Method and device for determining budget splitting strategy, electronic equipment and program product

Abstract

The invention discloses a budget splitting strategy determination method and device, electronic equipment and a program product, and relates to the technical field of artificial intelligence, and the determination method comprises the steps: extracting a business data set from a database, converting business data into a budget limit value based on a preset rule, and adding the budget limit value into a corresponding budget pool, under the condition that it is detected that a user submits a budget splitting strategy on a front-end page, calling a target model based on a preset interface, analyzing the budget splitting strategy, outputting a classification result of the budget splitting strategy and budget allocation data, and under the condition that the category to which the budget splitting strategy belongs is an abnormal category, performing budget allocation on the budget splitting strategy. And judging whether the detection score of the abnormal category is greater than a preset threshold, and when the detection score of the abnormal category is greater than the preset threshold, adjusting the budget splitting strategy based on the budget allocation data to obtain a target budget splitting strategy. The technical problem that the budget splitting strategy cannot be accurately determined in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more specifically, to a method, apparatus, electronic device, and program product for determining a budget splitting strategy. Background Technology

[0002] Currently, financial institutions use two types of technical solutions when splitting budgets: (1) Periodic budget distribution based on manual batch processing, usually at the end of the month or quarter, the budget splitting table is generated manually based on the data of each business system, and then the budget amount is split to the lower-level institutions or individuals according to fixed rules through manual means such as table statistics; (2) Semi-automatic budget distribution based on data warehouse + rule engine, business data is synchronized to the database every day, and the balance of the splittable "budget pool" of each institution is calculated on T+N days (usually N≥3 days) according to preset rules. Managers check their respective budget pool amounts, and then split the budget to the lower-level units at once according to a fixed template, for example, splitting according to the institution level multiplied by a fixed coefficient, only performing simple verification (such as whether the split ratio is 100%), without in-depth checking of the rationality of the split.

[0003] In the above method, if the budget split needs to be adjusted, it needs to be undone and the split process needs to be recalculated and executed. This process is time-consuming and many steps rely on manual operation, which is prone to errors and results in low accuracy of the budget split plan.

[0004] There is currently no effective solution to the above problems. Summary of the Invention

[0005] This invention provides a method, apparatus, electronic device, and program product for determining a budget splitting strategy, to at least solve the technical problem in the related art of being unable to accurately determine a budget splitting strategy.

[0006] According to one aspect of the embodiments of this application, a method for determining a budget splitting strategy is provided, comprising: extracting a set of business data within a preset time period from a database, and converting the business data in the set of business data into budget amount values ​​based on preset rules, and adding the budget amount values ​​to the corresponding budget pool; when a user submits a budget splitting strategy on a front-end page, calling a target model based on a preset interface to analyze the budget splitting strategy, and outputting the classification result of the budget splitting strategy and budget allocation data, wherein the budget splitting strategy is constructed based on the budget amount values ​​in the target budget pool, the target budget pool refers to the budget pool that allocates budget amount values ​​in the current budget splitting operation, and the classification result includes at least: the category to which the budget splitting strategy belongs and the detection score corresponding to the category; if the category to which the budget splitting strategy belongs is an abnormal category, determining whether the detection score of the abnormal category is greater than a preset threshold; if the detection score of the abnormal category is greater than the preset threshold, adjusting the budget splitting strategy based on the budget allocation data to obtain the target budget splitting strategy.

[0007] Furthermore, before adding the budget amount to the corresponding budget pool, the process includes: constructing multiple budget pools, wherein there is a hierarchical relationship between the budget pools, each budget pool is used to record the budget amount and to count the budget amount allocated to all other budget pools, and the other budget pools are the next level of the budget pool.

[0008] Furthermore, before calling the target model to analyze the budget splitting strategy and output the classification results of the budget splitting strategy and budget allocation data, the process also includes: collecting historical budget splitting data and extracting features from the historical budget splitting data to obtain a feature set. The feature set includes at least: a budget splitting ratio feature set, a time series feature set, a rule mapping feature set, and a combined feature set. The budget splitting ratio feature set is used to quantify the allocation of funds in the budget splitting, the time series feature set is used to characterize the change of the budget splitting ratio over time, the rule mapping feature set is obtained based on business rule mapping, and the combined feature set is synthesized through a calculation strategy.

[0009] Furthermore, after extracting features from the historical budget breakdown data to obtain a feature set, the process also includes: constructing an initial model using a preset fusion strategy, wherein the initial model includes at least: multiple model network structures, which are stacked to construct the initial model; and training the initial model based on the feature set to obtain the target model.

[0010] Furthermore, the steps for adjusting the budget splitting strategy based on budget allocation data to obtain the target budget splitting strategy include: determining preset budget allocation data based on a preset strategy; adjusting the budget splitting strategy based on the preset budget allocation data and the budget allocation data to obtain the adjusted budget splitting strategy; calling the target model to analyze the adjusted budget splitting strategy, outputting the classification results of the adjusted budget splitting strategy and the adjusted budget allocation data, and determining whether the category to which the adjusted budget splitting strategy belongs is an anomalous category; if the category to which the adjusted budget splitting strategy belongs is not an anomalous category, the adjusted budget splitting strategy is determined as the target budget splitting strategy.

[0011] Furthermore, after adjusting the budget allocation strategy based on the budget allocation data to obtain the target budget allocation strategy, the process also includes: monitoring the budget allocation data of the target model and verifying the budget allocation data to obtain the verification results; recording the budget allocation data when the verification results indicate that there are errors in the budget allocation data; correcting the budget allocation data to obtain corrected data and labeling the corrected data to obtain sample data; collecting a set of historical budget allocation strategies based on a preset frequency, and retraining the initial model based on the set of historical budget allocation strategies and the sample data to obtain the updated target model.

[0012] Furthermore, after adjusting the budget splitting strategy based on budget allocation data to obtain the target budget splitting strategy, the process also includes: configuring a target rule set for the new business when a new business is detected, wherein the new business is a business that is not recorded in the database, and the target rule set includes at least: multiple sub-rules, which are used to handle the calculation steps of the budget amount of the new business; configuring an effective time for the target rule set, and calculating the budget amount value of the new business based on the target rule set and the effective time.

[0013] According to another aspect of the embodiments of this application, a budget splitting strategy determination apparatus is also provided, comprising: a conversion unit, configured to extract a set of business data within a preset time period from a database, and convert the business data in the set of business data into budget amount values ​​based on preset rules, and add the budget amount values ​​to the corresponding budget pool; a calling unit, configured to, upon detecting that a user submits a budget splitting strategy on a front-end page, call a target model based on a preset interface to analyze the budget splitting strategy, and output the classification result of the budget splitting strategy and budget allocation data, wherein the budget splitting strategy is constructed based on the budget amount values ​​in the target budget pool, the target budget pool refers to the budget pool that allocates budget amount values ​​in the current budget splitting operation, and the classification result includes at least: the category to which the budget splitting strategy belongs and the detection score corresponding to the category; a judgment unit, configured to, if the category to which the budget splitting strategy belongs is an abnormal category, determine whether the detection score of the abnormal category is greater than a preset threshold; and an adjustment unit, configured to, if the detection score of the abnormal category is greater than the preset threshold, adjust the budget splitting strategy based on the budget allocation data to obtain the target budget splitting strategy.

[0014] Furthermore, the budget splitting strategy determination device also includes: a first construction module, used to construct multiple budget pools before adding budget amount values ​​to the corresponding budget pools, wherein the budget pools have a hierarchical relationship, each budget pool is used to record budget amount values ​​and count the budget amount values ​​allocated to all other budget pools, and the other budget pools are the next level of the budget pools.

[0015] Furthermore, the device for determining the budget splitting strategy also includes: a first extraction module, used to collect historical budget splitting data and extract features from the historical budget splitting data before calling the target model, analyzing the budget splitting strategy, and outputting the classification results and budget allocation data of the budget splitting strategy. The feature set includes at least: a budget splitting ratio feature set, a time series feature set, a rule mapping feature set, and a combined feature set. The budget splitting ratio feature set is used to quantify the allocation of funds in the budget splitting, the time series feature set is used to characterize the change of the budget splitting ratio over time, the rule mapping feature set is obtained based on business rule mapping, and the combined feature set is synthesized through a calculation strategy.

[0016] Furthermore, the budget splitting strategy determination device also includes: a second construction module, used to extract features from historical budget splitting data to obtain a feature set, and then use a preset fusion strategy to construct an initial model, wherein the initial model includes at least: multiple model network structures, which are stacked to construct the initial model; and a first training module, used to train the initial model based on the feature set to obtain a target model.

[0017] Furthermore, the adjustment unit includes: a first determining module, used to determine preset budget allocation data based on a preset strategy; a first adjusting module, used to adjust the budget splitting strategy based on the preset budget allocation data and the budget allocation data to obtain the adjusted budget splitting strategy; a first analyzing module, used to call the target model, analyze the adjusted budget splitting strategy, output the classification result of the adjusted budget splitting strategy and the adjusted budget allocation data, and determine whether the category to which the adjusted budget splitting strategy belongs is an anomalous category; and a second determining module, used to determine the adjusted budget splitting strategy as the target budget splitting strategy if the category to which the adjusted budget splitting strategy belongs is not an anomalous category.

[0018] Furthermore, the budget allocation strategy determination device also includes: a first verification module, used to monitor the budget allocation data of the target model and verify the budget allocation data after adjusting the budget allocation strategy based on the budget allocation data to obtain the target budget allocation strategy, and obtain the verification result; a first recording module, used to record the budget allocation data when the verification result indicates that there is an error in the budget allocation data; a first correction module, used to correct the budget allocation data to obtain corrected data, and to label the corrected data to obtain sample data; and a second training module, used to collect a set of historical budget allocation strategies based on a preset frequency, and to retrain the initial model based on the set of historical budget allocation strategies and the sample data to obtain the updated target model.

[0019] Furthermore, the budget splitting strategy determination device also includes: a first configuration module, used to adjust the budget splitting strategy based on budget allocation data to obtain a target budget splitting strategy, and then, upon detecting a new business, configure a target rule set for the new business, wherein the new business is a business not recorded in the database, and the target rule set includes at least: multiple sub-rules, which are used to handle the calculation steps of the budget amount for the new business; and a first calculation module, used to configure an effective time for the target rule set, and calculate the budget amount value for the new business based on the target rule set and the effective time.

[0020] According to another aspect of the embodiments of this application, a computer program product is also provided, including a non-volatile computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the method for determining any of the above-described budget splitting strategies.

[0021] According to another aspect of the embodiments of this application, an electronic device is also provided, including one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method for determining any of the above-described budget splitting strategies.

[0022] In this invention, a set of business data within a preset time period is extracted from the database. Based on preset rules, the business data in the set is converted into budget amounts, which are then added to the corresponding budget pool. When a user submits a budget splitting strategy on the front-end page, a target model is called based on a preset interface to analyze the budget splitting strategy. The classification results of the budget splitting strategy and budget allocation data are output. If the category to which the budget splitting strategy belongs is an anomaly category, it is determined whether the detection score of the anomaly category is greater than a preset threshold. If the detection score of the anomaly category is greater than the preset threshold, the budget splitting strategy is adjusted based on the budget allocation data to obtain the target budget splitting strategy. This solves the technical problem in related technologies where the budget splitting strategy cannot be accurately determined.

[0023] In this invention, a set of business data within a preset time period is extracted from the database. Then, according to preset rules, this business data is converted into specific budget amounts and automatically added to the associated budget pool. The balance information of each level of budget pool is updated in real time, ensuring accurate collection and hierarchical management of budget amounts. When a user submits a budget splitting strategy on the front-end page, a pre-trained target model is called through a preset interface to analyze the budget splitting strategy. The classification results of the budget splitting strategy (including its category and the corresponding detection score) and suggested budget allocation data are output. If the category to which the budget splitting strategy belongs is an anomalous category, it is determined whether the detection score of the anomalous category is greater than a preset threshold. If the detection score of the anomalous category is greater than the preset threshold, the budget splitting strategy is adjusted based on the budget allocation data to obtain the target budget splitting strategy. This achieves real-time risk identification and intelligent suggestions for budget splitting strategies, improving the accuracy and efficiency of budget allocation. Attached Figure Description

[0024] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0025] Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing a method for determining a budget splitting strategy is shown.

[0026] Figure 2This is a flowchart of the method for determining the budget splitting strategy according to Embodiment 1 of this application;

[0027] Figure 3 This is a flowchart of an optional model from training to application according to an embodiment of this application;

[0028] Figure 4 This is a schematic diagram of an optional four-level budget pool structure according to an embodiment of this application;

[0029] Figure 5 This is a schematic diagram of an optional budget splitting scheme approval process according to an embodiment of this application;

[0030] Figure 6 This is an optional system architecture diagram based on multi-level budget allocation according to an embodiment of this application;

[0031] Figure 7 This is a schematic diagram of an apparatus for determining an optional budget splitting strategy according to an embodiment of this application;

[0032] Figure 8 This is a structural block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0033] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0034] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0035] It should be noted that all related information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, and displayed data) collected and involved in this invention are information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of this data comply with the relevant laws, regulations, and standards of the relevant regions, necessary confidentiality measures have been taken, and it does not violate public order and good morals. Corresponding operation entry points are provided for users to choose to authorize or refuse. For example, this system has an interface with relevant users or organizations. Before obtaining relevant information, a request to obtain the information needs to be sent to the aforementioned user or organization through the interface, and the relevant information is obtained only after receiving consent from the aforementioned user or organization.

[0036] In this invention, by constructing a multi-level budget pool structure, business data is automatically collected and updated on T+1 day (i.e., the working day following the current transaction day or processing day (denoted as T day)). This ensures that managers at all levels can promptly grasp the budget status and promote the efficient flow and transparent transfer of budget resources. Secondly, by using a machine learning model to analyze and detect potential risks in user-submitted budget splitting schemes, it quickly identifies and intercepts abnormal splitting ratios. It also intelligently proposes optimization suggestions to assist users in adjusting to a more reasonable and fair budget allocation strategy, ensuring the reasonable, safe, and efficient allocation of budget resources. Furthermore, it supports the combined configuration of budget calculation rules to adapt to diverse business scenarios. Through a rule versioning management mechanism, it ensures the orderliness and traceability of rule changes, improving flexibility and scalability.

[0037] The present invention will now be described in detail with reference to various embodiments.

[0038] Example 1

[0039] According to an embodiment of this application, an embodiment of a method for determining a budget splitting strategy is also provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0040] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 A hardware block diagram of a computer terminal (or mobile device) for implementing a method for determining a budget splitting strategy is shown. Figure 1 As shown, computer terminal 10 (or mobile device) may include one or more ( Figure 1The processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission device 106 for communication functions may also be included. In addition, it may include: a display, a keyboard, a cursor control device, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera, wherein the network interface can be connected to wired and / or wireless networks. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0041] It should be noted that the aforementioned one or more processors 102 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10 (or mobile device). As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0042] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the budget splitting strategy determination method in this embodiment. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the aforementioned budget splitting strategy determination method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0043] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0044] The display can be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10 (or mobile device).

[0045] Under the aforementioned operating environment, this application provides the following: Figure 2 The method for determining the budget splitting strategy is shown. Figure 2 This is a flowchart of the method for determining the budget splitting strategy according to Embodiment 1 of this application, as follows: Figure 2 As shown, the method includes the following steps:

[0046] Step S201: Extract the business data set within the preset time period from the database, and based on preset rules, convert the business data in the business data set into budget amount values, and add the budget amount values ​​to the corresponding budget pool.

[0047] Optionally, a data lake or database stores detailed data on various business lines of a financial institution, including but not limited to customer transaction records, performance indicators, and expense consumption.

[0048] In this embodiment of the invention, a set of business data within a preset time period (e.g., day T-1) is extracted from the database; this is the latest data of business activities. Based on preset rules (i.e., budget pricing rules, rules for converting business data into budget amounts, such as the conversion formula between performance indicators and budget amounts), the business data in the set is converted into budget amount values, and these budget amounts are added to the corresponding budget pool (used for collecting and managing budget amounts at the corresponding level). For example, if a financial institution branch completes more deposit business than expected on a certain day, the corresponding incentive budget amount is calculated according to preset rules and automatically added to the branch's budget pool on day T+1.

[0049] Step S202: Upon detecting that a user has submitted a budget splitting strategy on the front-end page, the target model is called based on a preset interface to analyze the budget splitting strategy and output the classification results of the budget splitting strategy and budget allocation data. The budget splitting strategy is constructed based on the budget amount value in the target budget pool. The target budget pool refers to the budget pool that allocates budget amount values ​​in the current budget splitting operation. The classification results include at least the category to which the budget splitting strategy belongs and the detection score corresponding to the category.

[0050] In this embodiment of the invention, when it is detected that a user (such as a person in charge of a financial institution) submits a plan on the front-end page to split the budget amount in their budget pool to lower-level units (such as branches) (i.e., a budget splitting strategy, which is constructed based on the budget amount value in the target budget pool), the target model (a pre-trained machine learning model) is called based on a preset interface (a pre-set interface) to analyze the budget splitting strategy, identify potential abnormal splitting patterns, and output the classification result of the budget splitting strategy (e.g., normal or abnormal) and budget allocation data (i.e., the specific allocation amount and proportion of each receiving unit in the budget splitting strategy, such as suggested budget splitting: Department A 40%, Department B 30%, Branch C 30%). The target budget pool refers to the budget pool that allocates budget amount values ​​in the current budget splitting operation. The classification result includes at least: the category to which the budget splitting strategy belongs and the corresponding detection score. For example, if a branch submits a budget splitting strategy that allocates most of the budget to underperforming branches, the target model will analyze and provide a classification result for the abnormal category and a high detection score.

[0051] Step S203: If the category to which the budget splitting strategy belongs is an anomaly category, determine whether the detection score of the anomaly category is greater than a preset threshold.

[0052] In this embodiment of the invention, when the category to which the budget splitting strategy belongs is an anomaly category, it is determined whether the detection score of the anomaly category is greater than a preset threshold (which can be set by the user and is not limited here). By comparing it with the preset threshold, it is determined whether the budget splitting strategy needs to be intervened or adjusted, so as to avoid judging minor anomalies as situations that need to be handled, and to ensure the prudence and fairness of the decision.

[0053] Optionally, if the budget splitting strategy belongs to the normal category, the budget splitting strategy is allowed to enter the subsequent approval process, and the model score is recorded in the background for auditing purposes.

[0054] Step S204: If the detection score of the anomaly category is greater than the preset threshold, the budget splitting strategy is adjusted based on the budget allocation data to obtain the target budget splitting strategy.

[0055] In this embodiment of the invention, when the detection score of an anomaly category exceeds a preset threshold, the budget splitting strategy is adjusted based on budget allocation data to obtain a target budget splitting strategy. For example, if the preset threshold is 0.7, and the target model analysis shows that a certain branch has received an excessively high budget proportion and a detection score of 0.95 (greater than the preset threshold), and provides suggested budget allocation data, the budget splitting strategy for that branch can be adjusted based on this data to ensure a more balanced budget allocation, improve the fairness and efficiency of budget allocation, and prevent unreasonable budget allocations from entering the approval process.

[0056] Figure 3 This is a flowchart of an optional model from training to application according to an embodiment of this application, such as... Figure 3 As shown, the process includes a training phase, model storage and deployment, online inference service, and result application. The training phase involves the following steps: Records in the historical budget splitting database are first processed by a feature extraction module to convert them into training feature vectors and labels, thus constructing training data. This training data is divided into training and validation sets and fed into the model training engine. After the training engine produces a model, its accuracy is tested using the validation set. After multiple adjustments and iterations, the optimal model is finalized and output (i.e., the target model). Model storage and deployment involve saving the trained model files (including .pkl files, etc.) to a model library. System administrators use model management tools to deploy new models to the server, replacing old models or providing them for business use. The online inference service involves extracting identical feature vectors from newly submitted budget splitting records in real time using the online feature processing component. The inference service then loads the deployed model for calculation and returns a judgment result (normal / abnormal) and possible suggested allocation outputs. Finally, the business logic executes alert and optimization processes based on the model results, and the final decision is fed back to the user's front end. Meanwhile, these budget breakdown records, along with model judgments and final manual processing results, are recorded in the feedback log. The feedback log is used to summarize and regularly update the training dataset to enable continuous learning and improvement of the model.

[0057] In summary, by utilizing a real-time database update mechanism, business data is quickly converted into budget amounts and automatically added to the budget pool, ensuring the timeliness and accuracy of the budget pool. When users submit budget splitting strategies, the built-in machine learning model analyzes the rationality of the strategy in real time, outputs classification results and anomaly detection scores, effectively intercepts and assists in adjusting abnormal budget splitting strategies, improves the compliance and fairness of budget allocation, and thus solves the technical problem of not being able to accurately determine budget splitting strategies in related technologies.

[0058] In order to accurately construct multiple budget pools, in the method for determining the budget splitting strategy provided in Embodiment 1 of this application, multiple budget pools are constructed. There is a hierarchical relationship between the budget pools. Each budget pool is used to record the budget amount value and to count the budget amount value allocated to all other budget pools. The other budget pools are the next level of the budget pool.

[0059] In this embodiment of the invention, multiple budget pools are constructed, such as... Figure 4 As shown, Figure 4 This is a schematic diagram of an optional four-level budget pool structure according to an embodiment of this application, including: a primary budget pool, a secondary budget pool, a tertiary budget pool, and a quaternary budget pool. Each budget pool has its unique identifier and stores the budget amount value and allocation record of the corresponding level. There is a hierarchical relationship between the budget pools. Each budget pool is used to record the budget amount value and to statistically analyze the budget amount value allocated to all other budget pools, which are the next level down from the primary budget pool. For example, the primary budget pool, as the top level of budget management, is used to collect the total available budget for the entire jurisdiction, ensuring the visibility and controllability of the total budget. The secondary budget pool inherits the budget amount from the primary budget pool, performs initial splitting based on business needs and performance, and allocates it to the tertiary budget pool, and so on, forming a complete process from allocation to use, monitoring, and adjustment.

[0060] To accurately obtain the feature set, in the method for determining the budget splitting strategy provided in Embodiment 1 of this application, historical budget splitting data is collected, and features are extracted from the historical budget splitting data to obtain the feature set. The feature set includes at least: a budget splitting ratio feature set, a time-series feature set, a rule mapping feature set, and a combined feature set. The budget splitting ratio feature set is used to quantify the allocation of funds in the budget splitting, the time-series feature set is used to characterize the change of the budget splitting ratio over time, the rule mapping feature set is obtained based on business rule mapping, and the combined feature set is synthesized through a calculation strategy.

[0061] In this embodiment of the invention, historical budget split records (i.e., historical budget split data) are extracted from the database. These records include details of all budget split transactions, such as split date, total split amount, receiving unit, and respective split ratio. Feature extraction is performed on the collected historical split data to generate a feature set, including a budget split ratio feature set, a time-series feature set, a rule mapping feature set, and a combined feature set. The budget split ratio feature set (including but not limited to: the highest unit ratio (i.e., the maximum ratio obtained by a single receiving entity in each budget split), the lowest unit ratio (i.e., the minimum ratio obtained by a single receiving entity in each budget split), and the ratio of the top three units, etc.) is used to quantify the allocation of funds in the budget split. The time-series feature set (including but not limited to: the difference between the current ratio and the historical average, etc.) is used to characterize the change of the budget split ratio over time (this can be achieved by analyzing continuous data). The budget split ratio for each period is fitted with linear regression or difference calculation to extract the trend of the ratio changing over time. This can capture subtle skews that accumulate over time (such as the trend of a certain institution's share increasing over time) as well as sudden jumps or drops in the ratio. The rule mapping feature set includes, but is not limited to, the number of branches across which the budget split target is implemented, whether the budget split is skipped, and whether the upper limit of the budget split ratio is reached. These features are used to integrate business rule constraints into the model. The combined feature set (new features generated by combining two or more basic features) is synthesized through computational strategies (such as arithmetic operations (product, ratio, etc.)). By constructing rich nonlinear relationships, the model's discriminative ability is improved.

[0062] In order to accurately obtain the target model, the method for determining the budget splitting strategy provided in Embodiment 1 of this application adopts a preset fusion strategy to construct an initial model, wherein the initial model includes at least: multiple model network structures, which are stacked to construct the initial model; and the initial model is trained based on the feature set to obtain the target model.

[0063] Optionally, select and define the network structure of multiple machine learning models, including but not limited to XGBoost (gradient boosting decision tree), logistic regression, random forest, etc.

[0064] In this embodiment of the invention, a preset fusion strategy (i.e., stacked fusion) is adopted to construct an initial model, such as combining a gradient boosting decision tree model with a logistic regression model to form a two-layer model architecture. The gradient boosting decision tree model can first be used as a primary learner to train the aforementioned features (i.e., the feature set), outputting intermediate results (such as predicted scores or implicit representations generated by the tree model). Then, these intermediate results are used as new features, and the logistic regression model is used as a secondary learner to train, outputting the final verification judgment (i.e., obtaining the target model). Through stacked fusion, the powerful representational capabilities of complex models are utilized while avoiding overfitting, resulting in more accurate detection results.

[0065] It should be noted that the XGBoost model can be replaced by other algorithms that can produce equivalent outlier probability outputs, and no restrictions are imposed here.

[0066] After training is complete, the target model can be exported and stored in a model management repository (including model files, parameter weights, etc.). Then, a microservice architecture can be adopted to deploy the target model as an independent service, passing in features and outputting results through a preset interface call, which is convenient for subsequent replacement and upgrade.

[0067] Alternatively, inference time can be optimized using methods such as model distillation or offline pre-computation.

[0068] For situations where data distribution varies across regions, a multi-model heterogeneous hosting solution can be adopted. For example, different institutions can train their own optimal models, and during inference, the corresponding model can be dynamically selected based on the data source for prediction. Alternatively, the outputs of multiple models can be logically fused (e.g., voting mechanisms, weighted averaging, or more complex decision tree logic) to generate a final composite judgment result. These cross-domain joint training methods expand the effective sample size and the diversity of the feature space, improve the model's ability to identify rare and abnormal patterns and its adaptability to new environments, while simultaneously meeting the data segregation principles required by regulations, achieving a balance between technological innovation and compliance.

[0069] To accurately determine the target budget splitting strategy, the method for determining the budget splitting strategy provided in Embodiment 1 of this application involves: determining preset budget allocation data based on a preset strategy; adjusting the budget splitting strategy based on the preset budget allocation data and the budget allocation data to obtain the adjusted budget splitting strategy; calling the target model to analyze the adjusted budget splitting strategy, outputting the classification result of the adjusted budget splitting strategy and the adjusted budget allocation data, and determining whether the category to which the adjusted budget splitting strategy belongs is an abnormal category; if the category to which the adjusted budget splitting strategy belongs is not an abnormal category, the adjusted budget splitting strategy is determined as the target budget splitting strategy.

[0070] In this embodiment of the invention, based on a preset strategy, preset budget allocation data is determined (e.g., a candidate budget splitting scheme is calculated according to the historical average proportion of each recipient, or a budget splitting scheme is given according to a preset fairness principle (e.g., equal distribution among units of the same level)). The budget splitting strategy can be adjusted by combining the preset budget allocation data and the budget allocation data output by the target model to obtain the adjusted budget splitting strategy. The target model can be called again to analyze the adjusted budget splitting strategy, output the classification result of the adjusted budget splitting strategy and the adjusted budget allocation data, and determine whether the category to which the adjusted budget splitting strategy belongs is an abnormal category. If the category to which the adjusted budget splitting strategy belongs is not an abnormal category, the adjusted budget splitting strategy is determined as the target budget splitting strategy.

[0071] Optionally, it is also possible to retrieve budget splitting cases from similar historical business operations. For example, in the budget splitting of similar business operations, the average ratio is: 70% for Organization X and 30% for Organization Y. By comparing historical data horizontally, users can be helped to determine whether the current plan (i.e., the target budget splitting strategy) deviates too much from the normal range.

[0072] Optionally, if a user believes their allocation plan (i.e., budget splitting strategy) is reasonable, they can choose to force submission of the plan, but must enter a note explaining the reasons. In the case of forced submission, the plan can be marked as abnormal and submitted along with the model's judgment results and the user's explanation to the subsequent plan approver. To prevent abuse, forced submission permission can be granted only to management personnel, reducing the probability of unreasonable budget plans entering the approval process, achieving pre-emptive risk control, and reducing abnormal budget splitting plans.

[0073] To accurately obtain the updated target model, in the method for determining the budget allocation strategy provided in Embodiment 1 of this application, the budget allocation data of the target model is monitored and verified to obtain the verification result; if the verification result indicates that there is an error in the budget allocation data, the budget allocation data is recorded; the budget allocation data is corrected to obtain corrected data, and the corrected data is labeled to obtain sample data; based on a preset frequency, a set of historical budget allocation strategies is collected, and based on the set of historical budget allocation strategies and the sample data, the initial model is retrained to obtain the updated target model.

[0074] In this embodiment of the invention, the budget allocation data of the target model is monitored and verified to obtain verification results. If the verification results indicate that the budget allocation data contains errors (i.e., the budget allocation data output by the model is found to be incorrect after expert judgment), the budget allocation data is recorded and corrected to obtain corrected data. The corrected data is then labeled to obtain sample data. Furthermore, a historical budget splitting strategy set (i.e., labeled actual budget splitting cases) can be collected at a preset frequency (e.g., every quarter or every six months). Based on the historical budget splitting strategy set and the sample data, the initial model is retrained to obtain an updated target model. In terms of target model management, a smooth switch between old and new models can be achieved, ensuring model stability.

[0075] Once the budget splitting plan passes the target model validation (or is confirmed by the user after adjustments), it enters the formal approval process. Approver at all levels can receive the task online, and the approval interface clearly displays the details of the budget splitting plan, the model scoring results, and the user's notes. If a plan has been flagged as abnormal by the model and forced to resubmit, the higher-level approver will see a prominent red warning icon on the interface, along with the model-generated analysis report or risk warning, helping them to focus on this budget splitting plan and make a more cautious decision. Approver can choose to approve, reject, or request further explanation based on the actual situation. If rejected at any point in the approval chain, the plan will be returned to the applicant for modification and resubmitted. Finally, once approved at the highest level, the budget splitting plan officially takes effect, and the involved budget amount is updated in the database (i.e., deducted from the higher-level budget pool and added to the corresponding lower-level unit's budget pool balance).

[0076] Figure 5 This is a schematic diagram of an optional budget splitting scheme approval process according to an embodiment of this application, such as... Figure 5As shown, a budget splitting application is initiated by a manager at a certain level (such as a branch manager or sub-branch department). The manager fills out a plan on the front-end page to allocate a certain amount from their superior's pool to the subordinate level, and then submits it for approval. This application goes through the review and approval processes of the next higher level institution, and finally the final approval of the next higher level institution. For example, after the branch budget splitting plan is submitted, it is reviewed by the sub-branch. After the sub-branch approves it, it is submitted to the second-level branch for review. After the second-level branch approves it, it is submitted to the first-level branch for review. After the first-level branch approves it, it is submitted to the first-level branch for final approval, and so on, until the first-level branch approves it. For each level of review and approval, if it fails, it will be returned for modification of the budget breakdown plan. For example, if the branch fails to review or approve, it will be returned to the outlet for modification of the outlet's budget breakdown plan; if the second-level branch fails to review or approve, it will be returned to the branch for modification of the branch's budget breakdown plan; if the first-level branch fails to review or approve, it will be returned to the second-level branch for modification of the second-level branch's budget breakdown plan.

[0077] Optionally, approvers at all levels can view the application details and model validation results (with warnings displayed for any anomalies). Processing time limits can be configured for each review and approval stage. For example, there's a mechanism for automatic cancellation of budget splitting applications submitted by lower-level organizations if they fail to submit within the specified time. Each review and approval stage can also be configured to issue reminders or automatically reject applications if they are not processed within a certain number of days. Time limit control ensures the timely completion of the budget splitting process, preventing prolonged delays that could impact subsequent budget usage.

[0078] Optionally, the final approval result will be fed back to the machine learning module as a learning sample. If a solution is judged as anomalous by the model but is still approved by senior management, it is considered a case of model misjudgment; conversely, if the model judges it as anomalous and senior management also rejects or adjusts the solution, it indicates that the model's judgment is correct. By adding these real-world cases marked with the final result to the model training set, the model parameters are continuously iterated and optimized, reducing similar false alarms in subsequent judgments and improving the ability to identify real anomalies. This closed-loop feedback mechanism ensures that the model can continuously learn and improve over time.

[0079] In order to accurately calculate the budget amount of new services, in the method for determining the budget splitting strategy provided in Embodiment 1 of this application, when a new service is detected, a target rule set is configured for the new service, wherein the new service is a service that is not recorded in the database, and the target rule set includes at least: multiple sub-rules, which are used to handle the calculation steps of the budget amount of the new service; an effective time is configured for the target rule set, and the budget amount of the new service is calculated based on the target rule set and the effective time.

[0080] Optionally, the calculation of a business budget can be composed of multiple sub-rules. Each sub-rule can be independently version-managed and reused. Each rule configuration (such as a pricing formula for a product or an upper limit for a split ratio) has an effective date and version number.

[0081] In this embodiment of the invention, when a financial institution launches a new business or adjusts its incentive rate, no code modification is required. A corresponding set of rules (i.e., a target rule set) can be configured for the new business. For example, when a brand-new wealth management product first appears in the business record (i.e., a new business), various data related to this product are collected, including sales volume, customer feedback, and profit expectations. A sales budget sub-rule can be calculated based on the sales target for the first month, which could be 3% of the sales target amount. A reward budget sub-rule can also be calculated based on 10% of the product's expected profit. An effective time is set for the target rule set; that is, after the target rule set takes effect, the budget amount for the new business can be automatically calculated according to the new rules, while historical data retains the results of the old rules for traceability, ensuring data traceability.

[0082] Figure 6 This is an optional system architecture diagram based on multi-level budget allocation according to an embodiment of this application, such as... Figure 6 As shown, the system architecture may include: a data source and acquisition layer, a business logic layer, and an application presentation layer. The data source and collection layer connects to the data lake / data warehouse, collecting and aggregating data from various institutions and business lines daily. After cleaning and transformation, the data is written into the data tables of the primary, secondary, tertiary, and quaternary budget pools according to rules. This process ensures real-time updates of data at each level of the budget pool and performs data classification, grading, and security anonymization. The business logic layer includes core components such as a budget splitting engine, a rule engine, and a machine learning ratio verification module, realizing budget pool management, allocation process control, and anomaly verification. The machine learning ratio verification module is inserted as an intelligent verification link in the budget splitting scheme approval process. The business logic layer provides functions externally through service interfaces, such as submitting budget splitting and approval interfaces, and internally calls services to complete the corresponding functions. The application presentation layer provides management portals and approval workflow interfaces, providing users at all levels with exclusive budget pool views, including interfaces for branch and sub-branch managers, as well as secondary and primary branch management interfaces. Users can submit budget splitting applications and perform approval operations through the front end, and can also view various statistical reports and visualization dashboards. The application layer interacts with the business layer through interfaces to achieve real-time data display and operation submission.

[0083] The method for determining budget splitting strategies provided in this application embodiment can collect and analyze historical budget splitting data through a multi-model network structure constructed by a preset fusion strategy. It constructs a multi-level feature set including proportions, time series, rule mappings, and combined features, and trains an anomaly detection model based on this set. The model can evaluate the compliance of each budget splitting scheme in real time. For suspected anomaly-related budget splitting schemes, it can provide intelligent suggestions to help adjust them. Through highlighting warnings and closed-loop feedback in the approval process, it ensures the transparency and fairness of all budget splitting activities. Simultaneously, continuous model training and updates ensure the continuous optimization of the budget splitting strategy. Furthermore, in the face of new business introductions, it supports flexible configuration of budget rules and real-time budget amount calculation, ensuring the agility and adaptability of budget management, realizing the automation and intelligence of the budget management process, and improving the efficiency of budget allocation for financial institutions.

[0084] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0085] Example 2

[0086] This application also provides a budget splitting strategy determination apparatus. It should be noted that the budget splitting strategy determination apparatus of this application can be used to execute the budget splitting strategy determination method provided in this application. The budget splitting strategy determination apparatus provided in this application will be described below.

[0087] According to an embodiment of this application, an apparatus for implementing the above-described method for determining budget splitting strategies is also provided. Figure 7 This is a schematic diagram of an optional budget splitting strategy determination device according to an embodiment of this application, such as... Figure 7 As shown, the device for determining the budget splitting strategy may include: a conversion unit 70, a calling unit 71, a judgment unit 72, and an adjustment unit 73.

[0088] The conversion unit 70 is used to extract a set of business data within a preset time period from the database, and based on preset rules, convert the business data in the set of business data into budget amount values, and add the budget amount values ​​to the corresponding budget pool.

[0089] Calling unit 71 is used to call the target model based on a preset interface when it is detected that the user has submitted a budget splitting strategy on the front-end page, to analyze the budget splitting strategy, and output the classification result of the budget splitting strategy and the budget allocation data. The budget splitting strategy is constructed based on the budget amount value in the target budget pool. The target budget pool refers to the budget pool that allocates budget amount values ​​in the current budget splitting operation. The classification result includes at least: the category to which the budget splitting strategy belongs and the detection score corresponding to the category.

[0090] Judgment unit 72 is used to determine whether the detection score of the abnormal category is greater than a preset threshold when the category to which the budget splitting strategy belongs is an abnormal category.

[0091] The adjustment unit 73 is used to adjust the budget splitting strategy based on the budget allocation data when the detection score of the anomaly category is greater than a preset threshold, so as to obtain the target budget splitting strategy.

[0092] The budget splitting strategy determination device provided in this application embodiment can extract a set of business data within a preset time period from the database through the conversion unit 70, and convert the business data in the set of business data into budget amount values ​​based on preset rules, and add the budget amount values ​​to the corresponding budget pool. When the user submits a budget splitting strategy on the front-end page, the calling unit 71 can call the target model based on a preset interface to analyze the budget splitting strategy, output the classification result of the budget splitting strategy and the budget allocation data. When the category to which the budget splitting strategy belongs is an abnormal category, the judgment unit 72 can determine whether the detection score of the abnormal category is greater than a preset threshold. When the detection score of the abnormal category is greater than the preset threshold, the adjustment unit 73 can adjust the budget splitting strategy based on the budget allocation data to obtain the target budget splitting strategy.

[0093] Optionally, the budget splitting strategy determination device further includes: a first construction module, used to construct multiple budget pools before adding budget amount values ​​to the corresponding budget pools, wherein the budget pools have a hierarchical relationship, each budget pool is used to record budget amount values ​​and count the budget amount values ​​allocated to all other budget pools, and the other budget pools are the next level of the budget pools.

[0094] Optionally, the budget splitting strategy determination device further includes: a first extraction module, used to collect historical budget splitting data and extract features from the historical budget splitting data before calling the target model, analyzing the budget splitting strategy, and outputting the classification results of the budget splitting strategy and budget allocation data, to obtain a feature set. The feature set includes at least: a budget splitting ratio feature set, a time series feature set, a rule mapping feature set, and a combined feature set. The budget splitting ratio feature set is used to quantify the allocation of funds in the budget splitting, the time series feature set is used to characterize the change of the budget splitting ratio over time, the rule mapping feature set is obtained based on business rule mapping, and the combined feature set is synthesized through a calculation strategy.

[0095] Optionally, the budget splitting strategy determination device further includes: a second construction module, used to extract features from historical budget splitting data to obtain a feature set, and then use a preset fusion strategy to construct an initial model, wherein the initial model includes at least: multiple model network structures, which are stacked to construct the initial model; and a first training module, used to train the initial model based on the feature set to obtain a target model.

[0096] Optionally, the adjustment unit includes: a first determining module, used to determine preset budget allocation data based on a preset strategy; a first adjusting module, used to adjust the budget splitting strategy based on the preset budget allocation data and the budget allocation data to obtain the adjusted budget splitting strategy; a first analyzing module, used to call the target model, analyze the adjusted budget splitting strategy, output the classification result of the adjusted budget splitting strategy and the adjusted budget allocation data, and determine whether the category to which the adjusted budget splitting strategy belongs is an abnormal category; and a second determining module, used to determine the adjusted budget splitting strategy as the target budget splitting strategy if the category to which the adjusted budget splitting strategy belongs is not an abnormal category.

[0097] Optionally, the budget splitting strategy determination device further includes: a first verification module, used to monitor the budget allocation data of the target model and verify the budget allocation data after adjusting the budget splitting strategy based on the budget allocation data to obtain the target budget splitting strategy, and obtain the verification result; a first recording module, used to record the budget allocation data when the verification result indicates that there is an error in the budget allocation data; a first correction module, used to correct the budget allocation data to obtain corrected data, and to label the corrected data to obtain sample data; and a second training module, used to collect a set of historical budget splitting strategies based on a preset frequency, and to retrain the initial model based on the set of historical budget splitting strategies and the sample data to obtain the updated target model.

[0098] Optionally, the budget splitting strategy determination device further includes: a first configuration module, used to adjust the budget splitting strategy based on budget allocation data to obtain a target budget splitting strategy, and then, upon detecting a new business, configure a target rule set for the new business, wherein the new business is a business not recorded in the database, and the target rule set includes at least: multiple sub-rules, the sub-rules being used to handle the calculation steps of the budget amount for the new business; and a first calculation module, used to configure an effective time for the target rule set, and calculate the budget amount value for the new business based on the target rule set and the effective time.

[0099] The aforementioned budget splitting strategy determination device may also include a processor and a memory. The aforementioned conversion unit 70, calling unit 71, judgment unit 72, adjustment unit 73, etc., are all stored in the memory as program units, and the processor executes the aforementioned program units stored in the memory to realize the corresponding functions.

[0100] The aforementioned processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured. By adjusting kernel parameters, the budget allocation strategy is adjusted based on budget allocation data when the detection score of an anomaly category exceeds a preset threshold, thus obtaining the target budget allocation strategy.

[0101] The aforementioned memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0102] It should be noted that the conversion unit 70, calling unit 71, judging unit 72, and adjusting unit 73 mentioned above correspond to steps S201 to S204 in Embodiment 1. The instances and application scenarios implemented by the above units and corresponding steps are the same, but are not limited to the content disclosed in Embodiment 1. It should be noted that the above units can be hardware or software components stored in memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n). The above units can also be part of a device and run in the computer terminal 10 provided in Embodiment 1.

[0103] Example 3

[0104] Embodiments of this application may provide a computer terminal, which may be any computer terminal device in a group of computer terminals. Optionally, in this embodiment, the aforementioned computer terminal may also be replaced with a mobile terminal or an electronic device, etc.

[0105] Optionally, in this embodiment, the computer terminal may be located in at least one of a plurality of network devices in a computer network.

[0106] In this embodiment, the computer terminal can execute the following steps in the method for determining the budget splitting strategy: extracting a set of business data within a preset time period from the database, converting the business data in the set into budget amount values ​​based on preset rules, and adding the budget amount values ​​to the corresponding budget pool; when a user submits a budget splitting strategy on the front-end page, calling the target model based on a preset interface to analyze the budget splitting strategy, and outputting the classification result of the budget splitting strategy and budget allocation data, wherein the budget splitting strategy is constructed based on the budget amount values ​​in the target budget pool, the target budget pool refers to the budget pool that allocates budget amount values ​​in the current budget splitting operation, and the classification result includes at least: the category to which the budget splitting strategy belongs and the detection score corresponding to the category; when the category to which the budget splitting strategy belongs is an abnormal category, determining whether the detection score of the abnormal category is greater than a preset threshold; when the detection score of the abnormal category is greater than the preset threshold, adjusting the budget splitting strategy based on the budget allocation data to obtain the target budget splitting strategy.

[0107] Optionally, the computer terminal described above can execute the program code for the following steps in the method for determining the budget splitting strategy: constructing multiple budget pools, wherein there is a hierarchical relationship between the budget pools, each budget pool is used to record the budget amount value and to count the budget amount value allocated to all other budget pools, and the other budget pools are the next level of the budget pools.

[0108] Optionally, the aforementioned computer terminal can execute the program code for the following steps in the method for determining the budget splitting strategy: collecting historical budget splitting data and extracting features from the historical budget splitting data to obtain a feature set, wherein the feature set includes at least: a budget splitting ratio feature set, a time-series feature set, a rule mapping feature set, and a combined feature set. The budget splitting ratio feature set is used to quantify the allocation of funds in the budget splitting, the time-series feature set is used to characterize the change of the budget splitting ratio over time, the rule mapping feature set is obtained based on business rule mapping, and the combined feature set is synthesized through a calculation strategy.

[0109] Optionally, the aforementioned computer terminal can execute program code for the following steps in the method for determining the budget splitting strategy: using a preset fusion strategy to construct an initial model, wherein the initial model includes at least: multiple model network structures, which are stacked to construct the initial model; and training the initial model based on the feature set to obtain the target model.

[0110] Optionally, the aforementioned computer terminal may execute the program code for the following steps in the method for determining the budget splitting strategy: determining preset budget allocation data based on a preset strategy; adjusting the budget splitting strategy based on the preset budget allocation data and the budget allocation data to obtain the adjusted budget splitting strategy; calling the target model to analyze the adjusted budget splitting strategy, outputting the classification result of the adjusted budget splitting strategy and the adjusted budget allocation data, and determining whether the category to which the adjusted budget splitting strategy belongs is an abnormal category; if the category to which the adjusted budget splitting strategy belongs is not an abnormal category, determining the adjusted budget splitting strategy as the target budget splitting strategy.

[0111] Optionally, the aforementioned computer terminal can execute the program code for the following steps in the method for determining the budget allocation strategy: monitoring the budget allocation data of the target model and verifying the budget allocation data to obtain the verification result; recording the budget allocation data if the verification result indicates that there is an error in the budget allocation data; correcting the budget allocation data to obtain corrected data and labeling the corrected data to obtain sample data; collecting a set of historical budget allocation strategies based on a preset frequency, and retraining the initial model based on the set of historical budget allocation strategies and the sample data to obtain the updated target model.

[0112] Optionally, the aforementioned computer terminal may execute the program code for the following steps in the method for determining the budget splitting strategy: when a new business is detected, configure a target rule set for the new business, wherein the new business is a business that is not recorded in the database, and the target rule set includes at least: multiple sub-rules, which are used to handle the calculation steps of the budget amount of the new business; configure an effective time for the target rule set, and calculate the budget amount value of the new business based on the target rule set and the effective time.

[0113] Optionally, Figure 8 This is a structural block diagram of an electronic device according to an embodiment of this application. Figure 8 As shown, the electronic device may include: one or more ( Figure 8 Only one of the components is shown: processor 802, memory 804, memory controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module, and display.

[0114] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the budget splitting strategy determination method and apparatus in this application embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned budget splitting strategy determination method. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0115] The processor can access the information and application programs stored in the memory via the transmission device to execute the steps described above in the method for determining the budget splitting strategy.

[0116] This application provides a scheme for determining budget splitting strategies. By learning and analyzing historical budget splitting data through machine learning models, combined with real-time business data updates, the scheme can assess the rationality of budget splitting proposals and provide immediate feedback and suggestions based on multi-layered feature engineering and risk verification mechanisms when the proposals are submitted. Furthermore, flexible rule configuration and version management allow for adjustments to budget splitting rules according to different business scenarios, ensuring the timeliness and adaptability of the budget splitting strategy. This achieves comprehensive optimization of the budget allocation process, improving not only the accuracy of budget allocation but also enhancing the ability to identify and control risks, thereby solving the technical problem of accurately determining budget splitting strategies in related technologies.

[0117] Those skilled in the art will understand that Figure 8 The structure shown is for illustrative purposes only. Electronic devices can also be terminal devices such as smartphones, tablets, PDAs, and mobile internet devices (MIDs). Figure 8 This does not limit the structure of the aforementioned electronic device. For example, electronic devices may also include components that are more... Figure 8 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 8 The different configurations shown.

[0118] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0119] Example 4

[0120] Embodiments of this application also provide a storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the method for determining the budget splitting strategy provided in Embodiment 1.

[0121] Optionally, in this embodiment, the storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0122] This application also provides a computer program product, which, when executed on a data processing device, is suitable for performing the steps of a method for determining a budget splitting strategy.

[0123] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0124] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0125] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0126] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0127] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0128] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0129] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for determining a budget splitting strategy, characterized in that, include: Extract a set of business data within a preset time period from the database, and based on preset rules, convert the business data in the set of business data into budget amount values, and add the budget amount values ​​to the corresponding budget pool; Upon detecting that a user submits a budget splitting strategy on the front-end page, the target model is invoked based on a preset interface to analyze the budget splitting strategy, and the classification results and budget allocation data of the budget splitting strategy are output. The budget splitting strategy is constructed based on the budget amount value in the target budget pool, which refers to the budget pool that allocates the budget amount value in the current budget splitting operation. The classification results include at least: the category to which the budget splitting strategy belongs and the detection score corresponding to the category. If the category to which the budget splitting strategy belongs is an anomaly category, determine whether the detection score of the anomaly category is greater than a preset threshold; If the detection score of the anomaly category is greater than the preset threshold, the budget splitting strategy is adjusted based on the budget allocation data to obtain the target budget splitting strategy.

2. The method for determining the budget splitting strategy according to claim 1, characterized in that, Before adding the budget amount to the corresponding budget pool, the process also includes: Multiple budget pools are constructed, wherein there is a hierarchical relationship between the budget pools. Each budget pool is used to record the budget amount value and to count the budget amount value allocated to all other budget pools, wherein the other budget pools are the next level of the budget pool.

3. The method for determining the budget splitting strategy according to claim 1, characterized in that, Before calling the target model, analyzing the budget splitting strategy, and outputting the classification results and budget allocation data of the budget splitting strategy, the process also includes: Historical budget splitting data is collected, and features are extracted from the historical budget splitting data to obtain a feature set. The feature set includes at least: a budget splitting ratio feature set, a time series feature set, a rule mapping feature set, and a combined feature set. The budget splitting ratio feature set is used to quantify the allocation of funds in budget splitting. The time series feature set is used to characterize the change of the budget splitting ratio over time. The rule mapping feature set is obtained based on business rule mapping. The combined feature set is synthesized through a calculation strategy.

4. The method for determining the budget splitting strategy according to claim 3, characterized in that, After extracting features from the historical budget breakdown data to obtain a feature set, the process also includes: An initial model is constructed using a preset fusion strategy. The initial model includes at least multiple model network structures, which are stacked to construct the initial model. Based on the feature set, the initial model is trained to obtain the target model.

5. The method for determining the budget splitting strategy according to claim 1, characterized in that, The steps of adjusting the budget splitting strategy based on the budget allocation data to obtain the target budget splitting strategy include: Based on the preset strategy, the preset budget allocation data is determined; Based on the preset budget allocation data and the budget allocation data, the budget splitting strategy is adjusted to obtain the adjusted budget splitting strategy; The target model is invoked to analyze the adjusted budget splitting strategy, output the classification results of the adjusted budget splitting strategy and the adjusted budget allocation data, and determine whether the category to which the adjusted budget splitting strategy belongs is an abnormal category. If the adjusted budget splitting strategy does not belong to an abnormal category, the adjusted budget splitting strategy will be determined as the target budget splitting strategy.

6. The method for determining the budget splitting strategy according to claim 1, characterized in that, After adjusting the budget splitting strategy based on the budget allocation data to obtain the target budget splitting strategy, the method further includes: Monitor the budget allocation data of the target model, verify the budget allocation data, and obtain the verification results; If the verification result indicates that the budget allocation data contains an error, the budget allocation data shall be recorded. The budget allocation data is corrected to obtain corrected data, and the corrected data is labeled to obtain sample data; Based on a preset frequency, a set of historical budget splitting strategies is collected, and based on the set of historical budget splitting strategies and the sample data, the initial model is retrained to obtain an updated target model.

7. The method for determining the budget splitting strategy according to claim 1, characterized in that, After adjusting the budget splitting strategy based on the budget allocation data to obtain the target budget splitting strategy, the method further includes: Upon detecting a new service, a target rule set is configured for the new service, wherein the new service is a service not recorded in the database, and the target rule set includes at least: multiple sub-rules, which are used to handle the calculation steps of the budget amount for the new service; Configure an effective time for the target rule set, and calculate the budget amount for the new service based on the target rule set and the effective time.

8. A device for determining a budget splitting strategy, characterized in that, include: The conversion unit is used to extract a set of business data within a preset time period from the database, and based on preset rules, convert the business data in the set of business data into budget amount values, and add the budget amount values ​​to the corresponding budget pool. The calling unit is used to, upon detecting that a user has submitted a budget splitting strategy on the front-end page, call the target model based on a preset interface to analyze the budget splitting strategy, and output the classification result of the budget splitting strategy and budget allocation data. The budget splitting strategy is constructed based on the budget amount value in the target budget pool, which refers to the budget pool that allocates the budget amount value in the current budget splitting operation. The classification result includes at least: the category to which the budget splitting strategy belongs and the detection score corresponding to the category. The judgment unit is used to determine whether the detection score of the abnormal category is greater than a preset threshold when the category to which the budget splitting strategy belongs is an abnormal category; An adjustment unit is used to adjust the budget splitting strategy based on the budget allocation data when the detection score of the anomaly category is greater than the preset threshold, so as to obtain a target budget splitting strategy.

9. A computer program product, characterized in that, The method includes a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, implements the method for determining the budget splitting strategy as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, It includes one or more processors and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method for determining the budget splitting strategy as described in any one of claims 1 to 7.

Images