Model processing method and device for financial scene and computer program product

Abstract

The invention discloses a model processing method and device for a financial scene and a computer program product. Relates to the financial science and technology field or other related technology fields, and the method comprises the following steps: obtaining financial business demands, carrying out feature identification on the financial business demands, and obtaining N business demand features, the N business demand features comprising business types and N-1 business features, and N being a positive integer; m candidate models are determined according to the service type, a target model is screened out from the M candidate models according to the N-1 service features, and M is a positive integer; and predicting the service load of the target model by using the load prediction model to obtain service load prediction data, and sending the service load prediction data and the target model to the client. Through the method and the device, the problem of low model selection accuracy when a large model for processing financial business requirements is selected in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the field of financial technology or other related technical fields, and specifically, to a model processing method, device and computer program product for financial scenarios. Background Art

[0002] With the rapid development of artificial intelligence technology, big models, as a tool that can handle complex tasks and provide high-precision predictions, have shown great application potential in many fields such as finance and transportation. However, the widespread application and efficiency improvement of big models face a series of technical challenges and management difficulties, which limit their deployment and effectiveness in actual business scenarios.

[0003] First, the selection and adaptation of large models are highly dependent on the experience and judgment of human experts. Business personnel need to invest a lot of time and energy in model comparison and small-scale experiments to determine the model that best meets business needs. This process is cumbersome and time-consuming. There is a lack of systematic and automated tools to accurately match model characteristics with business needs, resulting in inefficient model selection and difficulty in achieving optimal performance matching. In addition, the data input into the large model needs to be preprocessed, labeled, updated, and other operations. These operations are mainly completed manually or semi-automatically, which is not only inefficient, but also easy to introduce human errors and affect data quality. In addition, large models have extremely high requirements for data real-time and accuracy. The existing data processing process is difficult to meet these requirements, which limits the model's real-time learning and rapid iteration capabilities. When processing sensitive data, it may touch multiple levels of risks, but these risks are mainly addressed through post-audits and remedial measures, thereby increasing the potential risks of model application.

[0004] Furthermore, the training and reasoning of large models consume a lot of computing resources, but the allocation and scheduling of resources currently mostly adopt static planning strategies, which are difficult to flexibly adjust according to business fluctuations, resulting in excess or insufficient resources, cost waste, and performance degradation. Cost management is usually a remedial measure taken in the later stage of model application. The lack of prior optimization planning makes it difficult to maximize resource efficiency and economic benefits. In addition, the performance of large models in actual business cannot be evaluated in a systematic and standardized manner, and it is difficult to quantify the specific contribution of the model to business value, which affects the decision-making process of model optimization, iteration, and return on investment analysis. At the same time, the method of using business indicators to directly quantitatively evaluate the role of the model is not yet mature, resulting in insufficient evaluation of the business effect of the model and unclear direction of model optimization.

[0005] When selecting large models to handle financial business needs in related technologies, there is a problem of low model selection accuracy, and no effective solution has been proposed yet. Summary of the invention

[0006] The main purpose of this application is to provide a model processing method, device and computer program product for financial scenarios to solve the problem of low model selection accuracy when selecting large models to process financial business needs in related technologies.

[0007] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a model processing method for financial scenarios is provided. The method includes: obtaining financial business needs, performing feature recognition on the financial business needs, and obtaining N business demand features, wherein the N business demand features include business types and N-1 business features, and N is a positive integer; determining M candidate models according to the business type, and selecting a target model from the M candidate models according to the N-1 business features, wherein M is a positive integer; using a load prediction model to predict the business load of the target model, obtaining business load prediction data, and sending the business load prediction data and the target model to the client.

[0008] Furthermore, the features of the financial business demand are identified to obtain N business demand features, including: using a language processing algorithm to parse the features of the financial business demand to obtain Y initial business demand features, where Y is greater than or equal to N and Y is a positive integer; using a machine learning algorithm to classify and label the Y initial business demand features to obtain Y processed business demand features; and adjusting the Y processed business demand features to obtain the N business demand features.

[0009] Furthermore, determining M candidate models according to the business type includes: obtaining K historical business demands in a historical time period, and obtaining a business model corresponding to each historical business demand, to obtain K business models, wherein K is greater than or equal to M and K is a positive integer; extracting the business type in each historical business demand to obtain K historical business types, and calculating the similarity between the business type and each historical business type to obtain K business similarity data; sorting the K business similarity data in descending order to obtain a similarity queue; obtaining M business similarity data in the similarity queue, and determining the business model associated with the M business similarity data as the M candidate models.

[0010] Furthermore, selecting a target model from the M candidate models according to the N-1 business features includes: performing feature conversion on the N-1 business features to obtain N-1 converted business features, and determining the feature weight of each converted business feature to obtain N-1 feature weights; acquiring model features of each candidate model to obtain M groups of model features; for each candidate model, using the N-1 feature weights to determine matching data between the N-1 converted business features and the model features associated with the candidate model to obtain model matching data; extracting the model matching data with the largest value from the M model matching data, and determining the candidate model associated with the model matching data with the largest value as the target model.

[0011] Furthermore, the business load of the target model is predicted using a load prediction model to obtain business load prediction data, including: obtaining P historical business demands, and obtaining business execution data corresponding to the P historical business demands, to obtain P groups of business execution data, wherein the business type of the P historical business demands is the same as the business type, and P is a positive integer; extracting historical business load data from each group of business execution data to obtain P historical business load data; determining the execution duration of the financial business corresponding to the financial business demand, inputting the execution duration and the P historical business load data into the load prediction model, and outputting the business load prediction data, wherein the business load prediction data includes multiple computing resources and a scheduling threshold for each computing resource.

[0012] Furthermore, after obtaining the business load forecast data, the method also includes: obtaining a resource scheduling list, wherein the resource scheduling list includes a plurality of business requirements and a relationship between the scheduling status of computing resources associated with each business requirement; using the business load forecast data and the execution duration to screen the resource scheduling list, determine the time window data of the financial business requirement, and send the time window data to the client, wherein the time window data refers to the time period during which the financial business requirement is allowed to be executed.

[0013] Furthermore, after obtaining N business demand features, the method also includes: performing differential privacy processing on the N business demand features to obtain N processed business demand features, and using the N processed business demand features to execute the step of screening out the target model from the M candidate models according to the N-1 business features.

[0014] In order to achieve the above-mentioned purpose, according to another aspect of the present application, a model processing device for financial scenarios is provided. The device includes: a first acquisition unit, used to acquire financial business needs, perform feature recognition on the financial business needs, and obtain N business demand features, wherein the N business demand features include business types and N-1 business features, and N is a positive integer; a determination unit, used to determine M candidate models according to the business type, and filter out a target model from the M candidate models according to the N-1 business features, wherein M is a positive integer; a prediction unit, used to predict the business load of the target model using a load prediction model, obtain business load prediction data, and send the business load prediction data and the target model to the client.

[0015] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided, wherein the computer-readable storage medium includes a stored executable program, wherein when the executable program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned model processing methods for financial scenarios.

[0016] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, wherein the memory stores an executable program, and the processor is used to run the program, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any one of the above-mentioned model processing methods for financial scenarios.

[0017] According to another aspect of an embodiment of the present invention, a computer program product is further provided, the computer program product comprising a computer program, wherein when the computer program is executed by a processor, any one of the above-mentioned model processing methods for financial scenarios is implemented.

[0018] In an embodiment of the present application, a model processing method for financial scenarios is adopted, by obtaining financial business needs, performing feature recognition on the financial business needs, and obtaining N business demand features, wherein the N business demand features include business types and N-1 business features, and N is a positive integer; M candidate models are determined according to the business type, and a target model is screened out from the M candidate models according to the N-1 business features, wherein M is a positive integer; a load prediction model is used to predict the business load of the target model to obtain business load prediction data, and the business load prediction data and the target model are sent to a client, and feature recognition is performed on the financial business needs, and a target model is determined according to the identified business demand features, and the load prediction model is used to predict the business load prediction data of the target model, thereby achieving the technical effect of accurately and efficiently selecting models and improving the efficiency of business processing, thereby solving the problem of low model selection accuracy when selecting a large model to process financial business needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0020] Figure 1 is a hardware structure block diagram of a computer terminal (or mobile device) for implementing a model processing method for a financial scenario;

[0021] Figure 2 is a flow chart of a model processing method for a financial scenario provided according to an embodiment of the present application;

[0022] Figure 3 is a schematic diagram of a model processing device for a financial scenario provided according to an embodiment of the present application;

[0023] Figure 4 It is a structural block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. For example, an interface is set up between this system and the relevant user or organization. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or organization through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or organization.

[0027] It should be noted that the collected information used in this application is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with the relevant laws, regulations and standards of the relevant regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse to use it.

[0028] Example 1

[0029] According to an embodiment of the present application, a method embodiment for model processing in financial scenarios is also provided. It should be noted that the steps shown in the flowchart of 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 can be executed in an order different from that shown here.

[0030] The method embodiment provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 It is a hardware structure diagram of a computer terminal (or mobile device) for implementing a model processing method for financial scenarios, such as Figure 1 As shown, the computer terminal 10 (or mobile device) may include one or more ( Figure 1102a, 102b, ..., 102n are used to illustrate) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU (Microcontroller Unit) or a programmable logic device FPGA (Field-Programmable Gate Array)), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a keyboard, a cursor control device, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0031] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10 (or mobile device). As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0032] The memory 104 can be used to store software programs and modules of application software, such as program instructions / data storage devices corresponding to the model processing method for financial scenarios in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, the above-mentioned model processing method for financial scenarios is realized. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0033] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC) and a network interface, which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0034] The display may be, for example, a touch screen liquid crystal display (LCD), which may enable a user to interact with a user interface of the computer terminal 10 (or mobile device).

[0035] Under the above operating environment, this application provides Figure 2 The model processing method for financial scenarios is shown. Figure 2 is a flow chart of a model processing method for a financial scenario provided in an embodiment of the present application, such as Figure 2 As shown, the method comprises the following steps:

[0036] Step S201, obtaining financial business needs, performing feature recognition on the financial business needs, and obtaining N business demand features, wherein the N business demand features include a business type and N-1 business features, and N is a positive integer.

[0037] Specifically, financial business needs may be business document processing needs, historical business data mining needs, etc. in financial institutions. In order to process the acquired financial business needs in a timely manner, first, after receiving the financial business needs, feature recognition may be performed on the financial business needs to identify business demand features including business types and business features, wherein business types may include document processing types, data mining types, etc., and business features may include data requirements, business rules, etc.

[0038] Step S202, determining M candidate models according to the business type, and screening out a target model from the M candidate models according to N-1 business characteristics, where M is a positive integer.

[0039] Specifically, after determining the business type according to the financial business needs, we can first extract multiple candidate models suitable for the business type from a model library including multiple models, and then analyze each model in the candidate models, including analyzing the model architecture, training data requirements, performance indicators (such as accuracy, response time), resource consumption, scalability, etc., and then determine the target model that is highly matched with the financial business needs based on the model information.

[0040] Step S203: Use the load prediction model to predict the service load of the target model, obtain service load prediction data, and send the service load prediction data and the target model to the client.

[0041] Specifically, after determining the target model, in order to ensure that the financial business needs can be stably met, it is necessary to determine the load capacity of the model. At this time, the historical business data of the model and related external environment data can be collected. The historical business data can include the number of model calls, processing time, response speed, resource consumption, etc., and the external environment data can include seasonal factors, holiday data, etc. Then, the above data is input into the load prediction model after data cleaning and other preprocessing. The load prediction model predicts the load capacity of the target model, that is, outputs the business load prediction data. Finally, the business pressure faced by the target model can be evaluated based on the data, providing a basis for subsequent resource scheduling and cost optimization.

[0042] The model processing method for financial scenarios provided in the embodiment of the present application obtains financial business needs, performs feature recognition on the financial business needs, and obtains N business demand features, wherein the N business demand features include business types and N-1 business features, where N is a positive integer; determines M candidate models according to the business type, and screens out a target model from the M candidate models according to the N-1 business features, wherein M is a positive integer; uses a load prediction model to predict the business load of the target model to obtain business load prediction data, and sends the business load prediction data and the target model to the client, thereby solving the problem of low model selection accuracy when selecting a large model for processing financial business needs in the related technology, and performs feature recognition on the financial business needs, determines the target model according to the identified business demand features, and uses the load prediction model to predict the business load prediction data of the target model, thereby achieving the technical effect of accurately and efficiently selecting the model and improving the efficiency of business processing.

[0043] Optionally, in the model processing method for financial scenarios provided in an embodiment of the present application, feature identification is performed on financial business needs to obtain N business demand features, including: using a language processing algorithm to parse the features of the financial business needs to obtain Y initial business demand features, where Y is greater than or equal to N and Y is a positive integer; using a machine learning algorithm to classify and label the Y initial business demand features to obtain Y processed business demand features; and adjusting the Y processed business demand features to obtain N business demand features.

[0044] Specifically, when identifying the business demand features corresponding to financial business demands, in order to reduce noise interference and improve the accuracy of subsequent analysis, the text description of the financial business demands can be preprocessed first, such as text cleaning, removal of stop words, etc. Then, language processing algorithms (such as bag-of-words model, word embedding model, etc.) are used to extract business-related features from the text description of the financial business demands to form multiple initial business demand features.

[0045] Further semantic understanding of the text is performed to identify the intent and core elements of the business requirements, ensuring that the initial business requirement features contain detailed information about the financial business. The above initial business requirement features are then classified and labeled using supervised learning machine learning algorithms (such as support vector machine models, random forest models, etc.), that is, the business requirement features are divided into different categories, making the features more structured and easier to handle. Finally, the importance of each feature for understanding business requirements and selecting models is evaluated through a feature selection algorithm to obtain the processed business requirement features. It should be noted that each feature can be associated with an importance score, and the feature selection algorithm can be feature importance calculation based on information entropy, feature weight analysis based on model coefficients, etc. Finally, based on the above importance scores, multiple features with importance scores greater than a preset score threshold are selected from the processed business requirement features to obtain the business requirement features. This embodiment determines business demand characteristics through text preprocessing and feature extraction of the language processing algorithm, which can improve the quality of business demand characteristics and the efficiency of the model selection and adaptation process, so that the model selection algorithm can more accurately understand the correlation between business needs and model capabilities, thereby providing the most matching model recommendation for financial business needs, reducing trial and error costs, and improving the efficiency and effectiveness of business processing.

[0046] Optionally, in the model processing method for financial scenarios provided in an embodiment of the present application, determining M candidate models based on business types includes: obtaining K historical business needs in a historical time period, and obtaining the business model corresponding to each historical business need, to obtain K business models, where K is greater than or equal to M and K is a positive integer; extracting the business type in each historical business need to obtain K historical business types, and calculating the similarity between the business type and each historical business type to obtain K business similarity data; sorting the K business similarity data in descending order to obtain a similarity queue; obtaining M business similarity data in the similarity queue, and determining the business model associated with the M business similarity data as M candidate models.

[0047] Specifically, when selecting candidate models based on business types, we can first obtain multiple historical business requirements within the historical time period, query the business models and related information used by each historical business requirement in the historical time period, such as querying historical data such as model usage records, model evaluation reports, and model performance logs for each historical business requirement, to ensure that the model information obtained is accurate and comprehensive. Then, we extract key information from historical business requirements through natural language processing technology or rule engines, identify the main business type of each requirement, and obtain the corresponding historical business type.

[0048] Furthermore, by using a similarity algorithm, the similarity between the current business type and the historical business type can be calculated, thereby obtaining multiple business similarity data, and these business similarity data are arranged in descending order to form a similarity queue, and the similarity data at the front of the queue are obtained, and the business model corresponding to these data is determined as a candidate model, wherein the similarity algorithm can be a text matching algorithm based on cosine similarity, a string comparison algorithm based on edit distance, a feature matching algorithm based on machine learning, etc. This embodiment recommends highly matching candidate models for current business needs by calculating the similarity between business types, which can more accurately match business needs, improve the pertinence and effect of model application, not only improve the accuracy and efficiency of model selection, but also promote the reuse of historical business knowledge and model application experience, and reduce trial and error costs.

[0049] Optionally, in the model processing method for financial scenarios provided in an embodiment of the present application, screening out a target model from M candidate models based on N-1 business features includes: performing feature conversion on the N-1 business features to obtain N-1 converted business features, and determining the feature weights of each converted business feature to obtain N-1 feature weights; obtaining model features of each candidate model to obtain M groups of model features; for each candidate model, using the N-1 feature weights to determine matching data between the N-1 converted business features and model features associated with the candidate model to obtain model matching data; extracting the model matching data with the largest value from the M model matching data, and determining the candidate model associated with the model matching data with the largest value as the target model.

[0050] Specifically, in order to select the target model based on business features, it is first necessary to standardize the unstandardized business features, unify the feature scale, and avoid the bias effect of features of different magnitudes on the model selection algorithm. Then, the standardized business features are converted to ensure that each feature can be converted into numerical data, which is convenient for subsequent weight calculation and matching analysis, so as to obtain the converted business features. Then, the machine learning model (such as random forest model, neural network weight analysis model, etc.) is used to determine the relative importance of the converted business features, that is, to determine the feature weight of each feature in the model selection process.

[0051] Further, model features are extracted from each candidate model, for example, performance indicators, resource requirements, data formats and type support of the model can be extracted to obtain multiple groups of model features. Then, the feature weight is used, combined with the model features of the model, and the matching degree between it and the business features is calculated through distance measurement (such as Euclidean distance, cosine similarity), correlation analysis or machine learning classification algorithm to obtain multiple model matching data. It should be noted that the numerical value of the data can reflect the matching degree between the model and the business requirements. The larger the numerical value, the higher the matching degree. Finally, the model matching data is sorted in descending order according to the numerical value, so as to quickly locate the model with the highest matching degree, extract the model matching data with the largest numerical value from the arrangement, and determine the candidate model associated with it as the target model. This embodiment screens the target model by using feature conversion, weight calculation, model feature matching and other methods to ensure that the final selected model can meet the business requirements to the greatest extent, avoid the subjectivity and trial and error cost of manually selecting the model, and improve the efficiency and effect of business processing.

[0052] Optionally, in the model processing method for financial scenarios provided in an embodiment of the present application, a load prediction model is used to predict the business load of a target model to obtain business load prediction data, including: obtaining P historical business demands, and obtaining business execution data corresponding to the P historical business demands, to obtain P groups of business execution data, wherein the business type of the P historical business demands is the same as the business type, and P is a positive integer; extracting historical business load data from each group of business execution data to obtain P historical business load data; determining the execution time of the financial business corresponding to the financial business demand, inputting the execution time and the P historical business load data into the load prediction model, and outputting business load prediction data, wherein the business load prediction data includes multiple computing resources and a scheduling threshold for each computing resource.

[0053] Specifically, in order to determine the load capacity of the target model, we can first extract historical business demands of the same type as the current financial business demands and their corresponding business execution data from the database. At this time, we can match historical business demands similar to the current financial business demands through information such as business codes, business description keywords or business type labels to improve prediction accuracy.

[0054] Furthermore, historical business load data can be extracted from each group of business execution data through log analysis, data mining and other technologies, and then statistical algorithms such as time series analysis and cluster analysis can be used to perform pattern recognition on the historical business load data to identify the periodicity, seasonality, trend and other characteristics of the business load, thereby providing data support for the training of the load prediction model. The historical business load data may include key indicators such as the number of historical business visits, data processing volume, computing resource usage (such as CPU (Central Processing Unit), GPU (Graphics Processing Unit), memory, etc.), and execution frequency.

[0055] After selecting the load prediction model for business load prediction, it is also necessary to determine the execution time in the current financial business demand that can reflect the complexity of the business and the amount of data processing. For example, when the financial business demand is an activity to issue points, the activity will last for one month, and the execution time is one month. The execution time and the characteristics corresponding to the historical business load data are then used as input parameters and input into the load prediction model, and the model outputs the business load prediction data, where the business load prediction data may include a variety of computing resources that may be required during the business execution process and the scheduling threshold of each computing resource. This embodiment predicts the resource demand of the current financial business by using historical business demand and execution time, and dynamically adjusts the allocation of computing resources based on business demand, thereby avoiding resource waste, improving resource utilization efficiency, reducing unnecessary costs, and providing solid data support for resource scheduling and cost optimization.

[0056] Optionally, in the model processing method for financial scenarios provided in an embodiment of the present application, after obtaining the business load forecast data, the method also includes: obtaining a resource scheduling list, wherein the resource scheduling list includes multiple business requirements and the relationship between the scheduling status of computing resources associated with each business requirement; using the business load forecast data and execution time to screen the resource scheduling list, determine the time window data of the financial business requirements, and send the time window data to the client, wherein the time window data refers to the time period allowed for the execution of the financial business requirements.

[0057] Specifically, in order to ensure efficient resource utilization and smooth business execution, resource scheduling optimization can be performed based on business load prediction and execution time. First, a resource scheduling list can be obtained, which records in detail the scheduling status of multiple business requirements and their associated computing resources (such as CPU (Central Processing Unit), GPU (Graphics Processing Unit), memory, etc.). The scheduling status can include information such as the current occupancy of resources, idle time, expected load, etc., providing basic data for intelligent resource scheduling.

[0058] Then, the resource scheduling list can be screened according to the business execution status and resource usage status corresponding to the business load prediction data, and the time period that allows the execution of financial business needs, that is, the time window data, can be identified. In this way, dynamic adjustment of resources can be achieved, wherein the time window data can be determined based on the idle period of resources, business peaks, and the matching degree of resource demand and supply, to ensure that the business is executed within the time period of the optimal resource configuration, and finally the data is sent to the client so that the business needs can be executed within the optimal time window. This embodiment can efficiently schedule computing resources and avoid resource waste by intelligently screening the resource scheduling list and determining the time window based on the predicted business load and execution duration. It can also realize the execution of business within the time window with the lowest cost. At the same time, by avoiding the peak of business load, the risk of system overload is reduced, the economic benefits and safety of resource utilization are improved, the overall cost of business operation is reduced, and the speed and stability of business processing are improved.

[0059] Optionally, in the model processing method for financial scenarios provided in an embodiment of the present application, after obtaining N business demand features, the method also includes: performing differential privacy processing on the N business demand features to obtain N processed business demand features, and using the N processed business demand features to execute the step of screening out the target model from M candidate models according to N-1 business features.

[0060] Specifically, in order to maintain data security, differential privacy processing can be performed on business demand features. To facilitate subsequent differential privacy processing, the business demand features are converted into numerical representations, and random noise is added to achieve differential privacy protection. In this way, the processed business demand features are obtained, thereby achieving the needs of protecting user privacy and data security. This embodiment performs differential privacy processing on business demand features. Even if there is a certain amount of data distortion in the processed business demand features, it can still accurately identify the model that best matches the business demand based on the machine learning algorithm, ensuring the accuracy and efficiency of model selection and adaptation, and also ensuring the efficiency and effect of business processing, achieving a balance between privacy protection and business performance.

[0061] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0062] Example 2

[0063] The embodiment of the present application also provides a model processing device for a financial scenario. It should be noted that the model processing device for a financial scenario in the embodiment of the present application can be used to execute the model processing method for a financial scenario provided in the embodiment of the present application. The model processing device for a financial scenario provided in the embodiment of the present application is introduced below.

[0064] According to an embodiment of the present application, a device for implementing the above-mentioned model processing method for financial scenarios is also provided. Figure 3 is a schematic diagram of a model processing device for a financial scenario provided according to an embodiment of the present application, such as Figure 3 As shown, the device includes: a first acquisition unit 30, a determination unit 31, and a prediction unit 32.

[0065] The first acquisition unit 30 is used to acquire financial service requirements, perform feature recognition on the financial service requirements, and obtain N service requirement features, wherein the N service requirement features include a service type and N-1 service features, where N is a positive integer;

[0066] A determination unit 31 is used to determine M candidate models according to the business type, and select a target model from the M candidate models according to N-1 business characteristics, where M is a positive integer;

[0067] The prediction unit 32 is used to predict the service load of the target model using the load prediction model, obtain service load prediction data, and send the service load prediction data and the target model to the client.

[0068] The model processing device for financial scenarios provided in the embodiment of the present application obtains financial business needs through a first acquisition unit 30, performs feature recognition on the financial business needs, and obtains N business demand features, wherein the N business demand features include business types and N-1 business features, and N is a positive integer; a determination unit 31 determines M candidate models according to the business type, and screens out a target model from the M candidate models according to the N-1 business features, wherein M is a positive integer; a prediction unit 32 uses a load prediction model to predict the business load of the target model to obtain business load prediction data, and sends the business load prediction data and the target model to the client, thereby solving the problem of low model selection accuracy when selecting a large model for processing financial business needs in the related art, by performing feature recognition on the financial business needs, determining the target model according to the identified business demand features, and using the load prediction model to predict the business load prediction data of the target model, thereby achieving the technical effect of accurately and efficiently selecting the model and improving the efficiency of business processing.

[0069] Optionally, in the model processing device for financial scenarios provided in an embodiment of the present application, the first acquisition unit 30 includes: a parsing module, used to use a language processing algorithm to parse the characteristics of financial business needs, and obtain Y initial business demand characteristics, where Y is greater than or equal to N and Y is a positive integer; a labeling module, used to use a machine learning algorithm to classify and label the Y initial business demand characteristics, and obtain Y processed business demand characteristics; an adjustment module, used to adjust the Y processed business demand characteristics to obtain N business demand characteristics.

[0070] Optionally, in the model processing device for financial scenarios provided in an embodiment of the present application, the determination unit 31 includes: a first acquisition module, used to obtain K historical business demands in a historical time period, and obtain the business model corresponding to each historical business demand, to obtain K business models, where K is greater than or equal to M and K is a positive integer; a first extraction module, used to extract the business type in each historical business demand, to obtain K historical business types, and calculate the similarity between the business type and each historical business type, to obtain K business similarity data; a sorting module, used to sort the K business similarity data in descending order to obtain a similarity queue; a second acquisition module, used to obtain M business similarity data in the similarity queue, and determine the business models associated with the M business similarity data as M candidate models.

[0071] Optionally, in the model processing device for financial scenarios provided in an embodiment of the present application, the determination unit 31 includes: a conversion module, used to perform feature conversion on N-1 business features to obtain N-1 converted business features, and determine the feature weight of each converted business feature to obtain N-1 feature weights; a third acquisition module, used to acquire the model features of each candidate model to obtain M groups of model features; a first determination module, used to determine, for each candidate model, the matching data of the N-1 converted business features and the model features associated with the candidate model using the N-1 feature weights to obtain model matching data; a second extraction module, used to extract the model matching data with the largest value from the M model matching data, and determine the candidate model associated with the model matching data with the largest value as the target model.

[0072] Optionally, in the model processing device for financial scenarios provided in an embodiment of the present application, the prediction unit 32 includes: a fourth acquisition module, used to obtain P historical business demands, and obtain business execution data corresponding to the P historical business demands, to obtain P groups of business execution data, wherein the business type of the P historical business demands is the same as the business type, and P is a positive integer; a third extraction module, used to extract historical business load data from each group of business execution data, to obtain P historical business load data; a second determination module, used to determine the execution time of the financial business corresponding to the financial business demand, input the execution time and the P historical business load data into the load prediction model, and output business load prediction data, wherein the business load prediction data includes multiple computing resources and a scheduling threshold for each computing resource.

[0073] Optionally, in the model processing device for financial scenarios provided in an embodiment of the present application, the device also includes: a second acquisition unit, used to obtain a resource scheduling list after obtaining the business load forecast data, wherein the resource scheduling list includes multiple business requirements and the relationship between the scheduling status of computing resources associated with each business requirement; a screening unit, used to use the business load forecast data and execution time to screen the resource scheduling list, determine the time window data of the financial business requirements, and send the time window data to the client, wherein the time window data refers to the time period allowed for the execution of the financial business requirements.

[0074] Optionally, in the model processing device for financial scenarios provided in an embodiment of the present application, the device also includes: a processing unit, which is used to perform differential privacy processing on the N business demand features after obtaining N business demand features, to obtain N processed business demand features, and use the N processed business demand features to execute the step of screening out the target model from M candidate models according to N-1 business features.

[0075] It should be noted that the above-mentioned first acquisition unit 30, determination unit 31, and prediction unit 32 correspond to steps S201 to S203 in Example 1, and the three units and the corresponding steps implement the same examples and application scenarios, but are not limited to the contents disclosed in the above-mentioned Example 1. It should be noted that the above-mentioned modules or units can be hardware components or software components stored in a memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n), and the above-mentioned units can also be part of the device and can be run in the computer terminal 10 provided in Example 1.

[0076] Example 3

[0077] The embodiment of the present application may provide a computer terminal, which may be any computer terminal device in a computer terminal group. Optionally, in this embodiment, the computer terminal may also be replaced by a terminal device such as a mobile terminal or an electronic device.

[0078] Optionally, in this embodiment, the computer terminal may be located in at least one network device among a plurality of network devices of the computer network.

[0079] In this embodiment, the above-mentioned computer terminal can execute the program code of the following steps in the model processing method for financial scenarios: obtaining financial business needs, performing feature recognition on the financial business needs, and obtaining N business demand features, wherein the N business demand features include business types and N-1 business features, and N is a positive integer; determining M candidate models according to the business type, and screening out a target model from the M candidate models according to the N-1 business features, wherein M is a positive integer; using the load prediction model to predict the business load of the target model, obtaining business load prediction data, and sending the business load prediction data and the target model to the client.

[0080] Optionally, the computer terminal may execute the program code of the following steps in the model processing method for financial scenarios: using a language processing algorithm to parse the characteristics of financial business needs to obtain Y initial business need characteristics, where Y is greater than or equal to N and Y is a positive integer; using a machine learning algorithm to classify and label the Y initial business need characteristics to obtain Y processed business need characteristics; and adjusting the Y processed business need characteristics to obtain N business need characteristics.

[0081] Optionally, the computer terminal may execute the program code of the following steps in the model processing method for financial scenarios: obtaining K historical business demands in a historical time period, and obtaining the business model corresponding to each historical business demand, to obtain K business models, where K is greater than or equal to M and K is a positive integer; extracting the business type in each historical business demand to obtain K historical business types, and calculating the similarity between the business type and each historical business type to obtain K business similarity data; sorting the K business similarity data in descending order to obtain a similarity queue; obtaining M business similarity data in the similarity queue, and determining the business models associated with the M business similarity data as M candidate models.

[0082] Optionally, the computer terminal may execute the program code of the following steps in the model processing method for financial scenarios: performing feature conversion on N-1 business features to obtain N-1 converted business features, and determining the feature weight of each converted business feature to obtain N-1 feature weights; obtaining the model features of each candidate model to obtain M groups of model features; for each candidate model, using the N-1 feature weights to determine the matching data of the N-1 converted business features and the model features associated with the candidate model to obtain model matching data; extracting the model matching data with the largest value from the M model matching data, and determining the candidate model associated with the model matching data with the largest value as the target model.

[0083] Optionally, the above-mentioned computer terminal can execute the program code of the following steps in the model processing method for financial scenarios: obtain P historical business demands, and obtain business execution data corresponding to the P historical business demands, to obtain P groups of business execution data, wherein the business type of the P historical business demands is the same as the business type, and P is a positive integer; extract historical business load data from each group of business execution data to obtain P historical business load data; determine the execution time of the financial business corresponding to the financial business demand, input the execution time and the P historical business load data into the load prediction model, and output business load prediction data, wherein the business load prediction data includes multiple computing resources and the scheduling threshold of each computing resource.

[0084] Optionally, the computer terminal may execute the program code of the following steps in the model processing method for financial scenarios: obtaining a resource scheduling list, wherein the resource scheduling list includes multiple business requirements and the relationship between the scheduling status of computing resources associated with each business requirement; screening the resource scheduling list using business load forecast data and execution duration, determining time window data for financial business requirements, and sending the time window data to the client, wherein the time window data refers to the time period during which the financial business requirements are allowed to be executed.

[0085] Optionally, the above-mentioned computer terminal can execute the program code of the following steps in the model processing method for financial scenarios: perform differential privacy processing on N business demand features to obtain N processed business demand features, and use the N processed business demand features to execute the step of screening out the target model from M candidate models according to N-1 business features.

[0086] Optionally, Figure 4 is a structural block diagram of an electronic device according to an embodiment of the present application. Figure 4 As shown, the electronic device may include: one or more ( Figure 4 (only one is shown) processor 402, memory 404, storage controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0087] Among them, the memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the model processing method and device for financial scenarios in the embodiments of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned model processing method for financial scenarios. The memory may include a high-speed random access memory, and may also include a 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 a memory remotely arranged relative to the processor, and these remote memories can be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0088] The processor can call the information and application programs stored in the memory through the transmission device to execute the above steps in the above model processing method for financial scenarios.

[0089] By adopting the embodiment of the present application, a solution for model processing in financial scenarios is provided. By acquiring financial business needs, the characteristics of financial business needs are identified to obtain N business demand characteristics, where the N business demand characteristics include business types and N-1 business characteristics, where N is a positive integer; M candidate models are determined according to the business type, and the target model is screened out from the M candidate models according to the N-1 business characteristics, where M is a positive integer; the business load of the target model is predicted using the load prediction model to obtain business load prediction data, and the business load prediction data and the target model are sent to the client, thereby achieving the technical effect of accurately and efficiently selecting models and improving the efficiency of business processing, thereby solving the problem of low model selection accuracy when selecting a large model to process financial business needs.

[0090] It can be understood by those skilled in the art that Figure 4The structure shown is for illustration only, and the electronic device may also be a smart phone, a tablet computer, a PDA, a mobile Internet device (MID), a PAD or other terminal device. Figure 4 The structure of the electronic device is not limited. Figure 4 More or fewer components (such as network interfaces, display devices, etc.) shown in, or having Figure 4 Different configurations are shown.

[0091] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0092] Example 4

[0093] The embodiment of the present application further provides a storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the model processing method for financial scenarios provided in the first embodiment.

[0094] Optionally, in this embodiment, the above storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0095] Optionally, in this embodiment, the storage medium is configured to store program code for executing the following steps: obtaining financial business needs, performing feature identification on the financial business needs, and obtaining N business demand features, wherein the N business demand features include a business type and N-1 business features, and N is a positive integer; determining M candidate models according to the business type, and screening out a target model from the M candidate models according to the N-1 business features, wherein M is a positive integer; predicting the business load of the target model using a load prediction model, obtaining business load prediction data, and sending the business load prediction data and the target model to the client.

[0096] The present application also provides a computer program product which, when executed on a data processing device, is suitable for executing the program of the model processing method steps for financial scenarios.

[0097] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0098] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0099] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0100] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0101] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0102] If the integrated unit is implemented in the form of 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 the present application, 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, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0103] The above are only preferred implementations of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A model processing method for financial scenarios, characterized in that: include: Acquire financial service requirements, perform feature recognition on the financial service requirements, and obtain N service requirement features, wherein the N service requirement features include a service type and N-1 service features, where N is a positive integer; Determine M candidate models according to the business type, and select a target model from the M candidate models according to the N-1 business characteristics, where M is a positive integer; The load prediction model is used to predict the business load of the target model to obtain business load prediction data, and the business load prediction data and the target model are sent to the client.

2. The method according to claim 1, characterized in that: The financial business demand is characterized and N business demand characteristics are obtained, including: Analyze the characteristics of the financial business demand by using a language processing algorithm to obtain Y initial business demand characteristics, where Y is greater than or equal to N and Y is a positive integer; Using a machine learning algorithm to classify and label the Y initial business demand features to obtain Y processed business demand features; The Y processed business demand characteristics are adjusted to obtain the N business demand characteristics.

3. The method according to claim 1, characterized in that Determining M candidate models according to the business type includes: Obtain K historical business requirements in a historical time period, and obtain a business model corresponding to each historical business requirement to obtain K business models, where K is greater than or equal to M and is a positive integer; Extracting the business type in each historical business demand to obtain K historical business types, and calculating the similarity between the business type and each historical business type to obtain K business similarity data; Sorting the K business similarity data in descending order to obtain a similarity queue; M business similarity data in the similarity queue are obtained, and business models associated with the M business similarity data are determined as the M candidate models.

4. The method according to claim 1, characterized in that: Screening out a target model from the M candidate models according to the N-1 business features includes: Performing feature conversion on the N-1 service features to obtain N-1 converted service features, and determining a feature weight of each converted service feature to obtain N-1 feature weights; Obtain the model features of each candidate model to obtain M groups of model features; For each candidate model, determining matching data between the N-1 converted business features and model features associated with the candidate model using the N-1 feature weights to obtain model matching data; The model matching data with the largest value is extracted from the M model matching data, and the candidate model associated with the model matching data with the largest value is determined as the target model.

5. The method according to claim 1, characterized in that The load prediction model is used to predict the service load of the target model, and the service load prediction data obtained includes: Acquire P historical business demands, and acquire business execution data corresponding to the P historical business demands, to obtain P groups of business execution data, wherein the business types of the P historical business demands are the same as the business type, and P is a positive integer; Extract historical service load data from each group of service execution data to obtain P historical service load data; Determine the execution time of the financial service corresponding to the financial service demand, input the execution time and the P historical service load data into the load prediction model, and output the service load prediction data, wherein the service load prediction data includes multiple computing resources and a scheduling threshold of each computing resource.

6. The method according to claim 5, characterized in that After obtaining the service load prediction data, the method further includes: Obtaining a resource scheduling list, wherein the resource scheduling list includes a relationship between multiple business requirements and a scheduling state of a computing resource associated with each business requirement; The resource scheduling list is screened using the business load forecast data and the execution duration to determine the time window data of the financial business demand, and the time window data is sent to the client, wherein the time window data refers to the time period during which the financial business demand is allowed to be executed.

7. The method according to claim 1, characterized in that After obtaining N business demand characteristics, the method further includes: Perform differential privacy processing on the N business demand features to obtain N processed business demand features, and use the N processed business demand features to execute the step of selecting a target model from the M candidate models based on the N-1 business features.

8. A model processing device for financial scenarios, characterized in that: include: A first acquisition unit is used to acquire financial service requirements, perform feature recognition on the financial service requirements, and obtain N service requirement features, wherein the N service requirement features include a service type and N-1 service features, where N is a positive integer; A determination unit, configured to determine M candidate models according to the service type, and select a target model from the M candidate models according to the N-1 service characteristics, wherein M is a positive integer; The prediction unit is used to predict the business load of the target model by using the load prediction model, obtain business load prediction data, and send the business load prediction data and the target model to the client.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored executable program, wherein when the executable program is run, the device where the computer-readable storage medium is located is controlled to execute the model processing method for financial scenarios described in any one of claims 1 to 7.

10. An electronic device, characterized in that: include: A memory storing an executable program; A processor is used to run the program, wherein the program, when running, executes the model processing method for financial scenarios described in any one of claims 1 to 7.

11. A computer program product comprising computer instructions, characterized in that: When the computer instructions are executed by a processor, the steps of the model processing method for financial scenarios described in any one of claims 1 to 7 are implemented.

Images