City operation management service platform case flow approval method and device

Abstract

The embodiment of the invention provides a city operation management service platform case flow approval method and device. The method comprises the following steps: when a city management case in a city operation management service platform is in an examination and approval stage, inputting current case information of the city management case and case flow node information into a pre-fine-tuned DeepSeek-R1 model to obtain a case examination and approval opinion of a current case flow node; if the approval mode of the current case flow node is an automatic approval mode, the case approval opinions are filled in a corresponding input field to automatically complete approval of the current case flow node, and if the approval mode of the current case flow node is a manual approval mode, the case approval opinions are filled in the corresponding input field to automatically complete approval of the current case flow node. Performing manual rechecking by a corresponding approver so as to complete approval of the current case flow node; and transferring the urban management case approved by the current case flow node to the next case flow node. In this way, the case process approval efficiency can be improved.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to a method and apparatus for case process approval on an urban operation management service platform. Background Technology

[0002] In the existing case approval process of the urban operation management service platform, due to the wide variety and large number of urban management cases, and the need for experienced personnel to provide approval opinions, there are generally shortcomings such as slow case flow, long processing cycle, and low efficiency of case approval process in the urban operation management service platform. Summary of the Invention

[0003] In a first aspect, embodiments of this disclosure provide a case process approval method for an urban operation management service platform, the method comprising:

[0004] When a case involving urban management is in the approval stage in the urban operation management service platform, the current case information and the case process node information are input into the pre-tuned DeepSeek-R1 model for the current case process node. The model then outputs the case approval opinion based on the current case information and the case process node information. The output case approval opinion is then used as the case approval opinion for the current case process node.

[0005] If the approval mode of the current case process node is automatic approval mode, the case approval opinion of the current case process node is filled in the corresponding input field, and the approval of the current case process node is completed automatically. If the approval mode of the current case process node is manual approval mode, the case approval opinion of the current case process node is filled in the corresponding input field, and the person in charge of the approval of the current case process node manually reviews the filled case approval opinion to complete the approval of the current case process node.

[0006] Urban management cases that have completed the approval process at the current stage will be transferred to the next stage.

[0007] In some possible implementations of the first aspect, the case process nodes include: verification feedback node, case filing node, case dispatch node, handling feedback node, dispatch verification node, verification feedback node, and case completion node.

[0008] In some possible implementations of the first aspect, the case approval opinion includes: opinions on the handling of urban management cases at the next case process node, the current handling status of the approval personnel, and an analysis of similar cases.

[0009] In some possible implementations of the first aspect, the fine-tuned DeepSeek-R1 model is obtained by fine-tuning it in the following way:

[0010] The system obtains case information, case process node information, and case approval opinions of urban management cases within a historical time period from the urban operation management service platform. Then, it constructs samples using the case information and case process node information, and uses the case approval opinions of the case process nodes as the labels corresponding to the samples, thereby constructing a dataset based on this.

[0011] The dataset is divided into a training set and a test set, and the case approval opinions in the test set are modified to the specified case approval opinion format.

[0012] Based on the training and test sets, the GRPO algorithm is used to fine-tune the DeepSeek-R1 model. In some implementations of the first aspect, the expected value of the GRPO algorithm during fine-tuning is expressed as:

[0013]

[0014] Where q represents the query, corresponding to case information and case process node information of urban management cases; O represents the observation sequence, representing the case approval opinions generated by the model; o i Let represent the i-th observation sequence, corresponding to the i-th case approval opinion; G represents the number of observation sequences, corresponding to the number of urban management cases; P(Q) represents the query distribution, representing the distribution of case information and case process node information, i.e. the probability of different case information and case process node information appearing; This represents the probability of the old strategy generating the observation sequence, and the probability of the old model generating case approval opinions based on case information and case process node information.

[0015] In some possible implementations of the first aspect, the main optimization terms of the GRPO algorithm during fine-tuning are represented as follows:

[0016]

[0017] Where G represents the number of observation sequences, corresponding to the number of urban management cases; |o i | represents the length of the i-th observation sequence, corresponding to the length of the i-th case approval opinion; t represents the time step, indicating the steps involved in generating the case approval opinion; π θ (o i,t |q,o i,<t ) represents the probability of the new strategy selecting the observation at time step t, which represents the probability of the new model predicting the next case approval opinion based on case information, case process node information, and previously generated case approval opinions; This represents the probability of the old strategy selecting the observation at time step t, and the probability of the old model predicting the next case approval opinion based on case information, case process node information, and previously generated case approval opinions. ε represents the advantage estimate, indicating the advantage of the case approval opinion generated by the new strategy compared to the case approval opinion generated by the old strategy; ε represents the trimming range, used to limit the extent of strategy updates.

[0018] In some possible implementations of the first aspect, during fine-tuning, the KL divergence penalty term of the GRPO algorithm is partially expressed as:

[0019] -βD KL [π θ ||π ref ]

[0020] Where β represents the weight of the KL divergence penalty term, used to balance the degree of policy update and deviation from the reference policy, D KL π represents the KL divergence, used to measure the difference between the old and new strategies. θ Indicates the new strategy, π ref Indicates a reference strategy.

[0021] Secondly, embodiments of this disclosure provide a case process approval device for an urban operation management service platform, the device comprising:

[0022] The processing module is used to input the current case information and case process node information of the urban management case into the pre-tuned DeepSeek-R1 model when the urban management case in the city operation management service platform is in the approval stage. The model then outputs the case approval opinion based on the current case information and case process node information, and then uses the output case approval opinion as the case approval opinion for the current case process node.

[0023] The approval module is used to automatically complete the approval of the current case process node by filling in the case approval opinion in the corresponding input field if the approval mode of the current case process node is automatic approval mode; if the approval mode of the current case process node is manual approval mode, the case approval opinion is filled in the corresponding input field and the person in charge of the approval of the current case process node manually reviews the filled case approval opinion to complete the approval of the current case process node.

[0024] The transfer module is used to transfer urban management cases that have completed the approval process at the current case process node to the next case process node.

[0025] Thirdly, embodiments of this disclosure provide an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; the memory storing instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method described above.

[0026] Fourthly, embodiments of this disclosure provide a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the methods described above.

[0027] Compared with the prior art, this disclosure has at least the following technical effects:

[0028] When urban management cases in the city operation management service platform are in the approval stage, the current case information and case process node information of the urban management cases can be processed based on the pre-tuned DeepSeek-R1 model, and the case approval opinion of the current case process node can be automatically generated, thereby speeding up the case flow, shortening the processing cycle, and improving the efficiency of case process approval.

[0029] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of this disclosure, nor is it intended to restrict the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0030] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the scope of this disclosure. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein:

[0031] Figure 1 A flowchart illustrating a case process approval method for an urban operation management service platform provided by an embodiment of this disclosure is shown.

[0032] Figure 2 A structural diagram of a case process approval device for an urban operation management service platform provided by an embodiment of this disclosure is shown;

[0033] Figure 3 A structural diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0035] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0036] To address the problems in the background art, embodiments of this disclosure provide a method, apparatus, device, and storage medium for case process approval in an urban operation management service platform. Specifically, when an urban management case in the urban operation management service platform is in the approval stage, the current case information and case process node information of the urban management case can be processed based on a pre-tuned DeepSeek-R1 model to automatically generate case approval opinions for the current case process node, thereby accelerating case flow, shortening the processing cycle, and improving the efficiency of case process approval.

[0037] The following detailed description, with reference to the accompanying drawings and specific embodiments, illustrates a case process approval method, apparatus, equipment, and storage medium for an urban operation management service platform provided by this disclosure.

[0038] Figure 1 A flowchart illustrating a case process approval method for an urban operation management service platform provided by an embodiment of this disclosure is shown, such as... Figure 1 As shown, method 100 may include the following steps:

[0039] S110, when a city management case in the city operation management service platform is in the approval stage, for the current case process node, the current case information and case process node information of the city management case are input into the pre-fine-tuned DeepSeek-R1 model, which outputs the case approval opinion based on the current case information and case process node information, and then the output case approval opinion is used as the case approval opinion of the current case process node.

[0040] The case process nodes during the approval stage of urban management cases may include: verification feedback node, case filing node, case dispatch node, handling feedback node, dispatch verification node, verification feedback node, and case closure node. Case information can be obtained from the reporting of urban management cases by relevant personnel, including: case number, source of problem, case category, case subcategory, area code, responsibility grid number, area code, street code, and community code. Case process node information may include: attribute information of the case process node and related handling information. Case approval opinions may include: opinions on the handling of urban management cases at the next case process node, the current handling status of the approving personnel, and analysis of similar cases.

[0041] As an example, the DeepSeek-R1 model described above can be fine-tuned in the following way:

[0042] Obtain case information, case process node information, and case approval opinions for all urban management cases within a historical time period (e.g., within the last year) from the urban operation management service platform. Then, use the case information and case process node information to construct samples, and use the case approval opinions for the case process nodes as the corresponding labels for the samples. Based on this, construct a dataset and save it in JSON format.

[0043] The dataset is divided into a training set and a test set according to a preset ratio (e.g., 4:1), and the case approval opinions in the test set are modified to the specified case approval opinion format.

[0044] Based on the training and test sets, the GRPO algorithm is used to fine-tune the pre-trained DeepSeek-R1 model.

[0045] During fine-tuning, the expected value of the GRPO algorithm is expressed as:

[0046]

[0047] Where q represents the query, corresponding to case information and case process node information of urban management cases; O represents the observation sequence, representing the case approval opinions generated by the model; o i Let represent the i-th observation sequence, corresponding to the i-th case approval opinion; G represents the number of observation sequences, corresponding to the number of urban management cases; P(Q) represents the query distribution, representing the distribution of case information and case process node information, i.e. the probability of different case information and case process node information appearing; This represents the probability of the old strategy generating the observation sequence, and the probability of the old model generating case approval opinions based on case information and case process node information.

[0048] During fine-tuning, the main optimization terms of the GRPO algorithm are represented as follows:

[0049]

[0050] Where G represents the number of observation sequences, corresponding to the number of urban management cases; |o i | represents the length of the i-th observation sequence, corresponding to the length of the i-th case approval opinion; t represents the time step, indicating the steps involved in generating the case approval opinion; π θ (o i,t |q,o i,<t) represents the probability of the new strategy selecting the observation at time step t, which represents the probability of the new model predicting the next case approval opinion based on case information, case process node information, and previously generated case approval opinions; This represents the probability of the old strategy selecting the observation at time step t, and the probability of the old model predicting the next case approval opinion based on case information, case process node information, and previously generated case approval opinions. ε represents the advantage estimate, indicating the advantage of the case approval opinion generated by the new strategy compared to the case approval opinion generated by the old strategy; ε represents the trimming range, used to limit the scope of strategy updates and prevent excessive strategy updates.

[0051] During fine-tuning, the KL divergence penalty term of the GRPO algorithm is partially expressed as:

[0052] -βD KL [π θ ||π ref ]

[0053] Where β represents the weight of the KL divergence penalty term, used to balance the degree of policy update and deviation from the reference policy, D KL π represents the KL divergence, used to measure the difference between the old and new strategies. θ Indicates the new strategy, π ref This represents the reference strategy, which is the strategy of the initial model.

[0054] S120, if the approval mode of the current case process node is automatic approval mode, the case approval opinion of the current case process node is filled in the corresponding input field, and the approval of the current case process node is completed automatically. If the approval mode of the current case process node is manual approval mode, the case approval opinion of the current case process node is filled in the corresponding input field, and the person in charge of the approval of the current case process node manually reviews the filled case approval opinion to complete the approval of the current case process node.

[0055] It is worth noting that for a specific case process node, the approval mode can be pre-set to either automatic or manual approval mode. For example, the approval mode for the case filing node, case dispatch node, dispatch verification node, and case completion node can be set to automatic approval mode to facilitate seamless approval later; while the approval mode for the verification feedback node, disposal feedback node, and investigation feedback node can be set to manual approval mode to improve approval accuracy.

[0056] Optionally, during the approval process, some case process nodes (such as verification feedback nodes, handling feedback nodes, and investigation feedback nodes) also require the uploading of approval attachments.

[0057] S130 will transfer urban management cases that have completed the approval process at the current case process node to the next case process node.

[0058] In other words, the subsequent case process nodes will also adopt the processing procedure described above until the case process approval is completed.

[0059] In summary, this disclosure achieves at least the following technical effects:

[0060] When urban management cases in the city operation management service platform are in the approval stage, the current case information and case process node information of the urban management cases can be processed based on the pre-tuned DeepSeek-R1 model, and the case approval opinion of the current case process node can be automatically generated, thereby speeding up the case flow, shortening the processing cycle, and improving the efficiency of case process approval.

[0061] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this disclosure is not limited to the described order of actions, because according to this disclosure, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this disclosure.

[0062] The above is an introduction to the method embodiments. The following describes the solution described in this disclosure further through device embodiments.

[0063] Figure 2 The diagram illustrates a structural diagram of a case process approval device for an urban operation management service platform according to an embodiment of this disclosure, as shown below. Figure 2 As shown, the device 200 may include:

[0064] The processing module 210 is used to input the current case information and case process node information of the urban management case into the pre-fine-tuned DeepSeek-R1 model when the urban management case in the city operation management service platform is in the approval stage. The DeepSeek-R1 model then outputs the case approval opinion based on the current case information and case process node information, and then uses the output case approval opinion as the case approval opinion of the current case process node.

[0065] The approval module 220 is used to automatically complete the approval of the current case process node by filling in the case approval opinion in the corresponding input field if the approval mode of the current case process node is automatic approval mode, and if the approval mode of the current case process node is manual approval mode, the case approval opinion of the current case process node is filled in the corresponding input field, and the person in charge of the approval of the current case process node manually reviews the filled case approval opinion to complete the approval of the current case process node.

[0066] The transfer module 230 is used to transfer urban management cases that have completed the approval of the current case process node to the next case process node.

[0067] Understandable Figure 2 Each module / unit in the illustrated device 200 has the ability to implement Figure 1 The functions of each step in method 100 shown, and their corresponding technical effects, will not be elaborated here for the sake of brevity.

[0068] Figure 3 A structural diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. Electronic device 300 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device 300 may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0069] like Figure 3 As shown, the electronic device 300 may include a computing unit 301, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 302 or a computer program loaded from a storage unit 308 into a random access memory (RAM) 303. The RAM 303 may also store various programs and data required for the operation of the electronic device 300. The computing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0070] Multiple components in electronic device 300 are connected to I / O interface 305, including: input unit 306, such as keyboard, mouse, etc.; output unit 307, such as various types of displays, speakers, etc.; storage unit 308, such as disk, optical disk, etc.; and communication unit 309, such as network card, modem, wireless transceiver, etc. Communication unit 309 allows electronic device 300 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0071] The computing unit 301 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 301 performs the various methods and processes described above, such as method 100. For example, in some embodiments, method 100 may be implemented as a computer program product, including a computer program tangibly contained in a computer-readable medium, such as storage unit 308. In some embodiments, part or all of the computer program may be loaded and / or installed on device 300 via ROM 302 and / or communication unit 309. When the computer program is loaded into RAM 303 and executed by the computing unit 301, one or more steps of method 100 described above may be performed. Alternatively, in other embodiments, the computing unit 301 may be configured to perform method 100 by any other suitable means (e.g., by means of firmware).

[0072] The various embodiments described above can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), payload programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0073] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0074] In the context of this disclosure, a computer-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of computer-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0075] It should be noted that this disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute method 100 and achieve the corresponding technical effects achieved by the embodiments of this disclosure in executing the method. For the sake of brevity, these will not be elaborated here.

[0076] In addition, this disclosure also provides a computer program product including a computer program that implements method 100 when executed by a processor.

[0077] To provide interaction with a user, the embodiments described above can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0078] The embodiments described above can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with the implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication (e.g., a communication network) of any form or medium. Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0079] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0080] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0081] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A case process approval method of an urban operation management service platform, characterized in that, The method comprises: When a case of urban management in the urban management service platform is in an approval stage, case information and case process node information of the current case are input into a pre-tuned DeepSeek-R1 model for the current case process node, and the DeepSeek-R1 model outputs a case approval opinion according to the current case information and the case process node information, and then the output case approval opinion is taken as the case approval opinion of the current case process node; If the approval mode of the current case process node is an automatic approval mode, the case approval opinion of the current case process node is filled in the corresponding input box, and the approval of the current case process node is automatically completed, if the approval mode of the current case process node is a manual approval mode, the case approval opinion of the current case process node is filled in the corresponding input box, and the filled case approval opinion is manually reviewed by an approval personnel responsible for the current case process node to complete the approval of the current case process node; The urban management case that completes the approval of the current case process node is transferred to a next case process node.

2. The method of claim 1, wherein, The case process node comprises: a verification feedback node, a case filing node, a case dispatching node, a disposal feedback node, a dispatching verification node, a verification feedback node, and a case settlement node.

3. The method of claim 1, wherein, The case approval opinion comprises: an opinion on the processing of the urban management case in the next case process node, a processing situation of the current approval personnel, and an analysis of similar cases.

4. The method of claim 1, wherein, The tuned DeepSeek-R1 model is obtained by tuning in the following way: Case information, case process node information, and case approval opinions of case process nodes of urban management cases of the urban management service platform in a historical time period are obtained, then samples are constructed using the case information and the case process node information, and the case approval opinions of the case process nodes are taken as labels corresponding to the samples, and then a data set is constructed based on this; The data set is divided into a training set and a test set, and the case approval opinions in the test set are modified into a specified case approval opinion format; Based on the training set and the test set, the DeepSeek-R1 model is tuned using a GRPO algorithm.

5. The method of claim 4, wherein, In the tuning process, the expectation part of the GRPO algorithm is represented as: wherein q represents a query, case information and case process node information corresponding to a case of city management; O represents an observation sequence, representing a case approval opinion generated by the model; o i represents the i-th observation sequence, corresponding to the i-th case approval opinion; G represents the number of observation sequences, corresponding to the number of cases of city management; P(Q) represents the distribution of the query, representing the distribution of case information and case process node information, i.e. the probability of occurrence of different case information and case process node information; represents the probability of generating an observation sequence by the old strategy, representing the probability of generating a case approval opinion by the old model according to case information and case process node information.

6. The method of claim 5, wherein, In the tuning process, the main optimization part of the GRPO algorithm is represented as: where G represents the number of observation sequences, corresponding to the number of urban management cases; |o i | represents the length of the i-th observation sequence, corresponding to the length of the i-th case review opinion; t represents the time step, representing the step of generating the case review opinion; π θ (o i,t |q,o i,<t ) represents the probability of the new strategy selecting the observation value at time step t, representing the probability of the new model predicting the next case review opinion according to the case information, case process node information, and previously generated case review opinions; represents the probability of the old strategy selecting the observation value at time step t, representing the probability of the old model predicting the next case review opinion according to the case information, case process node information, and previously generated case review opinions; represents the advantage estimate, representing the advantage of the case review opinion generated by the new strategy over the case review opinion generated by the old strategy; ε represents the clipping range, used to limit the amplitude of strategy update.

7. The method of claim 6, wherein, In the tuning process, the KL divergence penalty part of the GRPO algorithm is represented as: -βD KL [π θ ||π ref ] where β represents the weight of the KL divergence penalty term, used to balance the degree of policy update and deviation from the reference policy, D KL represents the KL divergence, used to measure the difference between the new and old policies, π θ represents the new policy, π ref represents the reference policy.

8. A case flow approval device of a city operation management service platform, characterized in that, The device comprises: A processing module configured to, when a case of urban management in the urban management service platform is in an approval stage, input case information and case process node information of the current case into a pre-tuned DeepSeek-R1 model for the current case process node, and output a case approval opinion according to the current case information and the case process node information, and then take the output case approval opinion as the case approval opinion of the current case process node; The approval module is configured to fill the case approval opinion of the current case process node into the corresponding input field if the approval mode of the current case process node is an automatic approval mode, and to automatically complete the approval of the current case process node; if the approval mode of the current case process node is a manual approval mode, the case approval opinion of the current case process node is filled into the corresponding input field, and the responsible personnel for the approval of the current case process node manually reviews the filled case approval opinion to complete the approval of the current case process node; The flow module is configured to flow the urban management case which has completed the approval of the current case process node to the next case process node.

9. An electronic device, comprising: The electronic device includes at least one processor; and a memory connected in communication with the at least one processor; The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.

10. A non-transitory computer-readable storage medium having stored thereon computer instructions, wherein, The computer instructions are used to enable the computer to perform the method of any one of claims 1-7.

Images