Process approval system and method based on big data and computer program product
By generating structured summaries and calculating risk indexes through the big data process approval system, the problems of inefficiency and insufficient risk control in enterprise process approval are solved, and rapid decision-making and risk identification are achieved.
Patent Information
- Application Number
- CN202510884905.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-30
- Publication Date
- 2025-10-17
AI Technical Summary
The existing enterprise process approval system relies on manual review of scattered data, resulting in inefficiency, insufficient risk control, difficulty in quickly locating key information and conducting structured integration, and lack of automated analysis of historical risk patterns.
A process approval system based on big data is adopted, including data layer, rule engine layer and algorithm layer. Structured summaries are generated through the rule factor library, and risk indexes are calculated by combining machine learning, dynamically integrating multi-source data for approval decisions.
Significantly reduce information retrieval time, improve approval efficiency, enhance risk control capabilities, mark potential risks in real time, and assist in quickly identifying anomalies.
Smart Images
Figure CN120806616A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of enterprise process approval automation, in particular to a process approval system and method based on big data and a computer program product. BACKGROUND
[0002] In the current enterprise process approval system, approval decisions usually rely on manual review of scattered historical data (such as work order records, supplier archives, contract terms, etc.), resulting in low efficiency and insufficient risk control. The existing technology mainly has the following problems: 1. The approver needs to manually search multiple data sources (such as ERP, CRM, financial system) to obtain complete approval basis, but due to scattered data and lack of intelligent summary, it is difficult to quickly locate key information (such as supplier historical performance rate, budget deviation), which affects decision-making speed.
[0003] 2. The existing system does not structurally integrate approval indicators (such as cost budget, supplier credit rating, compliance requirements), and the approver needs to check the data one by one, which not only consumes time but also is prone to misjudgment due to human negligence. For example, in the procurement approval, the contract amount, historical transaction records and risk events need to be repeatedly compared, and the process is cumbersome.
[0004] 3. Insufficient risk control: the current system relies on the subjective experience of the approver to judge the risk, lacks automated analysis of historical risk patterns (such as a supplier frequently delaying delivery, a specific department's budget overspending trend), and potential risks are ignored. For example, manual review may miss the supplier's past compliance violation records associated with the supplier, increasing the risk of enterprise operation.
[0005] Therefore, there is an urgent need for an intelligent approval decision-making method that dynamically integrates multi-source data, automatically generates structured summaries, and combines historical risk pattern analysis to improve approval efficiency and risk control capability. SUMMARY
[0006] The present application aims to at least one of the above technical problems in the technical field, and proposes a process approval system based on big data, comprising: a data layer, a rule engine layer and an algorithm layer; wherein, the algorithm layer is configured to: based on the rule factor library constructed by the rule engine layer, process the target process information received from the data layer to generate a structured summary; based on the target process information, respectively calculate a first risk index and a second risk index of the target process based on the rule factor library and machine learning, and dynamically calculate the comprehensive risk index of the target process; and perform process approval according to the comprehensive risk index.
[0007] Further, the rule factor library is a mapping relationship between an approval node, a core rule factor, a rule trigger condition, and a data source of the approval node; the core rule factor includes a core parameter corresponding to the approval node, and the rule trigger condition includes whether the core parameter exceeds a threshold value.
[0008] Further, the algorithm layer calculates a first risk index of the target process, and the method includes: determining a core parameter corresponding to the target process information based on the rule factor library; obtaining current data and historical data of each core parameter, and substituting the current data and the historical data into a first risk index calculation formula to obtain the first risk index of the target process; wherein the first risk index calculation formula is:
[0009] wherein, the first risk index is represented by R; the total number of the core parameters corresponding to the target process information is represented by N; the sub risk index corresponding to the i-th core parameter is represented by Ri; the dynamic weight of the sub risk index is represented by wi; the current value of the i-th core parameter is represented by xi; the historical mean value of the i-th core parameter is represented by xi, mean; the historical standard deviation of the current value of the i-th core parameter is represented by xi, std.
[0010] Further, the formula of the dynamic weight of the sub risk index is:
[0011] wherein, the sub risk index weight at the moment t is represented by wt; the initial weight of the i-th sub risk index is represented by wi, 0; the adjustment coefficient is represented by a; the mean value of the historical value of the i-th sub risk index is represented by wi, mean; the maximum value in the historical value of the i-th sub risk index is represented by wi, max; the minimum value in the historical value of the i-th sub risk index is represented by wi, min.
[0012] Furthermore, the algorithm layer is configured to determine the initial weight based on the hierarchical analysis method, and the corresponding method includes: Construct a hierarchical model corresponding to risk index weights; Calculating the eigenvector of the judgment matrix corresponding to the hierarchical structure model; A consistency check is performed on the judgment matrix. If the check result is passed, the judgment matrix is subjected to a hierarchical total sorting to obtain an initial weight of each sub-risk index.
[0013] Furthermore, the algorithm layer calculates a second risk index of the target process based on machine learning, and the execution method includes: inputting the first risk index into a trained LSTM model or XGBoost model to calculate the second risk index of the target process.
[0014] Furthermore, the process approval system based on big data also includes: The interaction layer is configured as follows: Visualizing the comprehensive risk index, including: displaying the risk level and highlighting the sub-risk indices exceeding the threshold; and Receive the approval command sent by the user and send it to the algorithm layer, so that the algorithm layer performs the approval operation according to the approval command; which includes: generating corresponding approval opinions based on the over-threshold sub-risk index when the approval is not passed.
[0015] This application also proposes a process approval method based on big data, including: Processing target process information according to a rule factor library to generate a structured summary; the rule factor library is a mapping relationship between approval nodes, core rule factors, rule triggering conditions, and approval node data sources; According to the target process information, respectively calculating a first risk index and a second risk index of the target process based on the rule factor library and machine learning, and weighting the two to calculate a comprehensive risk index of the target process; and Process approval is carried out based on the comprehensive risk index.
[0016] Furthermore, the first risk index calculation formula is:
[0017] in, represents the first risk index; Indicates the total number of core parameters corresponding to the target process information; Indicates the The sub-risk index corresponding to each core parameter; express Dynamic weight of represents the current value of the core parameter; represents the historical average value of the core parameter; represents the historical standard deviation of the current value of the core parameter; The dynamic weight formula of the sub-risk index is:
[0018] wherein, represents the sub-risk index weight at the moment; represents the initial weight of the sub-risk index; represents the adjustment coefficient; represents the average value of the historical value of the sub-risk index; represents the maximum value in the historical value of the sub-risk index; represents the minimum value of the historical value of the sub-risk index.
[0019] The application also provides a computer program product stored in a computer readable storage medium, which is executed by a processor to at least realize the above-mentioned process approval method based on big data.
[0020] Compared with the prior art, the application has the following beneficial effects: The application automatically generates a structured approval abstract based on a rule engine, significantly reduces the information retrieval time of the approver, and avoids information overload. In addition, the application also calculates the comprehensive risk index of the target process based on the machine learning model and the analytic hierarchy process, marks potential risks in real time, assists the approver in quickly identifying abnormalities, and enhances the enterprise risk control capability.
[0021] Other features and advantages of the present application will be described in the following description, and some will become apparent from the description, or will be learned by practice of the present application. The purpose and other advantages of the present application can be achieved and obtained by the structure specifically pointed out in the written description and the accompanying drawings. The technical solutions of the present application will be further described below by the accompanying drawings and examples. BRIEF DESCRIPTION OF DRAWINGS
[0022] The accompanying drawings are used to provide a further understanding of the present application, and constitute a part of the specification, and are used to explain the present application together with the embodiments of the present application, and do not constitute a limitation to the present application. In the drawings: Figure 1 A schematic diagram of a big data-based process approval system given for an embodiment; Figure 2 A schematic diagram of a comprehensive risk index calculation method given for an embodiment; Figure 3 A schematic diagram of an interactive layer interface given for an embodiment; Figure 4 A schematic diagram of a big data-based process approval system given for another embodiment; Figure 5 A working principle diagram of a big data-based process approval system given for an embodiment; Figure 6 A schematic diagram of a big data-based process approval method given for an embodiment; Figure 7 A schematic diagram of a computer-readable storage medium given for an embodiment. DETAILED DESCRIPTION
[0023] The application will be described in conjunction with the accompanying drawings, in which the preferred embodiments described herein are used to illustrate and explain the application, but not to limit the application.
[0024] Figure 1 A big data-based process approval system given for the application, comprising: a data layer, a rule engine layer, and an algorithm layer; wherein, the algorithm layer is configured to: based on a rule factor library constructed by the rule engine layer, process target process information received from the data layer to generate a structured summary; based on the rule factor library and machine learning, respectively calculate a first risk index and a second risk index of the target process according to the target process information, and dynamically calculate a comprehensive risk index of the target process based on the first risk index and the second risk index; and perform process approval according to the comprehensive risk index.
[0025] According to some embodiments of the application, the big data-based process approval system given by the application comprises a data layer, a rule engine layer, and an algorithm layer; wherein the data layer integrates a database of historical work orders, material procurement, financial posting, and other approval process related information; the rule engine layer is used to define and update a rule factor library; and the algorithm layer is used to extract key fields from to-be-approved process data by determining approval nodes of the approval process, generate a structured summary (such as a work order budget comparison table), and calculate a comprehensive risk index of the to-be-approved process by machine learning and the rule factor library.
[0026] Furthermore, the rule factor library is a mapping relationship between approval nodes, core rule factors, rule triggering conditions and approval node data sources; the core rule factors include core parameters of corresponding approval nodes, and the rule triggering conditions include whether the core parameters exceed a threshold.
[0027] According to some embodiments of the present application, the rule factor library construction scheme is shown in Table 1: Table 1
[0028] According to some embodiments of the present application, the rule factor library constructed by the rule engine layer adopts a dynamic maintenance mechanism, which is shown in Table 2: Table 2
[0029] Furthermore, the algorithm layer calculates the first risk index of the target process, and the execution method includes: Determining core parameters corresponding to the target process information based on the rule factor library; Obtain the current data and historical data of each core parameter, and substitute the two into the first risk index calculation formula to obtain the first risk index of the target process; wherein the first risk index calculation formula is:
[0030] in, represents the first risk index; Indicates the total number of core parameters corresponding to the target process information; Indicates the The sub-risk index corresponding to each core parameter; express Dynamic weight of Indicates the The current value of the core parameters; Indicates the The historical mean of the core parameters; Indicates the The historical standard deviation of the current value of a core parameter.
[0031] Furthermore, the formula for the dynamic weight of the sub-risk index is:
[0032] in, express Sub-risk index weight at the moment; Indicates the The initial weight of the item risk index; represents the adjustment coefficient; represents the mean of the historical values of the represents the maximum value of the historical values of the represents the minimum value of the historical values of the
[0033] According to some embodiments of the present application, a single approval process can involve multiple core parameters, each of which corresponds to a sub-risk index. A first risk index can be obtained by weighted sum and average of all sub-risk indexes. In this embodiment, after obtaining the first risk index, it is converted into a percentage score as a rule engine score. For each core parameter , the following calculation formula can be used to calculate its sub-risk index:
[0034] wherein, represents the sub-risk index corresponding to the thcore parameter; represents the current value of the thcore parameter; represents the historical mean of the thcore parameter; represents the historical standard deviation of the thcore parameter.
[0035] According to the above formula of dynamic weight, the first risk index can be obtained by weighted sum of each sub-risk index.
[0036] Further, the algorithm layer is configured to determine the initial weight based on the analytic hierarchy process, and the corresponding method comprises: constructing a hierarchical structure model corresponding to the risk index weight; calculating the eigenvector of the judgment matrix corresponding to the hierarchical structure model; performing consistency test on the judgment matrix, and if the test result is passed, performing hierarchical total sorting on the judgment matrix to obtain the initial weight of each sub-risk index.
[0037] According to some embodiments of the present application, the construction node risk index weight judgment matrix corresponds to safety, progress and cost in turn, and the matrix is as follows:
[0038] The consistency of the judgment matrix is checked, the checking result is passed, and according to the characteristic vector of the matrix, the safety risk weight is about equal to 0.637, the progress risk weight is about equal to 0.258, and the cost risk weight is about equal to 0.105.
[0039] Further, the algorithm layer calculates a second risk index of the target process based on machine learning, and the method comprises: inputting the first risk index into a trained LSTM model or XGBoost model to calculate the second risk index of the target process.
[0040] According to some embodiments of the present application, as shown in Figure 2 As shown in the figure, the present application adopts a hybrid risk model of rule engine + machine learning to calculate the comprehensive risk index, and the process is as follows: when the rule engine score is lower than the threshold, it is determined that the current approval process has risks; when the rule engine score is higher than the threshold, the score is input into the supervised learning model (LSTM or XGBoost), and the rule engine score (first risk index) and the supervised learning model (second risk index) are output for weighted fusion to obtain the comprehensive risk index and determine the final risk level. In this embodiment, the weight of the weighted fusion module is the rule confidence multiplied by the model accuracy, for example, the rule engine score for supplier credit is 70 (in percentage), the rule engine weight (historical accuracy) is 0.7, the LSTM model score is 65, and the LSTM model weight (recent prediction accuracy) is 0.65. The calculation can obtain the comprehensive score FinalRisk as follows: FinalRisk=0.7×70 + 0.65×65 = 94.25 In this embodiment, the threshold of supplier credit is 90, and since the FinalRisk exceeds the threshold, it is determined that the risk level of supplier credit is high.
[0041] Further, the process approval system based on big data further comprises: The interaction layer is configured to: visualize the comprehensive risk index, including: displaying the risk level and highlighting the sub-risk index exceeding the threshold; and receive the approval command sent by the user and send it to the algorithm layer, so that the algorithm layer performs the approval operation according to the approval command; including: generating the corresponding approval opinion according to the sub-risk index exceeding the threshold when the approval is not passed.
[0042] According to some embodiments of the present application, as shown in Figure 3As shown, the interactive layer interface includes: an approval interface, which is further divided into a structured summary and corresponding pass and reject buttons, the structured summary shows a budget comparison table, supplier risk and engineering team risk. Among them, the supplier risk and the engineering team risk are displayed in the form of radar chart.
[0043] According to some embodiments of the present application, as Figure 4 shown, the process approval system based on big data provided by the present application includes: a data layer, a rule engine layer, an algorithm layer and an interactive layer. The workflow of the system is as Figure 5 shown, including: inputting information related to the process to be approved, performing rule matching on the information related to the process to be approved based on a rule factor library, generating a summary according to the rule matching result, and calculating a risk index.
[0044] Based on the same technical concept, as Figure 6 shown, the present application also provides a process approval method based on big data, which includes: processing target process information according to a rule factor library to generate a structured summary; the rule factor library is a mapping relationship between an approval node, a core rule factor, a rule trigger condition and a data source of the approval node; based on the rule factor library and machine learning, respectively calculating a first risk index and a second risk index of the target process according to the target process information, and weighting the two to calculate a comprehensive risk index of the target process; and performing process approval according to the comprehensive risk index.
[0045] As Figure 7 shown, the present application provides a computer readable storage medium, which stores a computer program or instructions, and the computer program or instructions are executed by a processor to at least implement the above-mentioned process approval method based on big data.
[0046] The present application also provides a computer program product stored in a computer readable storage medium, which is executed by a processor to at least implement the above-mentioned process approval method based on big data.
[0047] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Therefore, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application also intends to include these modifications and variations.
Claims
1. A process approval system based on big data, characterized by: include: Data layer, rule engine layer and algorithm layer; among them, The algorithm layer is configured as follows: Based on the rule factor library constructed by the rule engine layer, the target process information received from the data layer is processed to generate a structured summary; According to the target process information, respectively calculating a first risk index and a second risk index of the target process based on the rule factor library and machine learning, and weighting the two to calculate a comprehensive risk index of the target process; and Process approval is carried out based on the comprehensive risk index.
2. The system according to claim 1, wherein The rule factor library is a mapping relationship between approval nodes, core rule factors, rule triggering conditions and approval node data sources; the core rule factors include core parameters of corresponding approval nodes, and the rule triggering conditions include whether the core parameters exceed a threshold.
3. The system according to claim 2, wherein: The algorithm layer calculates the first risk index of the target process, and the execution method includes: Determining core parameters corresponding to the target process information based on the rule factor library; Obtain the current data and historical data of each core parameter, and substitute the two into the first risk index calculation formula to obtain the first risk index of the target process; wherein the first risk index calculation formula is: in, represents the first risk index; Indicates the total number of core parameters corresponding to the target process information; Indicates the The sub-risk index corresponding to each core parameter; express Dynamic weight of Indicates the The current value of the core parameters; Indicates the The historical mean of the core parameters; Indicates the The historical standard deviation of the current value of a core parameter.
4. The system according to claim 3, wherein: The formula for the dynamic weight of the sub-risk index is: in, express Sub-risk index weight at the moment; Indicates the The initial weight of the item risk index; represents the adjustment coefficient; Indicates the The average of the historical values of the item risk index; Indicates the The maximum value among the historical values of the item risk index; Indicates the The minimum historical value of the item risk index.
5. The system according to claim 4, wherein: The algorithm layer is configured to determine the initial weight based on the hierarchical analysis method, and the corresponding method includes: Construct a hierarchical model corresponding to risk index weights; Calculating the eigenvector of the judgment matrix corresponding to the hierarchical structure model; A consistency check is performed on the judgment matrix. If the check result is passed, the judgment matrix is subjected to a hierarchical total sorting to obtain an initial weight of each sub-risk index.
6. The system according to claim 5, wherein: The algorithm layer calculates the second risk index of the target process based on machine learning, and the execution method includes: inputting the first risk index into the trained LSTM model or XGBoost model to calculate the second risk index of the target process.
7. The system according to any one of claims 1 to 6, wherein: Also includes: The interaction layer is configured as follows: Performing visualization on the comprehensive risk index; These include: risk level display and highlighting of sub-risk indices exceeding thresholds; and Receive the approval command sent by the user and send it to the algorithm layer, so that the algorithm layer performs the approval operation according to the approval command; which includes: generating corresponding approval opinions based on the over-threshold sub-risk index when the approval is not passed.
8. A process approval method based on big data, characterized in that: include: Process the target process information according to the rule factor library to generate a structured summary; The rule factor library is a mapping relationship between approval nodes, core rule factors, rule triggering conditions and approval node data sources; According to the target process information, a first risk index and a second risk index of the target process are calculated based on the rule factor library and machine learning respectively, and the two are weighted to calculate a comprehensive risk index of the target process; and Process approval is carried out based on the comprehensive risk index.
9. The process approval method based on big data according to claim 8, characterized in that: The calculation formula of the first risk index is: in, represents the first risk index; Indicates the total number of core parameters corresponding to the target process information; Indicates the The sub-risk index corresponding to each core parameter; express Dynamic weight of Indicates the The current value of the core parameters; Indicates the The historical mean of the core parameters; Indicates the The historical standard deviation of the current value of each core parameter; The dynamic weight formula of the sub-risk index is: in, express Sub-risk index weight at the moment; Indicates the The initial weight of the item risk index; represents the adjustment coefficient; Indicates the The average of the historical values of the item risk index; Indicates the The maximum value among the historical values of the item risk index; Indicates the The minimum historical value of the item risk index.
10. A computer program product, the computer program product being stored in a computer-readable storage medium, characterized in that: When the computer program product is executed by a processor, it is used to at least implement the method according to claim 8 or 9.