Process approval method and device, computer equipment, storage medium and program product
By combining an approval rule engine, a state machine engine, a risk prediction model, and a large language model, the process approval strategy is dynamically optimized, solving the problems of insufficient efficiency and accuracy in existing process approval technologies and achieving more efficient process management.
Patent Information
- Application Number
- CN202511769885.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-28
- Publication Date
- 2026-02-24
AI Technical Summary
Existing process approval solutions rely on fixed rules, which cannot adapt to the dynamic changes of complex processes, resulting in low efficiency and low accuracy.
By deeply integrating the approval rule engine, approval state machine engine, risk prediction model and large language model, it obtains information about the process to be approved, analyzes and processes the approval chain, monitors the status of nodes, predicts risks, and outputs optimized process approval strategies when necessary.
It improves the efficiency and accuracy of process approval, can dynamically respond to complex process changes, optimize approval strategies, and shorten approval time.
Smart Images
Figure CN121563431A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of automated workflow approval technology, and in particular to a workflow approval method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Technology
[0002] As enterprises deepen their digital transformation, business processes are becoming increasingly complex and diversified, and manual process handling methods are no longer sufficient to meet the process management needs of enterprises. In enterprise management and project development, the automation and intelligence of process management have become key to improving efficiency and reducing costs.
[0003] In the process approval schemes of related technologies, the reliance on fixed process approval rules results in poor adaptability and an inability to handle complex dynamic changes in processes, leading to low efficiency and low accuracy in process approval. Summary of the Invention
[0004] Therefore, it is necessary to provide a process approval method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve the efficiency and accuracy of process approval in response to the above-mentioned technical problems.
[0005] Firstly, this application provides a process approval method, which includes:
[0006] In response to the triggering operation of the process approval, obtain the information of the process to be approved;
[0007] The pre-defined approval rule engine parses and processes the information of the approval process to determine the approval process chain; the approval process chain includes multiple approval nodes and the flow relationship between each approval node.
[0008] In response to the approval operation of the information to be approved, the node status of each approval node is monitored through the preset approval state machine engine to obtain node status monitoring data; the node status monitoring data is used to represent the approval progress of the information to be approved.
[0009] Risk prediction results are obtained by using a pre-trained risk prediction model based on node status monitoring data and historical approval time data of each process approval node.
[0010] When the risk prediction results indicate that the approval time exceeds the preset time threshold, the information of the process to be approved is processed by a large language model to output the target process approval strategy. The target process approval strategy is used to optimize the approval strategy of the process approval node for the information of the process to be approved.
[0011] In one embodiment, a preset approval rule engine is used to parse and process the information of the approval process to determine the approval process chain, including:
[0012] The system uses a pre-defined approval rule engine to call a pre-defined approval rule library. It matches the information of the process to be approved with each approval rule in the pre-defined approval rule library to obtain the target approval rule. Each approval rule includes conditions and corresponding actions. Conditions are used to determine whether to trigger the approval rule, and actions are used to characterize the approval rule.
[0013] If the information of the pending approval process meets the target conditions of the target approval rules, the approval process chain is determined according to the target action corresponding to the target conditions.
[0014] In one embodiment, the node status monitoring data includes the current approval duration of each process approval node; based on the node status monitoring data and the historical approval duration data of each process approval node, a risk prediction result is obtained through a pre-trained risk prediction model, including:
[0015] By processing historical approval time data of each process approval node through a pre-trained risk prediction model, the predicted approval time of each process approval node is obtained.
[0016] The remaining approval time is determined based on the current approval time and the predicted approval time.
[0017] Risk prediction results are obtained based on the remaining approval time and the preset remaining approval time threshold.
[0018] In one embodiment, the pending approval process information is processed using a large language model to output a target process approval strategy, including:
[0019] The information about the approval process is vectorized to obtain the target vector;
[0020] Based on the target vector, target document data is retrieved from a pre-defined process approval vector database. The process approval vector database includes document data associated with process approval, and the target document data consists of a pre-defined number of documents that rank highly similar to the target vector.
[0021] Input the target document data into the large language model to obtain the target process approval strategy.
[0022] In one embodiment, prior to the step of the approval operation on the information to be approved in the approval process, the method further includes:
[0023] The pre-set approval weight calculation engine determines the approval resource allocation scheme based on the number of approval tasks of the responsible parties at each process approval node and the task priority corresponding to the pending approval process information.
[0024] According to the approval resource allocation plan, corresponding approval scheduling resources are allocated to each process approval node; the approval scheduling resources represent the resources required to complete the information of the pending approval process.
[0025] In one embodiment, an approval resource allocation scheme is determined using a pre-defined approval weighting engine, based on the number of approval tasks for the responsible parties at each process approval node and the task priority corresponding to the pending approval process information. This includes:
[0026] Based on the task completion rate, average response time, and approval resource utilization rate of each approval node, a reward function for the reinforcement learning agent is constructed.
[0027] Based on the number of approval tasks of the responsible parties at each approval node and the task priority corresponding to the pending approval process information, the state of the reinforcement learning agent is constructed, and actions are output according to the reward function and the state; the actions represent the approval resource allocation scheme.
[0028] Secondly, this application also provides a process approval device, the device comprising:
[0029] The information acquisition module is used to acquire information about the process to be approved in response to the triggering operation of the process approval.
[0030] The workflow approval link determination module is used to parse and process the information of the workflow to be approved through a preset approval rule engine to determine the workflow approval link; the workflow approval link includes multiple workflow approval nodes and the flow relationship between each workflow approval node.
[0031] The node status monitoring module is used to respond to the approval operation of the information to be approved in the process. Through the preset approval state machine engine, it monitors the node status of each process approval node and obtains node status monitoring data. The node status monitoring data is used to represent the approval progress of the information to be approved in the process.
[0032] The risk prediction result acquisition module is used to obtain risk prediction results based on node status monitoring data and historical approval time data of each process approval node through a pre-trained risk prediction model.
[0033] The strategy output module is used to process the information of the process to be approved through a large language model and output the target process approval strategy when the risk prediction result indicates that the approval time exceeds the preset time threshold. The target process approval strategy is used to optimize the approval strategy of the process approval node for the information of the process to be approved.
[0034] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method steps of the first aspect.
[0035] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method steps of the first aspect.
[0036] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the method steps of the first aspect.
[0037] The aforementioned process approval method, apparatus, computer equipment, computer-readable storage medium, and computer program product acquire information about the process to be approved in response to a trigger operation for process approval; they parse and process the information to be approved using a preset approval rule engine to determine the process approval chain; the process approval chain includes multiple process approval nodes and the flow relationship between each process approval node; in response to the approval operation of the information to be approved, they monitor the node status of each process approval node using a preset approval state machine engine to obtain node status monitoring data; the node status monitoring data is used to characterize the approval progress of the information to be approved; a pre-trained risk prediction model, based on the node status monitoring data and the historical approval duration data of each process approval node, obtains a risk prediction result; if the risk prediction result indicates that the approval duration exceeds a preset duration threshold, a large language model processes the information to be approved and outputs a target process approval strategy; the target process approval strategy is used to optimize the approval strategy of the process approval node for the information to be approved. As can be seen from the above, this application, by deeply integrating the approval rule engine, approval state machine engine, risk prediction model, and large language model, can predict the risks of process approval, optimize the process approval strategy, and thus improve the efficiency and accuracy of process approval. Attached Figure Description
[0038] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0039] Figure 1 This is a diagram illustrating the application environment of a workflow approval method in one embodiment.
[0040] Figure 2 This is a flowchart illustrating a process approval method in one embodiment;
[0041] Figure 3 This is a flowchart illustrating the process approval chain in one embodiment;
[0042] Figure 4 This is a flowchart illustrating the process of obtaining risk prediction results in one embodiment;
[0043] Figure 5 This is a flowchart illustrating the process of obtaining a target process approval strategy in one embodiment.
[0044] Figure 6 This is a structural block diagram of a process approval device in one embodiment;
[0045] Figure 7 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation
[0046] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0047] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.
[0048] The process approval method provided in this application embodiment can be applied to, for example, Figure 1In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on a cloud or other network server. Terminal 102, in response to a trigger operation for process approval, obtains the information of the process to be approved; it parses and processes the information of the process to be approved using a preset approval rule engine to determine the process approval link; the process approval link includes multiple process approval nodes and the flow relationship between each process approval node; in response to the approval operation of the information of the process to be approved, it monitors the node status of each process approval node using a preset approval state machine engine to obtain node status monitoring data; the node status monitoring data is used to characterize the approval progress of the information of the process to be approved; through a pre-trained risk prediction model, based on the node status monitoring data and the historical approval duration data of each process approval node, it obtains a risk prediction result; if the risk prediction result indicates that the approval duration exceeds a preset duration threshold, it processes the information of the process to be approved using a large language model and outputs a target process approval strategy scheme; the target process approval strategy scheme is used to optimize the approval strategy of the process approval node for the information of the process to be approved. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, drones, low-altitude aircraft, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, and projection equipment. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted displays. Head-mounted displays can be virtual reality (VR) devices, augmented reality (AR) devices, and smart glasses. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.
[0049] In one embodiment, such as Figure 2 As shown, a process approval method is provided. This embodiment applies this method to... Figure 1 Taking terminal 102 as an example, the method includes the following steps:
[0050] Step S210: In response to the triggering operation of the process approval, obtain the process information to be approved.
[0051] The triggering action is used to initiate the approval process. Specifically, the triggering action can be the user entering the information of the process to be approved on the terminal interface.
[0052] For example, in an employee onboarding process scenario, an HR specialist fills in new employee information (name, position, department, urgency level, etc.) in the recruitment system and uploads attachments such as resume and academic certificates, triggering the onboarding process in the recruitment system.
[0053] Step S220: The information to be approved is parsed and processed by the preset approval rule engine to determine the approval process link; the approval process link includes multiple approval process nodes and the flow relationship between each approval process node.
[0054] The preset approval rule engine uses the Rete algorithm to achieve efficient approval rule matching, supporting both forward and backward reasoning modes. Approval rules are represented in JSON format.
[0055] Among them, the process approval chain is used to indicate which process approval nodes the information to be approved needs to go through and the order in which the process approval nodes approve the information to be approved.
[0056] In this embodiment, a preset approval rule engine is used to match the approval rules corresponding to the information of the process to be approved. Based on the approval rules, the process approval chain is determined.
[0057] Step S230: In response to the approval operation of the process information to be approved, the node status of each process approval node is monitored through the preset approval state machine engine to obtain node status monitoring data; the node status monitoring data is used to characterize the approval progress of the process information to be approved.
[0058] In this context, the approval process involves the responsible party at each approval node reviewing the information to be approved. For example, a department manager or HR director might approve the onboarding information of a new employee.
[0059] In this embodiment, the pre-defined approval state machine engine is implemented based on the UML state diagram specification and supports advanced features such as composite states, historical states, and concurrent states. The pre-defined approval state machine engine manages the state transitions of each process approval node and monitors the approval progress of each node in real time. The approval progress includes, but is not limited to, approval duration, approval load, and approval status labels (not approved, in progress, approved).
[0060] Step S240: Using a pre-trained risk prediction model, risk prediction results are obtained based on node status monitoring data and historical approval duration data for each process approval node.
[0061] Among them, the pre-trained risk prediction model is used for process time sequence feature learning and risk prediction.
[0062] In this embodiment, a pre-trained risk prediction model is used to predict the approval time of each process approval node based on historical approval time data. The approval time of each process approval node is compared with the predicted approval time to determine the risk prediction result.
[0063] Step S250: If the risk prediction result indicates that the approval time is greater than the preset time threshold, the information of the process to be approved is processed by the large language model to output the target process approval strategy scheme; the target process approval strategy scheme is used to optimize the approval strategy of the process approval node for the information of the process to be approved.
[0064] The preset duration threshold can be set according to actual needs.
[0065] Among them, the large language model can be a customized model based on the GPT-4 architecture, fine-tuned for the process processing domain.
[0066] In this embodiment, when the approval time of one or more process approval nodes exceeds a preset approval time threshold, a target process approval strategy is output based on the information of the process to be approved using a large language model. The approval process is then optimized according to the target process approval strategy, shortening the approval time and thus improving process approval efficiency.
[0067] The aforementioned workflow approval method acquires pending workflow information in response to workflow approval trigger operations; it then parses and processes this information using a pre-set approval rule engine to determine the workflow approval chain; this chain includes multiple workflow approval nodes and the flow relationships between them; in response to approval operations on the pending workflow information, a pre-set approval state machine engine monitors the node status of each node to obtain node status monitoring data; this data represents the approval progress of the pending workflow information; a pre-trained risk prediction model, based on the node status monitoring data and historical approval duration data for each node, obtains risk prediction results; if the risk prediction results indicate that the approval duration exceeds a pre-set threshold, a large language model processes the pending workflow information to output a target workflow approval strategy; this strategy optimizes the approval strategies for the pending workflow information at each node. As can be seen from the above, this application, by deeply integrating the approval rule engine, approval state machine engine, risk prediction model, and large language model, can predict workflow approval risks and optimize workflow approval strategies, thereby improving workflow approval efficiency and accuracy.
[0068] In one embodiment, such as Figure 3 As shown, the pre-defined approval rule engine parses and processes the information of the approval process to determine the approval chain, including:
[0069] Step S310: Using a preset approval rule engine, a preset approval rule library is called to match the information to be approved with each approval rule in the preset approval rule library to obtain the target approval rule; each approval rule includes conditions and corresponding actions; conditions are used to determine whether an approval rule is triggered, and actions are used to characterize the approval rule.
[0070] The preset approval rule library includes several approval rules.
[0071] In this embodiment, keyword recognition can be performed on the information to be approved, and the identified keywords can be checked to see if they match the words in each approval rule. The approval rule with the matching keywords is then used as the target approval rule. Alternatively, semantic recognition can be performed on the information to be approved to obtain semantic recognition results. The semantic recognition results are then compared with the semantics of each approval rule, and the approval rule with the highest similarity is used as the target approval rule.
[0072] Step S320: If the information of the process to be approved meets the target conditions of the target approval rule, determine the process approval link according to the target action corresponding to the target conditions.
[0073] In this embodiment, taking the employee onboarding approval process as an example, the information to be approved includes the employee's basic information (name, position, department, salary, etc.). The preset approval rule library includes employee onboarding approval rules. Employee onboarding approval rules include an approval rule identifier, approval rule name, conditions, and actions. For example, the conditions are: [conditions: []
[0074] {"field": "employeeType", "operator": "equals", "value": "full-time"},
[0075] {"field": "department", "operator": "in", "value": ["Technical Department", "Product Department", "Marketing Department"; "Operator": "in", "value": ["Technical Department", "Product Department", "Marketing Department"; "Operator": "in"; ...
[0076] Field Department
[0077] {"field": "salary", "operator": "greaterThan", "value": 10000}
[0078] ].
[0079] Actions: [
[0080] {"name": "assignApprover", "params": {"role": "Department Manager"}},
[0081] {"name": "notifyHR", "params": {"template": "newEmployeeOnboarding"}}
[0082] ].
[0083] According to the above employee onboarding approval rules, it was detected that the employee is a full-time employee, belongs to the technical department, and has a salary of 12,000. The following actions were triggered: the department manager and HR director were automatically assigned as the responsible parties for the process approval node, and the department manager and HR director were notified to complete the employee onboarding process through a preset template.
[0084] In one embodiment, such as Figure 4 As shown, the node status monitoring data includes the current approval time of each process approval node; based on the node status monitoring data and the historical approval time data of each process approval node, a pre-trained risk prediction model is used to obtain risk prediction results, including:
[0085] Step S410: Process the historical approval time data of each process approval node through a pre-trained risk prediction model to obtain the predicted approval time of each process approval node.
[0086] The pre-trained risk prediction model is an LSTM model. The LSTM model includes an input layer, an embedding layer, LSTM layer 1, Dropout, LSTM layer 2, a batch normalization layer, LSTM layer 3, a fully connected layer, and an output layer (Sigmoid activation). Each LSTM layer contains 64 neurons.
[0087] During pre-training, the risk prediction model uses the Adam optimizer with a learning rate of 0.001, a batch size of 32, and 50 training epochs, employing an early stopping strategy to prevent overfitting.
[0088] Step S420: Determine the remaining approval time based on the current approval time and the predicted approval time.
[0089] In this embodiment of the application, the remaining approval time is obtained by subtracting the predicted approval time from the current approval time.
[0090] Step S430: Obtain risk prediction results based on the remaining approval time and the preset remaining approval time threshold.
[0091] The preset remaining approval time threshold can be set according to actual needs.
[0092] In this embodiment, a first risk prediction result is obtained when the remaining approval time is greater than or equal to a preset remaining approval time threshold. A second risk prediction result is obtained when the remaining approval time is less than the preset remaining approval time threshold. If a second risk prediction result is obtained, the process approval strategy needs to be optimized.
[0093] In one embodiment, such as Figure 5 As shown, the pending approval process information is processed through a large language model, and the target process approval strategy is output, including:
[0094] Step S510: Vectorize the information to be approved to obtain the target vector.
[0095] In this embodiment of the application, word embedding method can be used to convert the information of the process to be approved into a target vector.
[0096] Step S520: Based on the target vector, retrieve the target document data from the preset process approval vector database; the process approval vector database includes document data associated with process approval, and the target document data is a preset number of document data that rank highly similar to the target vector.
[0097] The preset process approval vector database can be a FAISS vector database, including but not limited to corporate rules and regulations, process approval manuals, and historical process approval cases.
[0098] In this embodiment of the application, target document data is retrieved from a preset process approval vector database by using Retrieval Enhanced Generation (RAG) technology based on the target vector.
[0099] Step S530: Input the target document data into the large language model to obtain the target process approval strategy scheme.
[0100] In this embodiment of the application, the target document data is processed by a large language model to obtain the target process approval strategy scheme.
[0101] In one embodiment, prior to the step of the approval operation on the information to be approved in the approval process, the method further includes:
[0102] Step S610: Using a preset approval weighting engine, an approval resource allocation scheme is determined based on the number of approval tasks of the responsible parties at each process approval node and the task priority corresponding to the pending approval process information.
[0103] The preset approval weighting engine is used to dynamically allocate approval scheduling resources for each process approval node, which can shorten the waiting time of high-priority tasks, avoid resource idleness or waste, and prevent low-priority tasks from accumulating for a long time.
[0104] In this embodiment, the workflow approval typically consists of multiple serial or parallel workflow approval nodes (such as preliminary review, secondary review, final review, and signing). Each workflow approval node requires computing, storage, and network resources to complete the task. A preset approval weighting engine adjusts the resource quotas of each workflow approval node in real time based on the system status.
[0105] For example, if the queue length of the initial review node surges (state: initial review queue = 15, CPU utilization = 90%), the approval weighting engine may allocate an additional 20% of CPU resources to it, while reclaiming idle resources from the final review node (state: final review queue = 0, CPU utilization = 30%).
[0106] Step S620: According to the approval resource allocation plan, allocate corresponding approval scheduling resources to each process approval node; the approval scheduling resources represent the resources required to complete the pending approval process information.
[0107] Among them, approval scheduling resources include, but are not limited to, human resources (the number of responsible parties involved in the process approval), computing and memory resources (IT system resources that support the operation of the approval process), time resources (the time window for approval tasks), and process dependency resources (preconditions or dependencies that need to be met in the approval process).
[0108] In this embodiment of the application, according to the approval resource allocation scheme, corresponding approval scheduling resources are allocated to each process approval node, and each process approval node completes the approval of the process information to be approved using the approval scheduling resources.
[0109] In one embodiment, an approval resource allocation scheme is determined using a pre-defined approval weighting engine, based on the number of approval tasks for the responsible parties at each process approval node and the task priority corresponding to the pending approval process information. This includes:
[0110] Step S612: Construct a reward function for the reinforcement learning agent based on the task completion rate, average response time, and approval resource utilization rate of each approval node.
[0111] In this embodiment of the application, the expression for the reward function is:
[0112]
[0113] in, , , These are the weighting coefficients, set to 0.5, 0.3, and 0.2 respectively.
[0114] Step S614: Based on the number of approval tasks of the responsible parties at each process approval node and the task priority corresponding to the pending approval process information, construct the state of the reinforcement learning agent, and output actions according to the reward function and the state; the actions represent the approval resource allocation scheme.
[0115] In this embodiment, the pre-defined approval weighting engine acts as a reinforcement learning agent. By observing the state, it outputs actions and reward values, and uses the reward values to optimize its own strategy, ultimately outputting an optimized approval resource allocation scheme. The pre-defined approval weighting engine can improve the configuration efficiency of approval scheduling resources and increase the response speed of the approval process.
[0116] To understand the above embodiments, the workflow approval system architecture of this application includes an infrastructure layer, a data layer, a dual-engine core layer, an AI intelligence layer, and an application interface layer. The infrastructure layer includes computing resources, a storage system, and a network environment, providing the hardware foundation for system operation. The data layer is responsible for data collection, storage, and preprocessing, including workflow data, rule data, and historical decision data. The dual-engine core layer includes an approval rule engine and an approval status machine alarm, enabling basic rule execution and workflow status management. The AI intelligence layer integrates LSTM neural networks, LLM, and RAG technologies to provide intelligent decision support. The application interface layer provides APIs and a user interface, supporting system integration with external applications and user interaction. Communication between layers is achieved through standardized interfaces, ensuring loose coupling and scalability of the system. This workflow approval system architecture is not limited to employee onboarding approval scenarios; it can also be applied to various workflow approval scenarios such as employee transfers, employee resignations, and employee travel application and expense reimbursement, to improve workflow approval efficiency and accuracy.
[0117] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.
[0118] In one exemplary embodiment, such as Figure 6 As shown, a process approval device is provided, the device comprising:
[0119] The information acquisition module 610 is used to acquire information about the process to be approved in response to the triggering operation of the process approval.
[0120] The process approval link determination module 620 is used to parse and process the information of the process to be approved through a preset approval rule engine to determine the process approval link; the process approval link includes multiple process approval nodes and the flow relationship between each process approval node.
[0121] The node status monitoring module 630 is used to respond to the approval operation of the information to be approved in the process. Through the preset approval state machine engine, it monitors the node status of each process approval node and obtains node status monitoring data. The node status monitoring data is used to represent the approval progress of the information to be approved in the process.
[0122] The risk prediction result acquisition module 640 is used to obtain risk prediction results through a pre-trained risk prediction model based on node status monitoring data and historical approval time data of each process approval node.
[0123] The strategy output module 650 is used to process the information of the process to be approved through a large language model and output the target process approval strategy when the risk prediction result indicates that the approval time exceeds the preset time threshold. The target process approval strategy is used to optimize the approval strategy of the process approval node for the information of the process to be approved.
[0124] In one embodiment, a preset approval rule engine is used to parse and process the information of the approval process to determine the approval process chain, including:
[0125] The system uses a pre-defined approval rule engine to call a pre-defined approval rule library. It matches the information of the process to be approved with each approval rule in the pre-defined approval rule library to obtain the target approval rule. Each approval rule includes conditions and corresponding actions. Conditions are used to determine whether to trigger the approval rule, and actions are used to characterize the approval rule.
[0126] If the information of the pending approval process meets the target conditions of the target approval rules, the approval process chain is determined according to the target action corresponding to the target conditions.
[0127] In one embodiment, node status monitoring data includes the current approval duration of each process approval node; based on the node status monitoring data and the historical approval duration data of each process approval node, a risk prediction result is obtained through a pre-trained risk prediction model, including:
[0128] By processing historical approval time data of each process approval node through a pre-trained risk prediction model, the predicted approval time of each process approval node is obtained.
[0129] The remaining approval time is determined based on the current approval time and the predicted approval time.
[0130] Risk prediction results are obtained based on the remaining approval time and the preset remaining approval time threshold.
[0131] In one embodiment, the pending approval process information is processed through a large language model to output a target process approval strategy, including:
[0132] The information about the approval process is vectorized to obtain the target vector;
[0133] Based on the target vector, target document data is retrieved from a pre-defined process approval vector database. The process approval vector database includes document data associated with process approval, and the target document data consists of a pre-defined number of documents that rank highly similar to the target vector.
[0134] Input the target document data into the large language model to obtain the target process approval strategy.
[0135] In one embodiment, prior to the step of the approval operation on the information to be approved in the approval process, the method further includes:
[0136] The pre-set approval weight calculation engine determines the approval resource allocation scheme based on the number of approval tasks of the responsible parties at each process approval node and the task priority corresponding to the pending approval process information.
[0137] According to the approval resource allocation plan, corresponding approval scheduling resources are allocated to each process approval node; the approval scheduling resources represent the resources required to complete the information of the pending approval process.
[0138] In one embodiment, an approval resource allocation scheme is determined using a pre-defined approval weighting engine, based on the number of approval tasks for the responsible parties at each process approval node and the task priority corresponding to the pending approval process information. This includes:
[0139] Based on the task completion rate, average response time, and approval resource utilization rate of each approval node, a reward function for the reinforcement learning agent is constructed.
[0140] Based on the number of approval tasks of the responsible parties at each approval node and the task priority corresponding to the pending approval process information, the state of the reinforcement learning agent is constructed, and actions are output according to the reward function and the state; the actions represent the approval resource allocation scheme.
[0141] Each module in the aforementioned process approval device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.
[0142] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows. Figure 7 As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores workflow approval data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When executed by the processor, the computer program implements a workflow approval method.
[0143] Those skilled in the art will understand that Figure 7 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the aforementioned workflow approval method. The steps of the workflow approval method described here may be steps from one of the workflow approval methods in the various embodiments described above.
[0144] In one embodiment, a computer-readable storage medium is provided, storing a computer program that, when executed by a processor, causes the processor to perform the steps of the aforementioned workflow approval method. The steps of this workflow approval method may be steps from one of the workflow approval methods described in the various embodiments above.
[0145] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, causes the processor to perform the steps of the aforementioned workflow approval method. The steps of this workflow approval method may be steps from one of the workflow approval methods described in the various embodiments above.
[0146] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.
[0147] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.
[0148] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.
[0149] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A process approval method, characterized in that, The method includes: In response to the triggering operation of the process approval, obtain the information of the process to be approved; The pre-set approval rule engine parses and processes the information of the process to be approved to determine the process approval link; the process approval link includes multiple process approval nodes and the flow relationship between each process approval node. In response to the approval operation of the pending approval process information, the node status of each approval node in the process is monitored through a preset approval state machine engine to obtain node status monitoring data; the node status monitoring data is used to characterize the approval progress of the pending approval process information. Risk prediction results are obtained by using a pre-trained risk prediction model based on the node status monitoring data and the historical approval time data of each process approval node. If the risk prediction result indicates that the approval time exceeds a preset time threshold, the pending approval process information is processed by a large language model to output a target process approval strategy scheme; the target process approval strategy scheme is used to optimize the approval strategy of the process approval node for the pending approval process information.
2. The method according to claim 1, characterized in that, The step of parsing and processing the pending approval process information through a preset approval rule engine to determine the process approval chain includes: The preset approval rule engine calls the preset approval rule library to match the pending approval process information with each approval rule in the preset approval rule library to obtain the target approval rule; each approval rule includes a condition and an action corresponding to the condition; the condition is used to determine whether the approval rule is triggered, and the action is used to characterize the approval rule; If the pending approval process information meets the target conditions of the target approval rule, the process approval link is determined according to the target action corresponding to the target conditions.
3. The method according to claim 1, characterized in that, The node status monitoring data includes the current approval time of each of the process approval nodes; the risk prediction result obtained by the pre-trained risk prediction model based on the node status monitoring data and the historical approval time data of each of the process approval nodes includes: The historical approval time data of each of the process approval nodes are processed by a pre-trained risk prediction model to obtain the predicted approval time of each of the process approval nodes. The remaining approval time is determined based on the current approval time and the predicted approval time. The risk prediction result is obtained based on the remaining approval time and the preset remaining approval time threshold.
4. The method according to claim 1, characterized in that, The process of processing the pending approval process information through a large language model and outputting a target process approval strategy includes: The pending approval process information is vectorized to obtain the target vector; Based on the target vector, target document data is retrieved from a preset process approval vector database; the process approval vector database includes document data associated with process approval, and the target document data consists of a preset number of document data that rank highly similar to the target vector. The target document data is input into the large language model to obtain the target process approval strategy scheme.
5. The method according to claim 1, characterized in that, Prior to the step of responding to the approval operation on the pending approval process information, the method further includes: The approval resource allocation scheme is determined by using a preset approval weighting engine based on the number of approval tasks of the responsible objects of each process approval node and the task priority corresponding to the process information to be approved. According to the approval resource allocation scheme, corresponding approval scheduling resources are allocated to each of the process approval nodes; the approval scheduling resources represent the resources required to complete the pending approval process information.
6. The method according to claim 5, characterized in that, The step of determining an approval resource allocation scheme through a preset approval weighting engine, based on the number of approval tasks of the responsible parties at each of the process approval nodes and the task priority corresponding to the pending approval process information, includes: Based on the task completion rate, average response time, and approval resource utilization rate of each process approval node, a reward function for the reinforcement learning agent is constructed. Based on the number of approval tasks of the responsible parties at each of the process approval nodes and the task priority corresponding to the pending approval process information, the state of the reinforcement learning agent is constructed, and an action is output based on the reward function and the state; the action represents the approval resource allocation scheme.
7. A process approval device, characterized in that, The device includes: The information acquisition module is used to acquire information about the process to be approved in response to the triggering operation of the process approval. The process approval link determination module is used to parse and process the information of the process to be approved through a preset approval rule engine to determine the process approval link; the process approval link includes multiple process approval nodes and the flow relationship between each process approval node. The node status monitoring module is used to respond to the approval operation of the pending approval process information. Through a preset approval state machine engine, it monitors the node status of each approval node of the process and obtains node status monitoring data. The node status monitoring data is used to represent the approval progress of the pending approval process information. The risk prediction result acquisition module is used to obtain risk prediction results based on the node status monitoring data and the historical approval duration data of each process approval node through a pre-trained risk prediction model. The strategy output module is used to process the pending approval process information through a large language model and output a target process approval strategy when the risk prediction result indicates that the approval time exceeds a preset time threshold. The target process approval strategy is used to optimize the approval strategy of the process approval node for the pending approval process information.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.