that haven't met the required thresholds or are not yet disclosed, the basic risk level is already medium. Therefore, according to the core indicator status constraint rules, the medium-risk level remains unchanged, resulting in a final risk level of medium.

[0069] Subsequently, the integrity assessment engine identifies the indicator deficiencies in the current candidate data assets. For example, insufficient completion rate of the monthly income field can be identified as a field integrity deficiency defect, undisclosed completion rate of the consumption category label coverage can be identified as a key statistical information undisclosed defect, and insufficient target population sample coverage multiple can be identified as a sample coverage deficiency defect.

[0070] The integrity assessment engine further retrieves corresponding business risk information from the knowledge graph of missing impacts based on the types of defects in the aforementioned indicators and the "precision marketing of financial products" scenario. For example, insufficient completion of the monthly income field may cause a decrease in the predictive performance of the marketing model; failure to disclose the coverage of consumer category tags may result in the inability to effectively verify the accuracy of population classification; and insufficient coverage of the target population sample may increase the risk of overfitting or reduced generalization ability during model training.

[0071] After obtaining the risk level and the negative impact on the business, corresponding risk mitigation directions are matched according to the different types of indicator deficiencies. For example, if some indicators are not performing well, suggestions can be generated to supplement the corresponding data dimensions; if key indicators are not disclosed, suggestions can be generated to require suppliers to supplement relevant statistical information or to further verify the data.

[0072] Step S106: Assemble the overall integrity score, risk level, quantified negative business impact and risk mitigation plan into a structured risk report, and send the structured risk report to the purchaser's terminal for display.

[0073] Specifically, step S106 primarily involves unifying the overall integrity score obtained in step S104 with the risk level, quantified negative business impact, and risk mitigation plan obtained in step S105 to form a structured risk report that can be identified and displayed by the purchasing party's terminal, thereby completing the output of the integrity assessment results to the purchasing party. Specifically, after completing the integrity scoring and risk analysis of the candidate data assets, the integrity assessment engine integrates the results generated at different processing stages according to preset data correlation relationships. The overall integrity score reflects the comprehensive integrity of the candidate data assets for the current application scenario; the risk level indicates the overall risk level when the candidate data asset is applied to the current business scenario; the quantified negative business impact describes the potential impact of non-compliance or undisclosed indicators on specific business operations; and the risk mitigation plan provides directions for subsequent processing of the corresponding indicator deficiencies.

[0074] When assembling the results, each candidate data asset can be used as a unit. The data asset identifier can be linked to its corresponding overall integrity score, risk level, indicator defect information, quantified negative business impact, and risk mitigation plan. A structured risk report can then be generated according to a pre-defined report structure. By organizing the results in a structured manner, the correspondence between different types of assessment information can be maintained, avoiding the separation of scoring results, risk information, and mitigation measures. This allows the purchaser to clearly identify the integrity level, main defects, and potential business consequences of a particular candidate data asset.

[0075] When the candidate data asset list contains multiple candidate data assets, evaluation results corresponding to each candidate data asset can be generated separately, and further summarized based on information such as the overall integrity score, so that the purchaser can make horizontal comparisons of multiple candidate data assets under the same application scenario and evaluation criteria.

[0076] After the structured risk report is generated, the integrity assessment engine sends the report to the purchasing party's terminal. Upon receiving the relevant data, the purchasing party's terminal parses and displays the report content, enabling the purchasing party to view the overall integrity score, risk level, and business risks and mitigation information related to indicator deficiencies.

[0077] For example, assuming that in the aforementioned "precision marketing of financial products" application scenario, for the candidate data asset "consumer profile dataset V3", step S104 obtains an overall completeness score of 72 points, and step S105 further determines that the risk level corresponding to the candidate data asset is medium risk, and identifies defects in indicators such as insufficient completion rate of monthly income field, undisclosed completeness rate of consumption category label coverage, and insufficient target population sample coverage multiple.

[0078] The integrity assessment engine then associates and assembles the asset identifier of the "Consumer Profile Dataset V3", the overall integrity score of 72 points, the medium risk level, the quantitative negative business impact corresponding to the defects of each indicator, and the corresponding risk mitigation solutions according to the preset structure to form a structured risk report corresponding to the candidate data asset.

[0079] After the structured risk report is sent to the purchaser's terminal, the purchaser can understand the overall integrity of the data asset, the main risks, and corresponding handling recommendations based on the displayed results. If the current assessment task also includes other candidate consumer profile data assets, corresponding assessment results can be generated separately in the same way. This allows the purchaser to compare the integrity level and business risks of each candidate data asset within the unified application scenario of "precision marketing of financial products," thereby providing a basis for the final data procurement selection.

[0080] In some embodiments of this application, in order to categorize the business requirements submitted by the purchaser into screening conditions and evaluation thresholds according to their purpose, some parameters are used to narrow down the range of candidate data assets, while others are used to adjust the subsequent integrity evaluation criteria. This avoids uniformly processing the requirements parameters for different purposes and improves the consistency between candidate asset screening and scenario-based integrity assessment. In step S101, the scenario requirement parameter set includes screening condition parameters and threshold constraint parameters. The filtering condition parameters are sent to the data asset registration system as the initial screening conditions for candidate data assets. The data asset registration system then removes asset records that do not meet the filtering condition parameters from the asset records that match the data asset category, thus obtaining the candidate data asset list. In step S102, based on the mapping relationship between the threshold constraint parameter and the integrity evaluation index, the integrity evaluation index corresponding to the threshold constraint parameter is determined, and the required threshold of the corresponding integrity evaluation index is determined based on the threshold constraint parameter. The determined required threshold is written into the corresponding required threshold field to replace the default required threshold preset in the scenario-integrity index mapping library for the current application scenario.

[0081] As mentioned above, the parameters in the scenario requirement parameter set are divided into screening condition parameters and threshold constraint parameters based on their role in the evaluation process. Screening condition parameters are mainly used to describe the purchaser's restrictions on the basic scope of candidate data assets, such as the age range of the target population and the time range corresponding to the data. These parameters are not directly used to calculate the integrity score, but are used as filtering conditions when determining candidate data assets.

[0082] After parsing the filtering condition parameters, the integrity assessment engine sends these parameters along with the target data asset category to the data asset registration system. The data asset registration system first determines the asset records matching the target data asset category, then filters these records based on the filtering condition parameters, eliminating data assets that clearly do not meet the purchaser's basic business requirements. The remaining assets form a candidate data asset list. By performing this pre-screening before the integrity assessment, the subsequent statistical summary collection and scoring processing for data assets that clearly do not meet business conditions can be reduced, narrowing the assessment scope.

[0083] Threshold-constrained parameters are primarily used to describe the specific quantitative requirements proposed by the purchaser for the current application scenario, such as the lower limit of equipment matching rate and the lower limit of sample size. These parameters are not used to directly exclude candidate data assets, but rather to adjust the required threshold of the integrity assessment indicators in step S102, so that the subsequent scoring criteria can reflect the actual requirements proposed by the purchaser for the current business task.

[0084] In practice, the integrity assessment engine determines the corresponding integrity assessment index for each threshold constraint parameter based on the pre-established mapping relationship between threshold constraint parameters and integrity assessment indices, and then determines the required threshold based on the quantitative relationship between the parameter and the corresponding index. When the threshold constraint parameter and the corresponding integrity assessment index use the same unit of measurement, the required threshold can be directly determined based on the parameter value; when they use different units of measurement, the threshold constraint parameter is converted into the required threshold of the corresponding index according to a preset conversion relationship. After determining the required threshold, it is written into the required threshold field of the corresponding index to replace the original default required threshold in the scenario-integrity index mapping library.

[0085] For example, when the purchaser requires that the equipment ID stability rate be no less than 90%, the requirement can be mapped to the integrity assessment indicator of equipment ID stability rate according to the pre-established indicator correspondence, and the demand threshold of the indicator can be set to 0.90. When the purchaser requires a lower limit for the sample size, the requirement can also be converted into the corresponding sample coverage assessment indicator according to the preset conversion relationship, and the corresponding demand threshold can be determined accordingly.

[0086] Through the above methods, the screening condition parameters are responsible for determining "which data assets are included in the evaluation scope", and the threshold constraint parameters are responsible for determining "what standards are used to evaluate the data assets included in the evaluation scope", so that the selection of candidate data assets and subsequent integrity scoring can be matched with the specific application scenarios and actual business needs of the purchaser.

[0087] In some embodiments of this application, in order to allow the purchaser to adjust the importance of each integrity assessment indicator according to actual business needs, the adjustment results are constrained by the minimum weight ratio of core indicators and the upper limit of the weight of auxiliary indicators. This avoids weakening the dominant role of core indicators in the overall integrity evaluation due to arbitrarily increasing or decreasing the weight of some indicators, thereby ensuring that the overall integrity score can still reflect the important data quality requirements in the current business scenario. Step S102, determining the weight corresponding to each integrity assessment indicator, includes: Obtain the default weights corresponding to each integrity assessment indicator from the scenario-integrity indicator mapping library; When the data query request message does not contain weight fine-tuning parameters, the default weight is determined as the weight of the corresponding integrity assessment index; When the data query request message contains a weight fine-tuning parameter, the default weight of the corresponding integrity assessment index is adjusted according to the weight fine-tuning parameter to form a weight to be normalized, and the sum of the weights to be normalized of all integrity assessment indices is calculated. The ratio of the unnormalized weight of each integrity assessment indicator to the sum of the unnormalized weights is determined as the normalized weight of the corresponding integrity assessment indicator. Calculate the sum of the normalized weights of the core indicators, and determine whether the sum of the normalized weights of the core indicators is lower than the preset minimum weight ratio threshold of the core indicators. At the same time, determine whether the normalized weight of any auxiliary indicator is higher than the lowest value among the normalized weights of the core indicators. When the sum of the normalized weights of the core indicators is not lower than the minimum weight percentage threshold of the core indicators, and the normalized weights of each auxiliary indicator are not higher than the lowest value among the normalized weights of each core indicator, the normalized weights are determined as the weights of the corresponding integrity assessment indicators. Otherwise, discard the weight fine-tuning parameters, use the corresponding default weights in the scenario-integrity index mapping library to perform subsequent overall integrity score calculations, and return a weight parameter constraint exception prompt to the purchasing terminal.

[0088] As described above, the integrity assessment engine first reads the integrity assessment indicators and their default weights corresponding to the current application scenario from the scenario-integrity indicator mapping library based on the application scenario identifier. The default weights represent the pre-defined importance of each indicator within the business scenario. When the purchaser does not request weight adjustments, the default weights can be directly used for subsequent overall integrity scoring. In one implementation, the default weights stored in the scenario-integrity indicator mapping library are normalized weights, and the sum of the default weights of all integrity assessment indicators corresponding to the same application scenario is 1. The pre-configured default weights satisfy the constraints of the minimum weight percentage of core indicators and the upper limit of the weight of individual auxiliary indicators; that is, the sum of the default weights of core indicators is not lower than the minimum weight percentage threshold of core indicators, and the default weights of each auxiliary indicator are not higher than the lowest value among the default weights of each core indicator.

[0089] When a data query request message contains weight fine-tuning parameters, it indicates that the purchaser wishes to adjust the importance of certain integrity assessment indicators to meet specific business needs. The integrity assessment engine increases or decreases the default weights of the corresponding indicators based on the weight fine-tuning parameters. The adjusted indicator weights serve as the weights to be normalized. Since the sum of all adjusted indicator weights may no longer meet the uniform total weight requirement, it is necessary to calculate the sum of all weights to be normalized and then divide the weight of each indicator to be normalized by the sum of these weights to obtain the corresponding normalized weight. This normalization process ensures that the adjusted weights are within a unified proportional system, providing directly usable weight parameters for subsequent weighted scoring.

[0090] After normalization, the adjustment results are further validated by weight constraints. First, the sum of the normalized weights of all core indicators is calculated and compared with the preset minimum weight percentage threshold for core indicators. This threshold is used to ensure that core indicators occupy a sufficiently large weight percentage in the overall evaluation, preventing the purchaser from excessively weakening the importance of core indicators through weight fine-tuning.

[0091] Simultaneously, the normalized weights of each auxiliary indicator are verified by comparing the normalized weight of each auxiliary indicator with the lowest normalized weight among the core indicators. If the weight of any auxiliary indicator is higher than the lowest weight of the core indicator, the current weight adjustment result is considered to violate the pre-set indicator level constraints. This constraint prevents a particular auxiliary indicator from having a greater impact on the overall score than the core indicator, thus maintaining the predetermined importance relationship between the core and auxiliary indicators.

[0092] When the sum of the normalized weights of the core indicators is not lower than the minimum weight percentage threshold of the core indicators, and the normalized weights of all auxiliary indicators are not higher than the minimum normalized weight among the core indicators, it indicates that the current weight adjustment conforms to the weight constraints corresponding to the current application scenario. At this time, the normalized weights of each indicator are determined as the indicator weights actually used in this evaluation, and the overall integrity score calculation in step S104 is performed accordingly.

[0093] Conversely, if any of the above weight constraints are not met, it is considered that this weight adjustment may lead to an imbalance in the distribution of indicator importance. The integrity assessment engine will not use the results of this weight adjustment, will discard the corresponding weight adjustment parameters, and will restore the default weights pre-configured in the scenario-integrity indicator mapping library for subsequent scoring. At the same time, an abnormal weight parameter constraint prompt will be returned to the purchaser's terminal to inform the purchaser that the currently submitted weight adjustment has not passed the verification, so that this assessment can still continue to be executed based on valid weights that meet the scenario constraints, and the assessment process will not be interrupted due to abnormal weight parameters.

[0094] For example, a certain application scenario corresponds to three core metrics and two auxiliary metrics. The scenario-completeness metric mapping library has already configured default weights for each of the five metrics. The purchasing party believes that this business requires more attention to one of the auxiliary metrics, so they increase the weight of that auxiliary metric by fine-tuning the parameters. The completeness assessment engine re-normalizes based on the adjustment result, calculates the sum of the weights of the three core metrics, and compares the weight relationship between the two auxiliary metrics and the lowest core metric.

[0095] If the sum of the weights of the normalized core indicators still meets the preset minimum percentage requirement, and the normalized weights of the increased auxiliary indicators are still not higher than the lowest value among the normalized weights of the core indicators, then the adjusted normalized weights will be used for overall integrity scoring. If the adjustment results in the overall weight percentage of the core indicators being too low, or the weight of a certain auxiliary indicator being higher than the minimum core indicator weight, then the weight adjustment will be rejected, the default weights in the mapping library will be used again, and the purchasing terminal will be notified that the weight setting does not meet the preset constraints.

[0096] In some embodiments of this application, a processing method of "pre-calculating general statistical results when data assets are put on the shelf, and performing partial recalculation based on scenario conditions after receiving an evaluation request" is adopted. This avoids re-statistically calculating all original data for each evaluation while ensuring that the returned statistical summary reflects the actual data situation under the current procurement scenario. Simultaneously, the conversion from reverse indicators to positive indicators is completed locally on the supplier's side, enabling the integrity assessment engine to use a unified scoring direction for subsequent processing. In step S103, the standardized feature statistical summary received by the integrity assessment engine is generated by the supplier-side local summary calculation node through the following steps: When data assets are put on the shelf, statistical calculations are performed on the full original dataset of the supplier's local machine according to a preset unified template to generate an initial statistical summary corresponding to the general statistical dimension, and the initial statistical summary is stored as baseline data. Upon receiving the statistical summary collection request, the system obtains the filtering conditional parameters determined by the integrity assessment engine based on the scenario requirement parameter set from the statistical summary collection request. For the conditional statistical dimensions related to the filtering conditional parameters, the system performs conditional filtering and statistical recalculation on the original data locally based on the filtering conditional parameters to obtain conditional adaptive statistical values. The condition-adaptive statistical values ​​are merged with the baseline data to form a standardized feature statistical summary corresponding to the current candidate data asset; For a negative indicator where a decrease in value indicates improved integrity, a positive indicator value corresponding to the negative indicator is calculated when generating the standardized feature statistical summary, and the positive indicator value replaces the negative indicator value in the standardized feature statistical summary.

[0097] As described above, the local summary computing node on the supplier side is deployed within the data supplier's data security domain and can access the local raw data corresponding to the data asset. When the data asset is uploaded to the data asset registration system, the local summary computing node performs statistical calculations on the full raw dataset corresponding to the data asset according to a preset unified template, obtains general statistical results that do not depend on specific procurement requests, forms an initial statistical summary, and saves the initial statistical summary as the baseline data of the data asset.

[0098] Among them, general statistical dimensions mainly refer to statistical dimensions that can be directly determined based on the overall data of the data asset and are generally not affected by temporary screening conditions of the purchaser. By generating such statistical results in advance during the listing stage, when a completeness assessment request is received later, the baseline data that has already been obtained can be directly reused without having to re-traverse and calculate the original data for all statistical dimensions, thereby reducing redundant statistical processing.

[0099] When the integrity assessment engine sends a statistical summary collection request to the local summary calculation node on the supplier side, it also sends the filtering condition parameters determined from the scenario requirement parameter set of the current assessment task along with the statistical summary collection request. After receiving the statistical summary collection request, the local summary calculation node on the supplier side retrieves the filtering condition parameters from the request and determines the conditional statistical dimensions affected by the filtering conditions. Subsequently, it filters the original data locally according to the filtering conditions and recalculates the corresponding statistical values ​​for the filtered data to obtain condition-adapted statistical values.

[0100] For statistical dimensions unaffected by the current screening criteria, the baseline data established during the uploading phase can continue to be used; for statistical dimensions that need to be redefined based on the current screening criteria, condition-adaptive statistical values ​​should be used. The local summary calculation node on the supplier side merges the two types of statistical results according to a unified data structure, ultimately forming a standardized feature statistical summary corresponding to the current candidate data assets and the current evaluation task. This allows for the reuse of existing statistical results while maintaining the relevance of indicators that need to adapt to specific scenario conditions.

[0101] Furthermore, to ensure a unified scoring direction for different integrity assessment indicators, the indicator direction is uniformly processed when generating standardized feature statistical summaries. For indicators where a larger statistical value indicates higher integrity, the corresponding statistical value can be directly written into the summary. For inverse indicators where a smaller statistical value indicates higher integrity, the local summary calculation node on the supplier side converts them into corresponding positive indicator values, and then writes the converted positive indicator values ​​into the standardized feature statistical summaries. For example, when the inverse indicator is a missing rate expressed as a value between 0 and 1, the difference between 1 and the missing rate can be used as the corresponding integrity rate; when the missing rate is expressed as a percentage, the difference between 100% and the missing rate can be used as the corresponding integrity rate. Through this conversion, the integrity assessment indicators entering step S104 uniformly satisfy the positive evaluation relationship that a larger statistically measured value indicates higher integrity.

[0102] In this way, the baseline data formed during the listing stage can be reused, and the relevant statistical values ​​can be updated according to the current screening conditions. Finally, only the standardized feature statistical summary is output to the integrity assessment engine.

[0103] In some embodiments of this application, to ensure that indicators not calculated or reported by the supplier are not directly ignored by the integrity assessment engine due to the lack of corresponding fields in the statistical summary, and to distinguish "undisclosed" from ordinary outliers and invalid values, thereby ensuring that step S104 can score undisclosed indicators according to established rules, and that step S105 can further identify such defects and perform risk analysis, step S103 marks integrity assessment indicators with unreported statistical values ​​as undisclosed, including: The local summary calculation node on the supplier side constructs index slots that correspond one-to-one with the integrity assessment index set determined in step S102 according to a preset fixed template. Each index slot includes at least an index unique identifier and an index statistical value field. For integrity assessment indicators that have not been calculated or reported, retain the corresponding indicator slot and the indicator's unique identifier, and write a preset undisclosed null value marker into the corresponding indicator statistical value field; The integrity assessment engine parses the standardized feature statistical summary, and when it identifies the undisclosed null value marker, it determines the integrity assessment indicator based on the corresponding indicator unique identifier and assigns an undisclosed status label to the integrity assessment indicator. When cleaning the standardized feature statistical summary, abnormal data in the reported indicator statistical values ​​are cleaned, and the undisclosed null value markers and corresponding undisclosed status labels with undisclosed status semantics are retained.

[0104] As described above, the local summary calculation node on the supplier side establishes corresponding indicator slots for each integrity assessment indicator according to the integrity assessment indicator set determined in step S102 and a preset fixed template, so that the indicator structure in the standardized feature statistical summary is consistent with the integrity assessment indicator set corresponding to the current scenario. Each indicator slot records at least an indicator unique identifier and an indicator statistical value. The indicator unique identifier is used to determine the specific integrity assessment indicator corresponding to the slot, and the indicator statistical value is used to record the statistical results calculated based on the supplier's local raw data.

[0105] When a certain integrity assessment indicator has completed statistical calculation and can be reported normally, the corresponding statistical result is written into the statistical value field of that indicator. For indicators that have not obtained statistical values ​​due to reasons such as not being calculated, not having the conditions for calculation, or the supplier not providing statistical results, their corresponding slots are not deleted. Instead, the unique identifier of the indicator is retained, and a preset undisclosed null value flag is written into the statistical value field. Thus, even if an indicator has no actual statistical value, the integrity assessment engine can still confirm that the indicator belongs to the current assessment indicator set based on the retained indicator slots. It simply means that the corresponding information is in an undisclosed state, rather than mistakenly believing that the indicator does not exist.

[0106] After receiving the standardized feature statistical summary, the integrity assessment engine identifies each indicator and its statistical results according to the indicator's unique identifier. When a pre-defined undisclosed null value flag exists in a statistical value field, the specific integrity assessment indicator is determined based on the corresponding indicator's unique identifier, and an undisclosed status label is assigned to that indicator. This label serves as status information in subsequent processing, distinguishing it from indicators that have already obtained normal statistical measurement values.

[0107] Furthermore, when performing data cleaning on standardized feature statistical summaries, it is necessary to distinguish between ordinary abnormal data and undisclosed tags with business semantics. For reported indicator statistical values, if there are abnormalities in format, illegal values, or other anomalies that do not conform to preset data rules, they can be processed according to the corresponding cleaning rules. For pre-set undisclosed null value tags, they are not directly deleted or corrected as ordinary null values, but their undisclosed semantics and corresponding status labels are retained. This can prevent the data cleaning process from mistakenly converting "supplier undisclosed" into ordinary null values, which would prevent subsequent scoring and risk assessment from recognizing this status.

[0108] For example, step S102 determines that the current business scenario requires evaluation of five completeness indicators. The supplier-side local summary calculation node establishes five corresponding indicator slots according to a fixed template. Four of the indicators can be calculated and reported normally, while the other indicator, "Consumer Category Tag Coverage Completeness Rate," does not obtain statistical results. At this time, the local summary calculation node still retains the slot and unique identifier of the indicator, only writing an undisclosed null value marker in its statistical value field.

[0109] When the integrity assessment engine parses the statistical summary, it can determine the "consumer category label coverage completeness rate" corresponding to the slot based on the indicator's unique identifier, and set the indicator to an undisclosed state based on the undisclosed null value flag. Even after subsequent data cleaning processes other abnormal statistical values, the undisclosed flag and status label for this indicator are retained. Thus, in step S104, the undisclosed indicator can be processed according to the scoring rules corresponding to it, and in step S105, the indicator can be identified as a corresponding information undisclosure defect, avoiding the omission of corresponding data integrity risks due to missing statistical fields.

[0110] In some embodiments of this application, in order to construct segmented scoring intervals through demand thresholds and minimum tolerance thresholds, so that statistically measured values ​​of different degrees of completeness can be converted into continuous and comparable single-indicator scores, and undisclosed indicators are treated as zero; and then, by combining the weight of each indicator in the current application scenario to form an overall completeness score, the evaluation results are avoided from being too coarse due to the simple binary judgment of "meeting the standard / not meeting the standard," and the overall score can simultaneously reflect the actual degree of completion and the importance of the indicator. In step S104, the demand threshold corresponding to each completeness assessment indicator is greater than the corresponding minimum tolerance threshold, and the single-indicator score corresponding to each completeness assessment indicator is determined according to the segmented scoring rules, including: When the statistical measured value of the integrity assessment indicator is greater than or equal to the corresponding demand threshold, the single indicator score of the integrity assessment indicator is determined as the full score of the single indicator. When the statistical measured value of the integrity assessment indicator is less than the corresponding minimum tolerance threshold, or when the integrity assessment indicator is in an undisclosed state, the single indicator score of the integrity assessment indicator shall be determined as zero. When the statistical measured value of the integrity assessment indicator is greater than or equal to the corresponding minimum tolerance threshold and less than the corresponding demand threshold, calculate the first difference between the statistical measured value and the minimum tolerance threshold and the second difference between the demand threshold and the minimum tolerance threshold, and multiply the ratio of the first difference to the second difference by the full score of the single indicator to obtain the single indicator score of the integrity assessment indicator. The overall integrity score is determined by summing the products of the weights corresponding to each integrity assessment indicator and the scores of the corresponding individual indicators.

[0111] As described above, the integrity assessment engine obtains the corresponding statistical measured value in the standardized feature statistical summary for each integrity assessment indicator and compares it with the demand threshold and minimum tolerance threshold determined in step S102. Since the indicators entering the scoring stage are all in positive indicator form, the larger the statistical measured value, the higher the corresponding integrity level, and segmented scoring can be performed according to a unified judgment direction. In this embodiment, the demand threshold is higher than the corresponding minimum tolerance threshold, so that a linear conversion interval is formed between the two thresholds for scoring the partial compliance status.

[0112] When the statistically measured value is greater than or equal to the demand threshold, it indicates that the candidate data asset has met the expected requirements of the current application scenario in this indicator dimension. Therefore, the single indicator score is determined as the full score for that single indicator. The full score for a single indicator can be determined according to the system's preset scoring benchmark, for example, using 100 points as the full score for a single item.

[0113] When the measured statistical value is less than the minimum tolerance threshold, it indicates that the indicator is below the minimum acceptable level for the current application scenario, and therefore the score for that indicator is set to zero. For indicators that are not disclosed, since there is a lack of valid statistical values ​​to verify their completeness, their scores are also set to zero, so that the undisclosed information can be reflected in the overall completeness score and will not be directly ignored due to the lack of statistical values.

[0114] When the measured value falls between the minimum tolerance threshold and the demand threshold, it indicates that although the indicator has not yet fully met business requirements, it is still within an acceptable range. In this case, the score of the single indicator is linearly adjusted based on the position of the measured value relative to the minimum tolerance threshold and the demand threshold. Specifically, the difference between the measured value and the minimum tolerance threshold is used as the first difference, and the difference between the demand threshold and the minimum tolerance threshold is used as the second difference. The ratio of the first difference to the second difference represents the degree of completion of the current indicator within the acceptable range. This ratio is then multiplied by the full score for the single indicator to obtain the corresponding single indicator score. Therefore, the closer the measured value is to the demand threshold, the closer the single indicator score is to the full score; the closer it is to the minimum tolerance threshold, the closer the single indicator score is to zero.

[0115] For example, if a certain indicator has a demand threshold of 95%, a minimum tolerance threshold of 80%, a statistically measured value of 90%, and a maximum score of 100, then this indicator falls between the minimum tolerance threshold and the demand threshold. The first difference is 10 percentage points, and the second difference is 15 percentage points. Therefore, the indicator score can be calculated by proportionally adjusting these two percentage points. In this way, different degrees of non-compliance can be expressed differently, rather than all being assigned the same non-compliance result.

[0116] After scoring each individual indicator, the integrity assessment engine multiplies the score of each individual indicator by the corresponding weight determined in step S102 to obtain the weighted contribution of that indicator to the overall integrity score. Then, it sums the weighted contributions of all indicators to obtain the overall integrity score of the current candidate data asset. For core indicators with higher weights, changes in their scores will have a greater impact on the overall integrity score; for auxiliary indicators with relatively lower weights, the impact is correspondingly smaller.

[0117] In some embodiments of this application, to avoid masking the defects of core indicators by determining the risk level solely based on the overall integrity score, and to further refine and quantify the business impact according to the actual degree to which the indicators deviate from business requirements, step S105, which determines the risk level and quantifies the negative business impact, includes: The basic risk level is determined based on the preset scoring range in which the overall integrity score falls, and the compliance status of each core indicator is detected. When at least one core indicator has a statistically measured value that is lower than the corresponding demand threshold or is not disclosed, and the basic risk level is low risk, the basic risk level will be adjusted to medium risk; when the basic risk level is medium risk or high risk, the basic risk level will remain unchanged. From the knowledge graph of missing impacts, which is established according to five dimensions of indicator defects, business scenarios, risk types, quantitative impacts and mitigation strategies, risk information matching the current indicator defect is retrieved based on the indicator defect type and the application scenario identifier, and the corresponding quantitative impact benchmark interval is read. For an integrity assessment index that has a statistically measured value and the statistically measured value is lower than the corresponding demand threshold, the deviation between the statistically measured value and the demand threshold is calculated, and a quantitative impact sub-interval is determined from the corresponding quantitative impact benchmark interval based on the deviation. The larger the deviation, the closer the quantitative impact sub-interval is to the upper limit of the quantitative impact benchmark interval. The quantitative negative impact on business is generated based on the retrieved risk type and the corresponding quantitative impact benchmark range or quantitative impact sub-range.

[0118] As described above, the integrity assessment engine first determines the basic risk level based on the preset scoring range where the overall integrity score obtained in step S104 falls. Different scoring ranges correspond to different levels of procurement risk. The lower the overall integrity score, the greater the deviation between the candidate data asset and the integrity requirements of the current business scenario, and the higher the corresponding risk level.

[0119] In one specific implementation, a percentage-based overall integrity score can be used. When the overall integrity score is greater than 80 but not exceeding 100, the basic risk level is determined to be low risk; when the overall integrity score is greater than or equal to 50 but less than or equal to 80, the basic risk level is determined to be medium risk; and when the overall integrity score is less than 50, the basic risk level is determined to be high risk. When the maximum score is not 100, the overall integrity score can be first converted to a percentage score according to the maximum score, and then the basic risk level can be determined according to the aforementioned score range.

[0120] After determining the basic risk level, further checks are conducted to determine whether each core indicator in the current integrity assessment indicator set meets the corresponding required threshold, and whether there are any undisclosed states. Core indicators represent key integrity dimensions that have a significant impact on the current application scenario; therefore, even with a high overall score, it is still necessary to use the status of core indicators to constrain the basic risk level.

[0121] When the basic risk level is low, but at least one core indicator's measured value is below the required threshold, or a core indicator is not disclosed, the risk level is adjusted from low to medium risk. This ensures that data assets with core indicator deficiencies cannot be classified as low risk solely because of higher scores on other indicators. If the basic risk level is already medium or high, the original risk level remains unchanged. This forms a risk assessment mechanism that involves both overall score and core indicator status. For example, if a candidate data asset has an overall integrity score of 85, which would normally be considered low risk according to the score range, but a core indicator is not disclosed, the final risk level is adjusted to medium risk.

[0122] After determining the risk level, the integrity assessment engine identifies the corresponding indicator defect type for indicators whose statistically measured values ​​are below the required threshold or are in an undisclosed state. It then accesses a pre-built knowledge graph of missing data impacts based on the application scenario identifier. This knowledge graph establishes connections along five dimensions: indicator defect, business scenario, risk type, quantified impact, and mitigation strategy. This allows the same data defect to be associated with different risk outcomes in different business scenarios. In one implementation, the knowledge graph is stored in a graph database. Indicator defects, business scenarios, risk types, quantified impacts, and mitigation strategies are used as nodes, node attributes, or related information within the graph. Edge relationships between nodes establish the correspondence between indicator defects and risk types, quantified impacts, and mitigation strategies in different business scenarios. The quantified impact baseline range is pre-written into the knowledge graph as an attribute of the corresponding risk information for subsequent retrieval based on indicator defect type and application scenario.

[0123] During retrieval, the current indicator defect type and application scenario are used as matching criteria to determine the risk type corresponding to the current defect, and the pre-stored quantitative impact benchmark range in the knowledge graph is retrieved. Quantitative impact can be represented in the form of model performance degradation, business indicator deviation, failure probability, or cost loss ratio. For example, in the scenario of precision marketing for financial products, insufficient completeness of the revenue field can correspond to the risk of model prediction accuracy decay, and is associated with the corresponding range of decline in model AUC (Area Under the Receiver Operating Characteristic); undisclosed tag coverage can correspond to the risk of unverifiable audience package accuracy, and is associated with quantitative impacts such as CTR (click-through rate) prediction deviation and failure probability.

[0124] For indicators with statistically measured values ​​but not yet reaching the demand threshold, the quantitative impact can be further refined based on the degree of deviation. Specifically, the deviation between the statistically measured value and the demand threshold can be calculated, and then a matching quantitative impact sub-interval can be selected from the corresponding quantitative impact benchmark interval in the knowledge graph based on the deviation amplitude. The smaller the deviation amplitude, the closer the selected sub-interval is to the lower limit of the benchmark interval; the larger the deviation amplitude, the closer the selected sub-interval is to the upper limit of the benchmark interval, thus allowing the negative impact of the quantitative business to change accordingly with the severity of the indicator's defect. In one optional implementation, the deviation amplitude of the indicator's statistically measured value relative to the demand threshold can be pre-divided into multiple progressively increasing deviation levels, and the quantitative impact benchmark interval can be divided into multiple quantitative impact sub-intervals arranged sequentially from the lower limit to the upper limit of the benchmark interval, so that the deviation levels and quantitative impact sub-intervals correspond in ascending order. Once the deviation amplitude of the current indicator falls into the corresponding deviation level, the quantitative impact sub-interval corresponding to that deviation level is selected. The larger the deviation amplitude, the closer the corresponding quantitative impact sub-interval is to the upper limit of the quantitative impact benchmark interval.

[0125] For example, the pre-defined baseline range for the quantitative impact of insufficient completeness of a certain field in a knowledge graph is a decrease in model AUC of 3% to 7%. When the statistically measured value of this indicator is 94% and the demand threshold is 95%, the deviation is only 1 percentage point, and 3% to 4% closer to the lower limit of the baseline range can be determined as the quantitative impact sub-range. When the statistically measured value is 85% and the demand threshold is still 95%, the deviation reaches 10 percentage points, and 6% to 7% closer to the upper limit of the baseline range can be selected as the quantitative impact sub-range.

[0126] For indicators that are not disclosed, since there are no valid statistically measured values ​​available for calculating the deviation magnitude, the above deviation calculation can be skipped. Instead, risk expression can be performed directly using the quantitative impact benchmark range obtained from knowledge graph retrieval or the pre-set quantitative risk information corresponding to the defect, based on the indicator defect type and application scenario.

[0127] Finally, the retrieved risk types are associated with the corresponding quantitative impact benchmark intervals or dynamically determined quantitative impact sub-intervals to form the quantitative business negative impacts corresponding to the candidate data assets.

[0128] In some embodiments of this application, in order to not only alert the purchaser to the corresponding risks after identifying non-compliant or undisclosed indicators of candidate data assets, but also further search for similar data assets that can remedy the current deficiencies from the data asset registration system, and quantify and rank them according to the degree of deficiencies remedied, thereby transforming the integrity evaluation results into actionable data supplementation suggestions and reducing the processing cost for the purchaser to manually search for supplementary data assets. In step S105, the risk mitigation scheme includes data supplementation recommendations, and the generation of the data supplementation recommendations includes: Extract unique identifiers from integrity assessment indicators whose statistically measured values ​​are below the corresponding demand thresholds or are in an undisclosed state to form a set of defect indicators; Based on the target data asset category and the set of defect indicators, a search is initiated with the data asset registration system to obtain candidate supplementary assets that are the same as the target data asset category and whose statistical measured value in at least one dimension of the defect indicator is greater than or equal to the corresponding demand threshold. For each candidate supplementary asset, a basic fit score is determined based on the ratio of the number of defect indicators that the candidate supplementary asset can compensate for to the total number of defect indicators in the defect indicator set. When at least one defect indicator in the candidate supplementary asset has a statistically measured value that exceeds the corresponding demand threshold by a preset excess range, a preset bonus is added to the basic fit score to obtain the fit score. When there are no defect indicators whose statistically measured values ​​exceed the corresponding demand threshold by the preset excess range, the basic fit score is determined as the fit score. The candidate supplementary assets are sorted according to their suitability scores, and a data supplement recommendation list is generated by selecting the candidate supplementary assets that meet the preset ranking conditions.

[0129] As described above, the integrity assessment engine first reads the indicator evaluation results generated in steps S104 and S105, extracts unique identifiers from indicators whose statistically measured values ​​are below the required threshold or those that are not disclosed, and summarizes these indicators to form a defect indicator set. The defect indicator set is used to characterize the specific dimensions in which the current candidate data asset is insufficient relative to the integrity requirements of this application scenario.

[0130] Subsequently, the data asset registration system is accessed using the target data asset category and the set of defect indicators as search criteria. The data asset registration system associates registered data assets with corresponding standardized feature statistical summaries or information indexable to those summaries. During candidate supplementary asset retrieval, the system obtains the statistical values ​​of the indicators corresponding to the defect indicator set from the relevant statistical summaries based on the data asset identifier. The integrity assessment engine then determines whether the measured statistical values ​​of the asset to be retrieved in each dimension of the defect indicator meet the corresponding requirement threshold. Data assets that meet the requirement threshold in at least one dimension of a defect indicator are identified as candidate supplementary assets. In other words, candidate supplementary assets do not need to meet the requirements in all defect indicators; as long as they can supplement at least one defect dimension of the current candidate data asset, they can proceed to the subsequent suitability evaluation.

[0131] For each candidate supplementary asset obtained, the integrity assessment engine further calculates its basic fit score. Specifically, it counts the number of indicators in the candidate supplementary asset that meet the demand threshold and thus can compensate for the current defects, then calculates the proportion of this number to the total number of defect indicators, and uses the obtained proportion as the basic fit score. Therefore, candidate supplementary assets that can simultaneously compensate for more defect dimensions have a higher basic fit.

[0132] Building upon this, the actual performance of candidate supplementary assets in the corresponding defect dimensions can also be considered. When at least one defect indicator in a candidate supplementary asset exceeds the corresponding requirement threshold by a preset excess margin, a preset bonus is added to the basic fit score to obtain the overall fit score; otherwise, the basic fit score is used directly as the overall fit score. The preset excess margin is used to determine whether the performance of the candidate supplementary asset in the corresponding defect dimension is significantly better than the requirement standard of the current application scenario. For example, exceeding the requirement threshold by 10 percentage points can be considered as a preset excess margin, and the preset bonus can be pre-configured by the system.

[0133] After calculating the fit, candidate supplementary assets are sorted from highest to lowest fit score, and a selection of assets is chosen based on preset ranking criteria to form a data supplement recommendation list. For example, the top three candidate supplementary assets can be selected. The recommendation list can record the asset's unique identifier, asset name, supplier information, the metrics it can compensate for, fit score, and asset detail redirection information, so that the purchaser can understand which deficiencies each recommended asset can address.

[0134] For example, the current candidate data asset has three deficient indicators: insufficient completion rate of the monthly income field, undisclosed completeness rate of consumption category tag coverage, and insufficient target population sample coverage multiple. If a candidate supplementary asset meets the corresponding requirement threshold in two of these indicator dimensions, its basic fit can be determined by the proportion of the two deficient indicators that can be remedied to the total three deficient indicators. If the statistically measured value of the asset in one of the deficient dimensions exceeds the corresponding requirement threshold by a preset excess amount, a preset bonus is added to the basic fit score to obtain the final fit score. Compared to other assets that can only remedy one deficient dimension, this candidate supplementary asset usually receives a higher recommendation priority.

[0135] In some embodiments of this application, in order to further transform the aforementioned integrity scoring and risk analysis results into risk management information that the purchaser can directly view and implement, corresponding negotiation suggestions are generated for defects that can be resolved through supplier supplementary disclosure or verification. Key risks are highlighted through hierarchical display and differentiated marking, while providing entry points for data and information supplementation. In step S105, the risk mitigation plan includes data supplementation recommendations and supplier negotiation suggestions. In step S106, the structured risk report is displayed, including: Based on the completeness assessment indicators that are not disclosed or whose statistical measured values ​​are lower than the corresponding demand threshold, the statistical indicators and verification parameters that need to be supplemented or verified by the supplier are determined, and the supplier negotiation suggestions are formed. The supplier negotiation suggestions include supplementary information requirements and re-evaluation of interface information. The structured risk report is presented in three layers: overview, details, and solutions. The overview layer displays the overall integrity score and risk level. The details layer displays the statistical measured values, individual index scores, weight contributions, and compliance status of each integrity assessment indicator. The solutions layer displays the quantified negative impact on business and risk mitigation solutions. The system provides differentiated prompts for core indicators that do not meet the demand threshold, marks undisclosed integrity assessment indicators that are not yet disclosed, and provides data supplement recommendation jumps corresponding to the data supplement recommendations, as well as supplier information supplement interaction entries corresponding to the supplier negotiation suggestions.

[0136] As described above, the integrity assessment engine identifies integrity assessment indicators that are not disclosed or whose measured statistical values ​​are below the required threshold, based on the indicator evaluation results generated in steps S104 and S105. For these indicators, it further determines the statistical information that needs to be supplemented by the supplier or the parameters that need to be re-verified, and formulates supplier negotiation suggestions accordingly. Specifically, for undisclosed indicators, the supplier can be required to supplement the corresponding statistical results; for indicators whose measured statistical values ​​do not reach the required threshold but require further confirmation of the actual data situation, verification requirements for the corresponding statistical dimensions can be proposed.

[0137] The supplementary information requirements in the supplier negotiation suggestions are used to clarify what the purchaser wants the data supplier to provide further. For example, when the consumer category label coverage completeness rate is not disclosed, the supplier can be asked to supplement the statistical results of this indicator and the related sample population stratification distribution information. The reassessment interface information is used to re-trigger the completeness assessment of the current candidate data assets after the supplier completes the supplementary disclosure or data verification, so that the updated statistical results can be processed again in steps S103 to S105.

[0138] After generating the structured risk report, the purchasing terminal displays the data in three layers: overview, details, and solutions. The overview layer provides the overall evaluation results of the candidate data assets, primarily showing the overall integrity score and risk level, enabling the purchasing party to quickly determine the overall integrity and risk level of the data asset for the current application scenario. The details layer further displays the specific evaluation results of each integrity assessment indicator, including statistical measured values, individual indicator scores, weight contributions, and compliance status, allowing the purchasing party to determine the specific composition of the overall integrity score and identify indicators that cause a decrease in score or an increase in risk. By simultaneously displaying statistical measured values ​​and scoring results, a correspondence can be established between the final score and the underlying statistical basis, improving the understandability of the evaluation results. The solutions layer displays the quantitative negative business impact and risk mitigation solutions generated in step S105, enabling the purchasing party to further understand the potential business impact of corresponding defects after confirming the source of risk, as well as possible handling methods such as data supplementation and supplier negotiation.

[0139] Furthermore, to enable purchasers to quickly identify abnormal indicators that have a significant impact on the current business scenario, core indicators whose actual measured values ​​have not reached the required threshold can be highlighted in the display interface using a different prompting method than normal indicators; for indicators that are in an undisclosed state, a special undisclosed mark will be used to distinguish between the two states of "indicator actually not meeting the standard" and "supplier has not yet provided statistical information", so as to avoid confusion for purchasers regarding different sources of risk.

[0140] In the display area corresponding to the risk mitigation plan, you can also set up a data supplement recommendation jump link and a supplier information supplement interaction link. Through the data supplement recommendation jump link, the purchaser can access the details of the corresponding candidate supplementary assets in the aforementioned data supplement recommendation list; through the supplier information supplement interaction link, the purchaser can initiate a supplementary information request for indicators that require further disclosure or verification.

[0141] Based on the same inventive concept as the above method, this application also proposes a data asset integrity assessment and risk warning system for business scenarios, which is deployed on a data trading platform. Figure 2 This is a schematic diagram of the system structure, which includes: The requirement parsing and retrieval module is used to receive data query request messages sent by the purchasing terminal, parse the target data asset category, application scenario identifier and scenario requirement parameter set from the data query request message, and obtain a candidate data asset list from the data asset registration system according to the target data asset category. The indicator matching and parameter initialization module is used to determine the corresponding integrity assessment indicator set and the requirement threshold, minimum tolerance threshold and weight of each integrity assessment indicator from the preset scenario-integrity indicator mapping library, using the application scenario identifier as an index. The integrity assessment indicator set includes core indicators and auxiliary indicators. The summary acquisition and status marking module is used to send a statistical summary acquisition request to the corresponding local summary calculation node on the supplier side for each candidate data asset in the candidate data asset list, and to receive a standardized feature statistical summary generated by the local summary calculation node on the supplier side after statistical calculation based on the supplier's local original data, wherein the supplier's local original data is not sent to the system; it is also used to mark the integrity assessment indicators of unreported statistical values ​​in the standardized feature statistical summary as undisclosed status; The quantitative scoring module is used to determine the individual index score corresponding to each integrity assessment index according to the statistical measured value, demand threshold and minimum tolerance threshold of each integrity assessment index expressed in the form of positive indicators in the standardized feature statistical summary, and to perform weighted summation of the individual index scores according to the weights corresponding to each integrity assessment index to obtain the overall integrity score of the candidate data asset. The risk reasoning module is used to determine the risk level of candidate data assets based on the overall integrity score and the compliance status of core indicators, determine the indicator defect type based on the integrity assessment indicators whose statistical measured values ​​have not reached the corresponding demand threshold or are in an undisclosed state, search in the pre-set knowledge graph of missing impact based on the indicator defect type, determine the corresponding quantitative business negative impact, and generate a risk mitigation plan based on the indicator defect type. The visualization output module is used to assemble the overall integrity score, risk level, quantified negative business impact and risk mitigation plan into a structured risk report, and send the structured risk report to the purchaser's terminal for display.

[0142] The specific data processing procedures performed by each of the above modules can be referred to in steps S101 to S106 and their corresponding implementation methods, and will not be repeated here.

[0143] The method steps and system modules in the various embodiments of this application can be implemented by a processor executing corresponding program instructions. The program instructions can be stored in a computer-readable storage medium and executed by a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to implement the corresponding data processing functions described in the above embodiments. Other software, hardware, or software-hardware combinations that can be implemented by those skilled in the art based on the disclosure of this application will not be elaborated here.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for data asset integrity assessment and risk warning in business scenarios, applied in a data asset transaction environment including a purchaser's terminal, a data asset registration system, an integrity assessment engine, and a supplier-side local summary computing node, characterized in that... The method includes: S101, the integrity assessment engine receives a data query request message sent by the purchaser's terminal, parses the target data asset category, application scenario identifier, and scenario requirement parameter set from the data query request message, and obtains a candidate data asset list from the data asset registration system according to the target data asset category; S102, using the application scenario identifier as an index, determine the corresponding integrity assessment index set and the requirement threshold, minimum tolerance threshold and weight corresponding to each integrity assessment index from the preset scenario-integrity index mapping library, wherein the integrity assessment index set includes core indicators and auxiliary indicators. S103, for each candidate data asset in the candidate data asset list, a statistical summary collection request is sent to the corresponding local summary calculation node on the supplier side, and a standardized feature statistical summary generated by the local summary calculation node on the supplier side after statistical calculation based on the supplier's local original data is received. The supplier's local original data is not sent to the integrity assessment engine. The integrity assessment engine marks the integrity assessment indicators of the standardized feature statistical summary that have not reported statistical values ​​as undisclosed. S104. Based on the statistical measured values, demand thresholds, and minimum tolerance thresholds of each integrity assessment indicator expressed in the form of positive indicators in the standardized feature statistical summary, the individual indicator scores corresponding to each integrity assessment indicator are determined according to the segmented scoring rules, and the individual indicator scores are weighted and summed according to the weights corresponding to each integrity assessment indicator to obtain the overall integrity score of the candidate data asset. S105. Based on the overall integrity score and the compliance status of the core indicators, determine the risk level corresponding to the candidate data assets. Based on the integrity assessment indicators whose statistical measured values ​​do not reach the corresponding demand threshold or are in an undisclosed state, determine the indicator defect type. Based on the indicator defect type, search in the preset knowledge graph of missing impacts to determine the corresponding quantitative business negative impact, and generate a risk mitigation plan based on the indicator defect type. S106, the overall integrity score, risk level, quantified negative business impact and risk mitigation plan are assembled into a structured risk report, and the structured risk report is sent to the purchaser's terminal for display.

2. The method according to claim 1, characterized in that, In step S101, the scenario requirement parameter set includes filtering condition parameters and threshold constraint parameters; The filtering condition parameters are sent to the data asset registration system as the initial screening conditions for candidate data assets. The data asset registration system then removes asset records that do not meet the filtering condition parameters from the asset records that match the data asset category, thus obtaining the candidate data asset list. In step S102, based on the mapping relationship between the threshold constraint parameter and the integrity evaluation index, the integrity evaluation index corresponding to the threshold constraint parameter is determined, and the required threshold of the corresponding integrity evaluation index is determined based on the threshold constraint parameter. The determined required threshold is written into the corresponding required threshold field to replace the default required threshold preset in the scenario-integrity index mapping library for the current application scenario.

3. The method according to claim 1, characterized in that, Step S102 determines the weights corresponding to each integrity assessment index, including: Obtain the default weights corresponding to each integrity assessment indicator from the scenario-integrity indicator mapping library; When the data query request message does not contain weight fine-tuning parameters, the default weight is determined as the weight of the corresponding integrity assessment index; When the data query request message contains a weight fine-tuning parameter, the default weight of the corresponding integrity assessment index is adjusted according to the weight fine-tuning parameter to form a weight to be normalized, and the sum of the weights to be normalized of all integrity assessment indices is calculated. The ratio of the unnormalized weight of each integrity assessment indicator to the sum of the unnormalized weights is determined as the normalized weight of the corresponding integrity assessment indicator. Calculate the sum of the normalized weights of the core indicators, and determine whether the sum of the normalized weights of the core indicators is lower than the preset minimum weight ratio threshold of the core indicators. At the same time, determine whether the normalized weight of any auxiliary indicator is higher than the lowest value among the normalized weights of the core indicators. When the sum of the normalized weights of the core indicators is not lower than the minimum weight percentage threshold of the core indicators, and the normalized weights of each auxiliary indicator are not higher than the lowest value among the normalized weights of each core indicator, the normalized weights are determined as the weights of the corresponding integrity assessment indicators. Otherwise, discard the weight fine-tuning parameters, use the corresponding default weights in the scenario-integrity index mapping library to perform subsequent overall integrity score calculations, and return a weight parameter constraint exception prompt to the purchasing terminal.

4. The method according to claim 1, characterized in that, In step S103, the standardized feature statistical summary received by the integrity assessment engine is generated by the local summary calculation node on the supplier side through the following steps: When data assets are put on the shelf, statistical calculations are performed on the full original dataset of the supplier's local machine according to a preset unified template to generate an initial statistical summary corresponding to the general statistical dimension, and the initial statistical summary is stored as baseline data. Upon receiving the statistical summary collection request, the system obtains the filtering conditional parameters determined by the integrity assessment engine based on the scenario requirement parameter set from the statistical summary collection request. For the conditional statistical dimensions related to the filtering conditional parameters, the system performs conditional filtering and statistical recalculation on the original data locally based on the filtering conditional parameters to obtain conditional adaptive statistical values. The condition-adaptive statistical values ​​are merged with the baseline data to form a standardized feature statistical summary corresponding to the current candidate data asset; For a negative indicator where a decrease in value indicates improved integrity, a positive indicator value corresponding to the negative indicator is calculated when generating the standardized feature statistical summary, and the positive indicator value replaces the negative indicator value in the standardized feature statistical summary.

5. The method according to claim 1, characterized in that, In step S103, the integrity assessment indicators for unreported statistical values ​​are marked as undisclosed, including: The local summary calculation node on the supplier side constructs index slots that correspond one-to-one with the integrity assessment index set determined in step S102 according to a preset fixed template. Each index slot includes at least an index unique identifier and an index statistical value field. For integrity assessment indicators that have not been calculated or reported, retain the corresponding indicator slot and the indicator's unique identifier, and write a preset undisclosed null value marker into the corresponding indicator statistical value field; The integrity assessment engine parses the standardized feature statistical summary, and when it identifies the undisclosed null value marker, it determines the integrity assessment indicator based on the corresponding indicator unique identifier and assigns an undisclosed status label to the integrity assessment indicator. When cleaning the standardized feature statistical summary, abnormal data in the reported indicator statistical values ​​are cleaned, and the undisclosed null value markers and corresponding undisclosed status labels with undisclosed status semantics are retained.

6. The method according to claim 1, characterized in that, In step S104, if the demand threshold corresponding to each integrity assessment indicator is greater than the corresponding minimum tolerance threshold, the individual indicator score for each integrity assessment indicator is determined according to the segmented scoring rules, including: When the statistical measured value of the integrity assessment indicator is greater than or equal to the corresponding demand threshold, the single indicator score of the integrity assessment indicator is determined as the full score of the single indicator. When the statistical measured value of the integrity assessment indicator is less than the corresponding minimum tolerance threshold, or when the integrity assessment indicator is in an undisclosed state, the single indicator score of the integrity assessment indicator shall be determined as zero. When the statistical measured value of the integrity assessment indicator is greater than or equal to the corresponding minimum tolerance threshold and less than the corresponding demand threshold, calculate the first difference between the statistical measured value and the minimum tolerance threshold and the second difference between the demand threshold and the minimum tolerance threshold, and multiply the ratio of the first difference to the second difference by the full score of the single indicator to obtain the single indicator score of the integrity assessment indicator. The overall integrity score is determined by summing the products of the weights corresponding to each integrity assessment indicator and the scores of the corresponding individual indicators.

7. The method according to claim 1, characterized in that, Step S105 involves determining the risk level and quantifying the negative impact on business operations, including: The basic risk level is determined based on the preset scoring range in which the overall integrity score falls, and the compliance status of each core indicator is detected. When at least one core indicator has a statistically measured value that is lower than the corresponding demand threshold or is not disclosed, and the basic risk level is low risk, the basic risk level will be adjusted to medium risk; when the basic risk level is medium risk or high risk, the basic risk level will remain unchanged. From the knowledge graph of missing impacts, which is established according to five dimensions of indicator defects, business scenarios, risk types, quantitative impacts and mitigation strategies, risk information matching the current indicator defect is retrieved based on the indicator defect type and the application scenario identifier, and the corresponding quantitative impact benchmark interval is read. For an integrity assessment index that has a statistically measured value and the statistically measured value is lower than the corresponding demand threshold, the deviation between the statistically measured value and the demand threshold is calculated, and a quantitative impact sub-interval is determined from the corresponding quantitative impact benchmark interval based on the deviation. The larger the deviation, the closer the quantitative impact sub-interval is to the upper limit of the quantitative impact benchmark interval. The quantitative negative impact on business is generated based on the retrieved risk type and the corresponding quantitative impact benchmark range or quantitative impact sub-range.

8. The method according to claim 1, characterized in that, In step S105, the risk mitigation plan includes data supplementation recommendations, and the generation of the data supplementation recommendations includes: Extract unique identifiers from integrity assessment indicators whose statistically measured values ​​are below the corresponding demand thresholds or are in an undisclosed state to form a set of defect indicators; Based on the target data asset category and the set of defect indicators, a search is initiated with the data asset registration system to obtain candidate supplementary assets that are the same as the target data asset category and whose statistical measured value in at least one dimension of the defect indicator is greater than or equal to the corresponding demand threshold. For each candidate supplementary asset, a basic fit score is determined based on the ratio of the number of defect indicators that the candidate supplementary asset can compensate for to the total number of defect indicators in the defect indicator set. When at least one defect indicator in the candidate supplementary asset has a statistically measured value that exceeds the corresponding demand threshold by a preset excess range, a preset bonus is added to the basic fit score to obtain the fit score. When there are no defect indicators whose statistically measured values ​​exceed the corresponding demand threshold by the preset excess range, the basic fit score is determined as the fit score. The candidate supplementary assets are sorted according to their suitability scores, and a data supplement recommendation list is generated by selecting the candidate supplementary assets that meet the preset ranking conditions.

9. The method according to claim 1, characterized in that, In step S105, the risk mitigation plan includes data supplementation recommendations and supplier negotiation suggestions. In step S106, the structured risk report is displayed, including: Based on the completeness assessment indicators that are not disclosed or whose statistical measured values ​​are lower than the corresponding demand threshold, the statistical indicators and verification parameters that need to be supplemented or verified by the supplier are determined, and the supplier negotiation suggestions are formed. The supplier negotiation suggestions include supplementary information requirements and re-evaluation of interface information. The structured risk report is presented in three layers: overview, details, and solutions. The overview layer displays the overall integrity score and risk level. The details layer displays the statistical measured values, individual index scores, weight contributions, and compliance status of each integrity assessment indicator. The solutions layer displays the quantified negative impact on business and risk mitigation solutions. The system provides differentiated prompts for core indicators that do not meet the demand threshold, marks undisclosed integrity assessment indicators that are not yet disclosed, and provides data supplement recommendation jumps corresponding to the data supplement recommendations, as well as supplier information supplement interaction entries corresponding to the supplier negotiation suggestions.

10. A data asset integrity assessment and risk warning system for business scenarios, deployed on a data trading platform, characterized in that... include: The requirement parsing and retrieval module is used to receive data query request messages sent by the purchasing terminal, parse the target data asset category, application scenario identifier and scenario requirement parameter set from the data query request message, and obtain a candidate data asset list from the data asset registration system according to the target data asset category. The indicator matching and parameter initialization module is used to determine the corresponding integrity assessment indicator set and the requirement threshold, minimum tolerance threshold and weight of each integrity assessment indicator from the preset scenario-integrity indicator mapping library, using the application scenario identifier as an index. The integrity assessment indicator set includes core indicators and auxiliary indicators. The summary acquisition and status marking module is used to send a statistical summary acquisition request to the corresponding local summary calculation node on the supplier side for each candidate data asset in the candidate data asset list, and to receive a standardized feature statistical summary generated by the local summary calculation node on the supplier side after statistical calculation based on the supplier's local original data, wherein the supplier's local original data is not sent to the system; it is also used to mark the integrity assessment indicators of unreported statistical values ​​in the standardized feature statistical summary as undisclosed status; The quantitative scoring module is used to determine the individual index score corresponding to each integrity assessment index according to the statistical measured value, demand threshold and minimum tolerance threshold of each integrity assessment index expressed in the form of positive indicators in the standardized feature statistical summary, and to perform weighted summation of the individual index scores according to the weights corresponding to each integrity assessment index to obtain the overall integrity score of the candidate data asset. The risk reasoning module is used to determine the risk level of candidate data assets based on the overall integrity score and the compliance status of core indicators, determine the indicator defect type based on the integrity assessment indicators whose statistical measured values ​​have not reached the corresponding demand threshold or are in an undisclosed state, search in the pre-set knowledge graph of missing impact based on the indicator defect type, determine the corresponding quantitative business negative impact, and generate a risk mitigation plan based on the indicator defect type. The visualization output module is used to assemble the overall integrity score, risk level, quantified negative business impact and risk mitigation plan into a structured risk report, and send the structured risk report to the purchaser's terminal for display.

Images