An intelligent evaluation system and method applied to a financial data link

Abstract

The application discloses an intelligent evaluation system and method applied to a financial data link, relates to the technical field of financial data quality monitoring, and evaluation comprises the following steps: calling a financial data operation log, extracting operation record information and combing a flow data link of target data; generating a data flow time axis according to a flow time, calculating error propagation probability and data matching degree among nodes, and calculating a data distortion risk value of each node according to the error propagation probability and the data matching degree; identifying a target flow node whose risk exceeds a threshold value, and tracing upstream nodes thereof; calculating a distortion influence range of each upstream node, constructing a distortion source head scoring model, and locating an initial link and a reason of data distortion. The application realizes automatic, quantitative evaluation and accurate tracing of the financial data link risk, and greatly improves positioning efficiency and management precision of the data distortion problem.

Description

Technical Field

[0001] This invention relates to the field of financial data quality monitoring technology, specifically an intelligent evaluation system and method applied to the financial data chain. Background Technology

[0002] With the deepening of enterprise digital transformation, the generation and flow of financial data have become increasingly frequent, forming a multi-level, multi-stage financial data chain from raw business transaction data to final financial statement data. Financial data exhibits dependencies within this data chain; for example, financial statement data depends on account summary data, account summary data depends on underlying accounting voucher data, and accounting voucher data depends on the front-end business transaction data. However, when the data chain is long, due to the numerous data flow stages and complex dependencies, existing technologies struggle to quickly and accurately pinpoint the initial stage and specific cause of errors when derived data distortion is detected. Extensive manual investigation is required, which is time-consuming and labor-intensive. Therefore, an intelligent evaluation system and method for financial data chains are needed. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent evaluation system and method applied to financial data links to solve the problems raised in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: an intelligent evaluation method applied to financial data links, the intelligent evaluation method comprising:

[0005] Retrieve all operation records related to financial data from the operation log of the financial data storage system, and extract information from each operation record; based on the extracted operation record information, trace the data flow path of each target financial data.

[0006] The data flow path of each target financial data is sorted out according to the flow time to generate a financial data flow timeline; the error propagation probability and data matching degree between each flow node on the timeline are calculated; and the data distortion risk value of each flow node is calculated by combining the error propagation probability and data matching degree.

[0007] Identify the target flow node and trace its upstream flow nodes and flow links;

[0008] Calculate the distortion impact range of each upstream circulation node of the target circulation node, construct a distortion source scoring model, and locate the initial circulation node and the cause of data distortion.

[0009] Furthermore, when analyzing the data flow path of each target financial data, a reverse mapping technique is used to generate a data flow path expression for the target financial data. The data flow path expression is Way={target financial data→operation type→related data→operation type→original business data}; wherein, the operation type includes collection operation, summary operation, and transformation operation.

[0010] In the above steps, reverse mapping technology is the core implementation method in the process of sorting out the flow of target financial data. Its main purpose is to start from the final target financial data (such as financial statement data, account summary data, and other derived data) and trace back the entire process of its generation and flow, ultimately forming a standardized flow data link expression Way={target financial data → operation type → related data → operation type → original business data}. The specific meanings of each component are as follows:

[0011] Target financial data refers to the end-point financial data that requires risk assessment. It can be output data such as financial statements and analysis reports that are directly used for decision-making, or intermediate derived data in the process of circulation (such as account summary data, classification statistics data, etc.).

[0012] Operation type refers to the key action identifier that connects each flow node, clarifying how the data is processed during the flow process;

[0013] Related data refers to the upstream data that the target financial data directly depends on. It is an intermediate node in the data flow chain. It can be intermediate data after the original business data has been preliminarily processed, or it can be derived data (such as when generating gross profit analysis data for a product, related data includes, but is not limited to, the product's sales revenue data, cost data, etc.).

[0014] By using reverse mapping technology and standardized link expressions, the correspondence and operational logic between target data and related data, as well as original business data, can be clearly marked, and the flow trajectory of data from source to terminal can be fully restored, avoiding the difficulty of tracing the source due to distortion caused by link breakage.

[0015] Furthermore, the data acquisition operation refers to the operation of obtaining raw transaction data from the business system; the data aggregation operation refers to the operation of calculating multiple basic financial data according to preset accounting rules; and the data conversion operation refers to the operation of standardizing and adjusting the format of financial data.

[0016] In the above steps, "data collection" refers to the process of extracting raw, unprocessed transaction data from various business systems of an enterprise (such as sales management systems, procurement management systems, expense reimbursement systems, and fund payment systems). Raw transaction data is the source of the financial data chain and has authenticity and originality. Examples include sales order data, purchase invoice data, employee expense reimbursement voucher data, and bank transfer record data. Data collection does not change the content of the data itself; it only enables data access from business scenarios to the financial data system.

[0017] "Summary operation" refers to the operation of merging, accumulating, and integrating multiple basic financial data of the same dimension and type according to preset calculation rules and statistical standards (such as subject classification, department classification, project aggregation, time period division, etc.). For example, the monthly sales revenue data of each sales region can be summarized into the company's total monthly sales revenue by the main business revenue subject, and the office expense data of each department can be summarized into the company's total monthly office expenses by the management expense subject. The core of the summary operation is to realize the aggregation and summarization of data to form more comprehensive financial data.

[0018] "Conversion operation" refers to the standardization operation of financial data from different sources and in different formats to unify the format; for example, converting CSV format data exported from the sales system into JSON format data compatible with the financial system, adjusting business data statistically based on tax-inclusive amount to accounting rules based on tax-exclusive amount according to accounting standards, and converting data statistically based on calendar month into data statistically based on accounting period (such as quarter or year). The conversion operation does not change the core business meaning of the data, but only realizes the adaptation and compatibility of data in different scenarios.

[0019] Furthermore, the financial data flow timeline is generated by sorting all flow nodes in the target financial data flow data link in ascending order of flow timestamps; each flow node is marked with a unique corresponding flow timestamp t. i , t i This represents the timestamp of the i-th transfer node;

[0020] The above steps clarify the generation logic of the financial data flow timeline. Its main purpose is to structurally organize the entire financial data flow chain through the time dimension. The generation logic of the financial data flow timeline is based on the complete flow data chain of the target financial data. It extracts all nodes (i.e., flow nodes) involved in the data generation, processing, and flow in the chain, and sorts them in ascending order according to the flow timestamps corresponding to each flow node (arranged from earliest to latest time). Finally, a visualized and orderly financial data flow timeline is formed. The flow timestamp is the core identifier of each flow node, accurately recording the specific time when the financial data corresponding to the flow node completes processing (such as collection, summary, and transformation) and enters the next stage (such as the time when the collection operation is completed, the time when the summary calculation ends, and the time when the transformation adjustment takes effect).

[0021] Furthermore, the error propagation probability between two adjacent flow nodes is calculated as follows:

[0022] ;

[0023] Where P i Let n represent the error propagation probability between the i-th and (i+1)-th flow nodes, k represent the weight coefficient corresponding to the operation type, and n represent the error propagation probability between the i-th and (i+1)-th flow nodes. j Let t represent the cumulative total number of operations of the j-th operation type at the i-th flow node, N represent the cumulative total number of operations of all operation types at the i-th flow node, and t represent the cumulative total number of operations of all operation types at the i-th flow node. i Let t be the generation timestamp of the i-th transfer node. i+1 This is the generation timestamp of the (i+1)th transfer node;

[0024] The error propagation probability P in the above steps i This measures the likelihood that erroneous or distorted data from the i-th flow node (preceding node) will propagate to the (i+1)-th flow node (following node); a higher value indicates a higher risk of error propagation. The weighting coefficient k for each operation type is set according to the risk level of errors introduced by different operation types (e.g., data collection operations directly relate to the original data, resulting in the widest impact of errors, so weighting coefficient k=3; aggregation and transformation operations have relatively lower risks, so weighting coefficient k=2), used to distinguish the risk levels of different operation types; n j This parameter reflects the maturity and stability of the j-th type of operation. The more times the operation is performed, the more mature the operation process becomes and the lower the probability of error. This parameter can be used to correct the error propagation probability in reverse.

[0025] Furthermore, the data matching degree is calculated in the following way:

[0026] ;

[0027] Where Mi S represents the data matching degree between the i-th flow node and the (i+1)-th flow node. i C represents the data fit between two adjacent flow nodes. i This represents the credibility of the i-th transfer node;

[0028] Among them, C i The value of C is set based on the cumulative total frequency of the flow node. The higher the cumulative total frequency of the flow node, the higher the value of C. i The larger the value, the higher the reliability of the corresponding transfer node;

[0029] Among them, S i This is a weighted composite value of data numerical overlap and consistency with calculation rules; data numerical overlap is calculated by extracting corresponding data fields from two adjacent flow nodes and using a relative error rate correction method.

[0030] ;

[0031] Where S1 represents the data numerical overlap, x i Let x be the target data value of the i-th transfer node. i+1 Let x be the associated data value corresponding to the (i+1)th flow node; if x i or x i+1 If the result is 0, take another non-zero value as the denominator; if both are 0, then S1 = 1.

[0032] Consistency in calculation rules is achieved by analyzing the data processing rules of adjacent flow nodes and using a rule matching method.

[0033] ;

[0034] Where S2 represents the consistency of calculation rules, A represents the number of matched rule items, and B represents the total number of rule items; where a rule item represents the calculation logic involved in data calculation, such as summation, average calculation, etc.

[0035] Si is obtained by weighted summation of data numerical overlap and calculation rule consistency:

[0036] ;

[0037] in, , These are the weights corresponding to the overlap of data values ​​and the consistency of calculation rules, respectively.

[0038] Furthermore, the data distortion risk value is calculated by considering all upstream flow nodes e that point to flow node e. i Calculate the upstream flow node e respectively i Error propagation probability Pi Data mismatch degree 1−M i The product of Q is obtained by summing all the product results; where Q e The data distortion risk value for node e is e. i This represents all upstream flow nodes e that provide data input to flow node e. i P i For each upstream transfer node e of transfer node e i The error propagation probability between M i For each upstream transfer node e of transfer node e i Data matching degree between them, 1-M i Characterizing each upstream flow node e i Data mismatch between them;

[0039] In the aforementioned financial data flow chain, data distortion at the terminal or intermediate flow nodes is often not caused by a single upstream flow node, but rather by the combined effect of multiple upstream flow nodes. By accumulating the risk contribution values ​​of all upstream flow nodes, the comprehensive risk from multiple sources can be quantified, making the risk assessment more in line with the actual scenario of multi-node dependence and multi-path transmission in the financial data flow chain, and avoiding risk misjudgment caused by single-dimensional assessment.

[0040] Furthermore, when identifying target flow nodes, a data distortion risk threshold F is preset, all flow nodes are traversed, and the distortion risk value Q of each flow node is compared with the data distortion risk threshold F. If the distortion risk value Q of the flow node is greater than or equal to F, then the flow node is marked as the target flow node.

[0041] Among them, the data distortion risk threshold F is determined by analyzing the risk value distribution of past data distortion cases in the system's financial data (such as the lowest risk value of historical high-risk distortion nodes, the risk value range of common distortion nodes, etc.) to ensure that the threshold F matches the risk characteristics of the actual operation of the system.

[0042] The aforementioned financial data flow chain typically includes several flow nodes. If no nodes are screened and all nodes are traced and investigated directly, it will result in indiscriminate investment of manpower and time costs, which is extremely inefficient. By identifying target flow nodes, high-risk nodes with distortion risk values ​​exceeding the acceptable range can be quickly identified, and subsequent work can be focused on core risk objects, avoiding wasting resources on low-risk nodes and significantly improving the efficiency of data quality control.

[0043] Furthermore, when calculating the range of distortion influence, the calculation formula is C. i =Q i ×m; where C i Q represents the distortion impact range of the i-th associated flow node. irepresents the distortion risk value of the i-th associated circulation node, and m represents the number of downstream associated circulation nodes of the i-th associated circulation node;

[0044] The constructed distortion source scoring model is G. i =(1-M i )×P i ×C i Among them, G i M represents the distortion score of the i-th associated flow node. i P represents the data matching degree between the i-th associated flow node and the downstream flow node. i This represents the error propagation probability from the i-th associated flow node to the downstream flow node;

[0045] The distortion scores of all related flow nodes are sorted in descending order. The flow node with the highest distortion score is the initial stage of data distortion. The cause of distortion is determined by combining the operation records of the flow node.

[0046] The above steps standardize the tracing process into calculation, scoring, sorting, location, and cause finding, eliminating the need for manual investigation of all nodes one by one. At the same time, through the correlation analysis between operation records and distorted nodes, the root cause of the distortion can be directly pointed to (such as incorrect collection rules or data matching deviations), avoiding the ineffective work of repeatedly checking the cause after the location step, and greatly shortening the problem-solving cycle.

[0047] Furthermore, to better implement the above method, an intelligent evaluation system for financial data links is also provided. This intelligent evaluation system includes: an information acquisition module, a data link sorting module, a timeline generation and calculation module, a target node identification and tracing module, and a distortion source location module.

[0048] The information extraction module is used to retrieve all operation records related to financial data in the operation log of the financial data storage system and extract information from each operation record.

[0049] The data link analysis module is used to analyze the flow data link of each target financial data based on the extracted operation record information.

[0050] The timeline generation and calculation module is used to sort out the data flow links of each target financial data according to the flow time and generate a financial data flow timeline; calculate the error propagation probability and data matching degree between each flow node on the timeline; and calculate the data distortion risk value of each flow node by combining the error propagation probability and data matching degree.

[0051] The target node identification and tracing module is used to identify target circulation nodes and trace the upstream circulation nodes and circulation links of the target circulation nodes.

[0052] The distortion source localization module is used to calculate the distortion impact range of each upstream circulation node of the target circulation node, construct a distortion source scoring model, and locate the initial circulation node and the cause of data distortion.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] 1. Traditional financial data problem investigation relies on manual backtracking, which is time-consuming and prone to omissions. This invention automatically extracts operation logs, generates standardized data links through reverse mapping, and dynamically calculates error propagation probability and data matching degree based on the time axis. This enables a quantitative assessment of data distortion risk. The system can automatically identify high-risk nodes, locate the source of distortion through a scoring model, and analyze the causes. This overcomes the time-consuming nature of manual investigation and significantly improves the efficiency and accuracy of data governance.

[0055] 2. Addressing the multi-node, multi-path, and multi-rule characteristics of financial data flow, this invention integrates multiple factors, including operation type risk weights, historical operation frequency, time intervals, data consistency, rule matching degree, and the scope of node influence, to establish a dynamic risk assessment model. This model not only reflects the risk of a single link but also quantifies the cumulative effect of errors from multiple sources, making the assessment results more comprehensive and more consistent with the complexities of data dependence and risk transmission in actual business operations, supporting more precise decision-making interventions.

[0056] 3. This invention proposes a unified data link expression and a structured traceability process, which transforms scattered operation records into a clear and traceable flow map. Through steps such as threshold screening, scoring and sorting, and operation record association, a closed-loop processing mechanism from risk identification to cause location is constructed. This not only improves the interpretability and operability of the system, but also facilitates its portability and reuse in different enterprise or business systems, providing a systematic and engineered solution for financial data quality monitoring. Attached Figure Description

[0057] Figure 1 This is a schematic diagram of the method flow of an intelligent evaluation system and method applied to financial data links according to the present invention;

[0058] Figure 2 This is a schematic diagram illustrating an application scenario of the intelligent evaluation system and method of the present invention applied to the financial data link. Detailed Implementation

[0059] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0060] Example 1: As Figure 1As shown, the present invention provides a technical solution, an intelligent evaluation method applied to financial data links, the intelligent evaluation method comprising:

[0061] Retrieve all operation records related to financial data from the operation log of the financial data storage system, and extract information from each operation record; based on the extracted operation record information, trace the data flow path of each target financial data.

[0062] The data flow path of each target financial data is sorted out according to the flow time to generate a financial data flow timeline; the error propagation probability and data matching degree between each flow node on the timeline are calculated; and the data distortion risk value of each flow node is calculated by combining the error propagation probability and data matching degree.

[0063] Identify the target flow node and trace its upstream flow nodes and flow links;

[0064] Calculate the distortion impact range of each upstream circulation node of the target circulation node, construct a distortion source scoring model, and locate the initial circulation node and the cause of data distortion;

[0065] Specifically, when analyzing the data flow path of each target financial data, a reverse mapping technique is used to generate a data flow path expression for the target financial data. The data flow path expression is Way={target financial data→operation type→related data→operation type→original business data}; whereby the operation type includes collection operation, summary operation, and transformation operation.

[0066] The data acquisition operation involves obtaining raw transaction data from the business system; the data aggregation operation involves calculating multiple basic financial data according to preset accounting rules; and the data conversion operation involves standardizing and adjusting the format of the financial data.

[0067] The financial data flow timeline is generated by sorting all flow nodes in the target financial data flow data link in ascending order of flow timestamps; each flow node is marked with a unique corresponding flow timestamp t. i , t i This represents the timestamp of the i-th transfer node;

[0068] The error propagation probability between two adjacent flow nodes is calculated as follows:

[0069] ;

[0070] Where P i Let n represent the error propagation probability between the i-th and (i+1)-th flow nodes, k represent the weight coefficient corresponding to the operation type, and n represent the error propagation probability between the i-th and (i+1)-th flow nodes. jLet t represent the cumulative total number of operations of the j-th operation type at the i-th flow node, N represent the cumulative total number of operations of all operation types at the i-th flow node, and t represent the cumulative total number of operations of all operation types at the i-th flow node. i Let t be the generation timestamp of the i-th transfer node. i+1 This is the generation timestamp of the (i+1)th transfer node;

[0071] The data matching degree is calculated in the following way:

[0072] ;

[0073] Where M i S represents the data matching degree between the i-th flow node and the (i+1)-th flow node. i C represents the data fit between two adjacent flow nodes. i This represents the credibility of the i-th transfer node;

[0074] Among them, the data distortion risk value is determined by considering all upstream flow nodes e that point to flow node e. i Calculate the upstream flow node e respectively i Error propagation probability P i Data mismatch degree 1−M i The product of Q is obtained by summing all the product results; where Q e The data distortion risk value for node e is e. i This represents all upstream flow nodes e that provide data input to flow node e. i P i For each upstream transfer node e of transfer node e i The error propagation probability between M i For each upstream transfer node e of transfer node e i Data matching degree between them, 1-M i Characterizing each upstream flow node e i Data mismatch between them;

[0075] When identifying a target flow node, a data distortion risk threshold F is preset. All flow nodes are traversed and the distortion risk value Q of each flow node is compared with the data distortion risk threshold F. If the distortion risk value Q of the flow node is greater than or equal to F, the flow node is marked as the target flow node.

[0076] When calculating the range of distortion influence, the calculation formula is C. i =Q i ×m; where C i Q represents the distortion impact range of the i-th associated flow node. irepresents the distortion risk value of the i-th associated circulation node, and m represents the number of downstream associated circulation nodes of the i-th associated circulation node;

[0077] The distortion source scoring model is G. i =(1-M i )×P i ×C i Among them, G i M represents the distortion score of the i-th associated flow node. i P represents the data matching degree between the i-th associated flow node and the downstream flow node. i This represents the error propagation probability from the i-th associated flow node to the downstream flow node;

[0078] The distortion scores of all associated flow nodes are sorted in descending order. The flow node with the highest distortion score is the initial stage of data distortion. The cause of distortion is determined by combining the operation records of the flow node.

[0079] Example 2: Figure 2 As shown, in order to better implement the above method, an intelligent evaluation system for financial data links is also provided. The intelligent evaluation system includes: an information acquisition module, a data link sorting module, a time axis generation and calculation module, a target node identification and tracing module, and a distortion source location module.

[0080] The information extraction module is used to retrieve all operation records related to financial data in the operation log of the financial data storage system and extract information from each operation record.

[0081] The data link analysis module is used to analyze the flow data link of each target financial data based on the extracted operation record information.

[0082] The timeline generation and calculation module is used to sort out the data flow links of each target financial data according to the flow time and generate a financial data flow timeline; calculate the error propagation probability and data matching degree between each flow node on the timeline; and calculate the data distortion risk value of each flow node by combining the error propagation probability and data matching degree.

[0083] The target node identification and tracing module is used to identify target circulation nodes and trace the upstream circulation nodes and circulation links of the target circulation nodes.

[0084] The distortion source localization module is used to calculate the distortion impact range of each upstream circulation node of the target circulation node, construct a distortion source scoring model, and locate the initial circulation node and the cause of data distortion.

[0085] In one embodiment of the present invention, the finance department discovers an anomaly in the "Operating Revenue" item data of the month-end consolidated profit statement, which differs significantly from the sales figures compiled by the business system. To pinpoint the initial stage and cause of the data distortion, the intelligent evaluation method of the present invention is employed, specifically implemented as follows:

[0086] Retrieve all operation log records related to the "Operating Revenue" account from the financial data storage system, including data collection, summarization, and conversion operations; using reverse mapping technology, starting from the "Operating Revenue" account data, trace its data link backwards to generate a standardized link expression Way={Operating Revenue → Summarization Operation → Gross Profit of Each Product → Collection Operation → Original Sales Order Data}, clearly marking the complete flow path and operation type from the final report data to the original business data;

[0087] Based on the timestamps of each flow node, a financial data flow timeline is generated in ascending order; the error propagation probability between adjacent nodes is calculated, with the weight coefficient of the data collection operation set to 3 and the weight coefficient of the data aggregation operation set to 2, and the results are adjusted based on the cumulative number of operations and time intervals of each node; at the same time, the data matching degree is calculated, and the consistency and reliability of the data between nodes are comprehensively evaluated by comparing the relative error of the data values ​​of adjacent nodes and the consistency of the calculation rules.

[0088] By multiplying and summing the error propagation probability and the data mismatch degree, the data distortion risk value Q of the "Operating Revenue" node and its upstream nodes is calculated; the risk threshold F is set as the high-risk critical value in historical cases. After traversing, it is found that the risk value Q of the "Product A Gross Profit Summary" node is significantly higher than the threshold F, so it is marked as the target flow node.

[0089] For this target flow node, trace its associated data collection nodes upstream (such as product A sales revenue collection, product A sales cost collection, etc.), and calculate the distortion impact range C of each upstream flow node, using the formula C. i = Q i ×m, where m is the number of downstream associated nodes; and then a distortion source scoring model G is constructed. i = (1-M i )×P i ×C i The upstream nodes are scored and ranked.

[0090] The scoring results showed that the "Product A Sales Revenue Collection" node had the highest distortion score, G. Analysis of its operation records revealed that due to a system interface configuration error during data collection, the tax amount of some orders was not correctly separated, resulting in the tax-inclusive inclusion of the collected sales revenue data, which triggered a chain reaction of deviations in the subsequent aggregation process. Thus, the source of the distortion was accurately located in the data collection stage, and the specific configuration error was identified.

[0091] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent evaluation method applied to financial data links, characterized in that: The intelligent evaluation method includes: Retrieve all operation records related to financial data from the operation log of the financial data storage system, and extract information from each operation record; based on the extracted operation record information, trace the data flow path of each target financial data. The data flow path of each target financial data is sorted out according to the flow time to generate a financial data flow timeline; the error propagation probability and data matching degree between each flow node on the timeline are calculated; and the data distortion risk value of each flow node is calculated by combining the error propagation probability and data matching degree. Identify the target flow node and trace its upstream flow nodes and flow links; Calculate the distortion impact range of each upstream circulation node of the target circulation node, construct a distortion source scoring model, and locate the initial circulation node and the cause of data distortion.

2. The intelligent evaluation method applied to financial data links according to claim 1, characterized in that: When analyzing the data flow path of each target financial data item, a reverse mapping technique is used to generate a data flow path expression for the target financial data. The data flow path expression is Way={target financial data→operation type→related data→operation type→original business data}; where the operation type includes collection operation, summary operation, and transformation operation.

3. The intelligent evaluation method applied to financial data links according to claim 2, characterized in that: The data acquisition operation is the operation of obtaining raw transaction data from the business system; the data aggregation operation is the operation of calculating multiple basic financial data according to preset accounting rules; the data conversion operation is the operation of standardizing and adjusting the format of financial data.

4. The intelligent evaluation method applied to financial data links according to claim 1, characterized in that: The financial data flow timeline is generated by sorting all flow nodes in the target financial data flow data link in ascending order of flow timestamps; each flow node is marked with a unique corresponding flow timestamp t. i , t i This represents the timestamp of the i-th transfer node.

5. The intelligent evaluation method applied to financial data links according to claim 1, characterized in that: The error propagation probability between two adjacent flow nodes is calculated as follows: ； Where P i Let n represent the error propagation probability between the i-th and (i+1)-th flow nodes, k represent the weight coefficient corresponding to the operation type, and n represent the error propagation probability between the i-th and (i+1)-th flow nodes. j Let t represent the cumulative total number of operations of the j-th operation type at the i-th flow node, N represent the cumulative total number of operations of all operation types at the i-th flow node, and t represent the cumulative total number of operations of all operation types at the i-th flow node. i Let t be the generation timestamp of the i-th transfer node. i+1 This is the generation timestamp of the (i+1)th transfer node.

6. The intelligent evaluation method applied to financial data links according to claim 1, characterized in that: Data matching degree is calculated in the following way: ； Where M i S represents the data matching degree between the i-th flow node and the (i+1)-th flow node. i C represents the data fit between two adjacent flow nodes. i This represents the credibility of the i-th transfer node.

7. The intelligent evaluation method applied to financial data links according to claim 1, characterized in that: The data distortion risk value is determined by considering all upstream flow nodes e that point to flow node e. i Calculate the upstream flow node e respectively. i Error propagation probability P i Data mismatch degree 1−M i The product of Q is obtained by summing all the product results; where Q e The data distortion risk value for node e is e. i This represents all upstream flow nodes e that provide data input to flow node e. i P i For each upstream transfer node e of transfer node e i The error propagation probability between M i For each upstream transfer node e of transfer node e i Data matching degree between them, 1-M i Characterizing each upstream flow node e i The degree of data mismatch between them.

8. The intelligent evaluation method applied to financial data links according to claim 1, characterized in that: When identifying a target flow node, a data distortion risk threshold F is preset. All flow nodes are traversed and the distortion risk value Q of each flow node is compared with the data distortion risk threshold F. If the distortion risk value Q of the flow node is greater than or equal to F, then the flow node is marked as the target flow node.

9. The intelligent evaluation method applied to financial data links according to claim 1, characterized in that: When calculating the range of distortion influence, the formula is C. i =Q i ×m; where C i Q represents the distortion impact range of the i-th associated flow node. i represents the distortion risk value of the i-th associated circulation node, and m represents the number of downstream associated circulation nodes of the i-th associated circulation node; The distortion source scoring model is G. i =(1-M i )×P i ×C i Among them, G i M represents the distortion score of the i-th associated flow node. i P represents the data matching degree between the i-th associated flow node and the downstream flow node. i This represents the error propagation probability from the i-th associated flow node to the downstream flow node; The distortion scores of all associated flow nodes are sorted in descending order. The flow node with the highest distortion score is the initial stage of data distortion. The cause of distortion is determined by combining the operation records of the flow node.

10. An intelligent evaluation system applied to a financial data link, used to execute the intelligent evaluation method applied to a financial data link as described in any one of claims 1-9, characterized in that: The intelligent assessment system includes: an information acquisition module, a data link analysis module, a timeline generation and calculation module, a target node identification and tracing module, and a distortion source location module. The information extraction module is used to retrieve all operation records related to financial data in the operation log of the financial data storage system, and to extract information from each operation record. The data link sorting module is used to sort out the flow data link of each target financial data based on the extracted operation record information; The timeline generation and calculation module is used to sort out the flow data link of each target financial data according to the flow time, generate a financial data flow timeline; calculate the error propagation probability and data matching degree between each flow node on the timeline; and combine the error propagation probability and data matching degree to calculate the data distortion risk value of each flow node. The target node identification and tracing module is used to identify target circulation nodes and trace the upstream circulation nodes and circulation links of the target circulation nodes. The distortion source localization module is used to calculate the distortion impact range of each upstream circulation node of the target circulation node, construct a distortion source scoring model, and locate the initial circulation node and the cause of data distortion.

Images