Method, device and electronic equipment for obtaining service feature information

Abstract

The application discloses a kind of acquisition methods, device and electronic equipment of business feature information.It involves artificial intelligence field, the method includes: the business data of target business and target feature set are input to target large language model, and the preset tool information and instruction information of each target feature are obtained by processing, wherein, target feature set includes M target features, and M target features are the characteristics required for executing the preset business operation of target business;According to the instruction information and preset tool information of each target feature respectively, the corresponding preset tool is called, and the characteristic value of target feature is generated by preset tool, and M characteristic values are obtained;M target features and corresponding M characteristic values are determined as the business feature information of preset business operation, wherein, business feature information is used to support target model to execute preset business operation.Through the present application, the problem that the accuracy and efficiency of obtaining business feature information in related technology are low is solved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and more specifically, to a method, apparatus, and electronic device for acquiring business characteristic information. Background Technology

[0002] With the gradual application of neural network models in financial institutions, more and more business operations are being processed using models. The automated execution of some business operations (such as credit approval and risk identification) depends on the accurate processing of business feature information by the target model. Therefore, in order to improve the accuracy of the results and efficiency of business operations, it is necessary to ensure the accuracy of the business feature information input to the model and improve the efficiency of acquiring such information.

[0003] However, currently, the acquisition of business feature information mainly relies on manual processing. Data analysts or engineers manually analyze business logic and write code to obtain the required business feature information from multiple sources of data. This process is highly dependent on domain knowledge, has a low degree of automation, is prone to generating erroneous feature information, and reduces the efficiency of acquiring business feature information.

[0004] There is currently no effective solution to the problem of low accuracy and efficiency in obtaining business characteristic information in related technologies. Summary of the Invention

[0005] The main objective of this application is to provide a method, apparatus, and electronic device for acquiring business characteristic information, so as to solve the problem of low accuracy and efficiency in acquiring business characteristic information in related technologies.

[0006] To achieve the above objectives, according to one aspect of this application, a method for obtaining business feature information is provided. The method includes: inputting business data and a set of target features into a target large language model; processing to obtain preset tool information and instruction information for each target feature; wherein the target feature set includes M target features, which are features required to perform preset business operations of the target business, and M is a positive integer; calling the corresponding preset tool according to the instruction information and preset tool information of each target feature, and generating feature values ​​for the target features through the preset tool to obtain M feature values; and determining the M target features and their corresponding M feature values ​​as business feature information for the preset business operation, wherein the business feature information is used to support the target model in performing the preset business operation.

[0007] Optionally, inputting the business data and target feature set of the target business into the target large language model includes: obtaining a prompt word template associated with a preset business operation, and adding the business data and target feature set to the prompt word template to obtain an initial prompt word; obtaining the business version of the target business, and querying the preset requirements under the business version, wherein the preset requirements are used to limit the requirements for obtaining the feature values ​​of each target feature; updating the initial prompt word according to the preset requirements to obtain the target prompt word, and inputting the target prompt word into the target large language model.

[0008] Optionally, the instruction information is a code instruction. After obtaining the instruction information, the method further includes: for any instruction information, obtaining the code metadata of the instruction information; determining test data to test the instruction information based on the code metadata; using the test data to test the instruction information and obtaining test results; and generating a first prompt message if the test results indicate an anomaly, wherein the first prompt message indicates that the instruction information is abnormal.

[0009] Optionally, calling the corresponding preset tool based on the instruction information and preset tool information of each target feature includes: dividing the instruction information according to function to obtain multiple instruction segments; determining the preset tool information corresponding to each instruction segment; and calling the preset tool corresponding to each instruction segment in sequence according to the execution order of the instruction segments in the instruction information.

[0010] Optionally, generating feature values ​​of the target feature using a preset tool includes: for any target feature, inputting the instruction information of the target feature into the preset tool and obtaining the output result of the preset tool, wherein the output result is obtained by running the instruction information of the preset tool; preprocessing the output result to obtain an initial value and obtaining the feature requirement information of the target feature; determining whether the initial value meets the feature requirement information based on the feature requirement information; if the initial value does not meet the feature requirement information, outputting a second prompt message, wherein the second prompt message indicates that the initial value is abnormal; if the initial value meets the feature requirement information, determining the initial value as the feature value of the target feature.

[0011] Optionally, determining the M target features and their corresponding M feature values ​​as the business feature information for the preset business operation includes: obtaining the input conditions of the target model and determining whether the M target features and M feature values ​​meet the input conditions; if the M target features and / or M feature values ​​do not meet the input conditions, processing the M target features and / or M feature values ​​according to the input conditions to obtain input data; and determining the input data as the business feature information.

[0012] Optionally, the target large language model is trained as follows: multiple sample features and a flowchart for acquiring each sample feature are obtained, and the tool information and instruction information of each node in the acquisition flowchart are determined; each sample feature and its corresponding acquisition flowchart, as well as the tool information and instruction information of each node in the acquisition flowchart, are determined as a set of sample data to obtain multiple sets of sample data; the language model is trained using multiple sets of sample data to obtain the target large language model.

[0013] To achieve the above objectives, according to another aspect of this application, a device for acquiring business feature information is provided. The device includes: a first processing unit, configured to input business data and a set of target features into a target large language model, and process to obtain preset tool information and instruction information for each target feature, wherein the set of target features includes M target features, which are features required to perform preset business operations of the target business, and M is a positive integer; a calling unit, configured to call the corresponding preset tool according to the instruction information and preset tool information of each target feature, and generate feature values ​​of the target features through the preset tool to obtain M feature values; and a second processing unit, configured to determine the M target features and the corresponding M feature values ​​as business feature information for preset business operations, wherein the business feature information is used to support the target model in performing preset business operations.

[0014] Optionally, the first processing unit includes: a first acquisition module, used to acquire a prompt word template associated with a preset business operation, and add business data and a target feature set to the prompt word template to obtain an initial prompt word; a second acquisition module, used to acquire the business version of the target business, and query the preset requirements under the business version, wherein the preset requirements are used to limit the acquisition requirements of the feature values ​​of each target feature; and an update module, used to update the initial prompt word according to the preset requirements to obtain the target prompt word, and input the target prompt word into the target large language model.

[0015] Optionally, the instruction information is a code instruction. After obtaining the instruction information, the device further includes: a first acquisition unit, configured to acquire the code metadata of any instruction information; a first determination unit, configured to determine test data for testing the instruction information based on the code metadata; a testing unit, configured to test the instruction information using the test data and obtain test results; and a generation unit, configured to generate a first prompt message when the test results indicate an abnormality, wherein the first prompt message indicates that the instruction information is abnormal.

[0016] Optionally, the calling unit includes: a division module, used to divide the instruction information according to function to obtain multiple instruction segments; a first determination module, used to determine the preset tool information corresponding to each instruction segment; and a calling module, used to call the preset tool corresponding to each instruction segment in sequence according to the execution order of the instruction segments in the instruction information and the preset tool information.

[0017] Optionally, the calling unit includes: a third acquisition module, used to input the instruction information of any target feature into a preset tool and acquire the output result of the preset tool for any target feature, wherein the output result is obtained through the preset tool information running instruction information; a first processing module, used to preprocess the output result to obtain an initial value and acquire the feature requirement information of the target feature; a first judgment module, used to judge whether the initial value meets the feature requirement information according to the feature requirement information; an output module, used to output a second prompt information when the initial value does not meet the feature requirement information, wherein the second prompt information indicates that the initial value is abnormal; and a second determination module, used to determine the initial value as the feature value of the target feature when the initial value meets the feature requirement information.

[0018] Optionally, the second processing unit includes: a second judgment module, used to obtain the input conditions of the target model and judge whether the M target features and M feature values ​​meet the input conditions; a second processing module, used to process the M target features and / or M feature values ​​according to the input conditions to obtain input data when the M target features and / or M feature values ​​do not meet the input conditions; and a third determination module, used to determine the input data as business feature information.

[0019] Optionally, the target large language model is trained using the following apparatus: a second acquisition unit, used to acquire multiple sample features and an acquisition flowchart for each sample feature, and to determine the tool information and instruction information of each node in the acquisition flowchart; a second determination unit, used to determine each sample feature and its corresponding acquisition flowchart, as well as the tool information and instruction information of each node in the acquisition flowchart, as a set of sample data, to obtain multiple sets of sample data; and a training unit, used to train the language model using the multiple sets of sample data to obtain the target large language model.

[0020] To achieve the above objectives, according to another aspect of this application, an electronic device is provided, the electronic device including a memory storing an executable program; and a processor for running the program, wherein the program executes the above-described method for obtaining business characteristic information during runtime.

[0021] To achieve the above objectives, according to another aspect of this application, a computer program product is provided, including computer instructions, which, when executed by a processor, implement the steps of the above-described method for obtaining business characteristic information.

[0022] In this embodiment, the business data and target feature set of the target business are input into a target large language model. The model processes the data to obtain preset tool information and instruction information for each target feature. The target feature set includes M target features, which are features required to execute preset business operations of the target business, where M is a positive integer. The model calls the corresponding preset tool based on the instruction information and preset tool information for each target feature, and generates feature values ​​for the target feature using the preset tool, resulting in M ​​feature values. The M target features and their corresponding M feature values ​​are then determined as the business feature information for the preset business operation. This business feature information supports the target model in executing the preset business operation. By parsing the business data using the target large language model, the instruction information and preset tool information corresponding to each business feature are determined. The preset tool executes the instruction information to obtain feature values. Finally, the target features and their corresponding feature values ​​are determined as the business feature information. This achieves the goal of determining the feature values ​​of each target feature through a large model and obtaining specific feature values ​​by calling tools, thereby improving the accuracy and efficiency of obtaining business feature information and solving the technical problem of low accuracy and efficiency in obtaining business feature information in related technologies. Attached Figure Description

[0023] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0024] Figure 1 A hardware structure block diagram of a computer terminal for implementing a method for acquiring business characteristic information is shown.

[0025] Figure 2 This is a flowchart of a method for obtaining business feature information according to Embodiment 1 of this application;

[0026] Figure 3 This is a schematic diagram of a device for acquiring business feature information according to Embodiment 2 of this application;

[0027] Figure 4 This is a structural block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0028] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0029] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0030] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application 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 this application 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 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.

[0031] It should be noted that the methods, devices, and electronic devices for obtaining business characteristic information as defined in this disclosure can be used in the field of artificial intelligence, or in any field other than artificial intelligence, such as the field of fintech. The application fields of the methods, devices, and electronic devices for obtaining business characteristic information as defined in this disclosure are not limited.

[0032] It should be noted that all information, user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, and displayed data) used in this application 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 related data all comply with the relevant regulations and standards of the relevant regions, have taken necessary measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse use. If the user chooses to refuse, the process proceeds to the expert decision-making process. For example, this system has interfaces 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. After receiving consent from the aforementioned user or organization, the relevant information is obtained. Users can view the purpose of data use in real time through the authorization interface and have the right to withdraw authorization or delete data at any time. After authorization is withdrawn, the system will terminate the relevant data processing within 24 hours.

[0033] The embodiments or examples disclosed herein are not exhaustive, but merely illustrative of some embodiments or examples, and are not intended to limit the scope of protection of this disclosure. Unless otherwise specified, each step in a particular embodiment or example can be implemented as an independent embodiment, and the steps can be arbitrarily combined. For example, a solution after removing some steps in a particular embodiment or example can also be implemented as an independent embodiment, and the order of the steps in a particular embodiment or example can be arbitrarily interchanged. Furthermore, optional methods or examples in a particular embodiment or example can be arbitrarily combined; moreover, embodiments or examples can be arbitrarily combined. For example, some or all steps of different embodiments or examples can be arbitrarily combined, and a particular embodiment or example can be arbitrarily combined with optional methods or examples of other embodiments or examples.

[0034] Example 1

[0035] According to an embodiment of this application, an embodiment of a method for obtaining business feature information 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.

[0036] 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 for implementing a method for acquiring business characteristic information is shown. Figure 1 As shown, the computer terminal 10 (or mobile device) may include one or more processors 102 (shown as 102a, 102b, ..., 102n in the figure) 102 (processor 102 may include, but is not limited to, processing devices such as microprocessors or programmable logic devices), 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, a universal serial bus port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. 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.

[0037] 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).

[0038] 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 method for obtaining business feature information 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 above-mentioned method for obtaining business feature information. 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.

[0039] 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.

[0040] The display may be, for example, a touchscreen LCD display that allows the user to interact with the user interface of the computer terminal 10 (or mobile device).

[0041] Under the aforementioned operating environment, this application provides the following: Figure 2 The method for obtaining business characteristic information is shown. Figure 2 This is a flowchart of a method for obtaining business feature information according to Embodiment 1 of this application, as shown below. Figure 2 As shown, the method includes:

[0042] Step S201: Input the business data and target feature set of the target business into the target large language model, and process to obtain the preset tool information and instruction information of each target feature. The target feature set includes M target features, which are the features required to perform the preset business operations of the target business, and M is a positive integer.

[0043] It should be noted that the execution entity in this embodiment can be a business feature information acquisition system. After the user inputs a preset business operation, the system can automatically acquire the business data and target feature set of the target business, and determine the business feature information required to execute the preset business operation based on the business data and target feature set of the target business.

[0044] It should be noted that business data refers to relevant data of the target business within a financial institution. The target business can be any business within the financial institution, including external business such as loans and deposits, and internal business such as operation log processing or code testing. This embodiment does not limit the target business. The target feature set refers to a predefined set of features containing M target features, which are the feature information required to execute subsequent preset business operations. The target large language model refers to the artificial intelligence model used as the inference unit. Preset tool information refers to the information of the software tool or framework type used to perform specific feature calculation tasks, such as tool name and tool version number. Instruction information refers to the specific execution code or query statement generated based on the preset tool information, such as the code for inputting tool A or the query statement for inputting tool B.

[0045] For example, when performing a preset business operation of a target business, it is first necessary to obtain and parse the business data of the target business, and also to obtain a set of target features associated with the preset business operation. M target features are predefined in this set for performing the preset business operation.

[0046] Furthermore, after obtaining the business data and target feature set, the business data and target feature set can be input into the target large language model. Using the target large language model as the core reasoning unit, combined with its built-in domain knowledge base and memory components, the input data is deeply analyzed to obtain preset tool information and instruction information.

[0047] During the analysis, the target large language model can use the thinking chain technology to decompose and plan the acquisition path of each feature, and identify the computational logic required for each target feature. For example, for target feature A, the computational logic can be to first obtain intermediate variables through tool A, and then obtain the feature value of target feature A based on tool B and intermediate variables.

[0048] After determining the calculation logic for each target feature, corresponding preset tool information and instruction information can be generated based on the calculation logic for each target feature, so that subsequent steps can obtain the feature value of each target feature based on the preset tool information and instruction information.

[0049] It should be noted that in the above process, the target large language model can be configured with a self-checking mechanism to ensure that the generated instruction information has correct syntax and logical integrity. If potential errors are found, the prompt words can be automatically adjusted and regenerated until the instruction information meets the execution standard, thereby improving the accuracy of the preset tool information and instruction information.

[0050] This step leverages the intelligent reasoning capabilities of the target large language model to automatically generate executable feature value acquisition logic and corresponding instruction and tool information, thereby improving the accuracy and efficiency of subsequent feature value acquisition.

[0051] Step S202: Call the corresponding preset tool according to the instruction information and preset tool information of each target feature, and generate the feature value of the target feature through the preset tool to obtain M feature values.

[0052] For example, after obtaining the instruction information and preset tool information corresponding to each target feature, the feature value of each target feature can be determined separately. For any target feature, the corresponding preset tool can be dynamically selected and assigned according to the preset tool information. The preset tools include, but are not limited to, tools that support data processing libraries, tools that support query languages, and real-time computing tools that support stream processing frameworks.

[0053] After determining the preset tools, instruction information can be sent to the preset tools to trigger the tools to execute the corresponding calculation tasks. For example, if the instruction information is type A code, then type A toolset is invoked to execute the code; if the instruction information is type S query statement, then type S toolset is invoked to execute the query.

[0054] During the execution of corresponding tasks based on instruction information, the preset tool can read relevant data from business data, perform data cleaning, transformation, aggregation or statistical calculation according to instruction information, and finally generate specific numerical values ​​of target features, i.e. feature values.

[0055] It should be noted that if any of the preset tools encounters an error during task execution, the error information can be captured, the error handling mechanism can be triggered, and the instruction information can be regenerated to try again until the correct feature value is obtained.

[0056] Finally, the system can follow the above process to calculate the feature values ​​of all M target features in sequence, and obtain the return results of each preset tool to obtain the feature value of each target feature.

[0057] Step S203: Determine the M target features and their corresponding M feature values ​​as the business feature information for the preset business operation, wherein the business feature information is used to support the target model in performing the preset business operation.

[0058] It should be noted that the target model refers to a trained machine learning model that can be used to execute corresponding preset business operations based on business feature information. Different preset business operations correspond to different target models.

[0059] For example, after obtaining the feature values ​​of each target feature, the M target features and their corresponding M feature values ​​can be determined as the business feature information of the preset business operation, thereby completing the business feature information acquisition operation. At this time, the business feature information can be input into the target model, and the target model can process the business feature information to obtain the operation result of the preset business operation.

[0060] It should be noted that in the above process, the system can also combine model output results and model performance indicators to continuously optimize and iterate the target model to improve recognition accuracy. Simultaneously, the model's memory component stores the recognition results and related feature values, forming new historical experience for feature construction and model training in subsequent similar scenarios.

[0061] The method for obtaining business feature information provided in this application embodiment involves inputting the business data and target feature set of the target business into a target large language model, processing them to obtain preset tool information and instruction information for each target feature. The target feature set includes M target features, which are features required to execute preset business operations of the target business, where M is a positive integer. The method calls the corresponding preset tool based on the instruction information and preset tool information of each target feature, and generates feature values ​​for the target feature using the preset tool, resulting in M ​​feature values. The M target features and their corresponding M feature values ​​are then determined as the business feature information for the preset business operation. This business feature information supports the target model in executing the preset business operation. By parsing the business data using the target large language model, the instruction information and preset tool information corresponding to each business feature are determined. The preset tool executes the instruction information to obtain feature values. Finally, the target features and their corresponding feature values ​​are determined as the business feature information. This achieves the goal of determining the feature values ​​of each target feature through a large model and obtaining specific feature values ​​by calling tools, thereby improving the accuracy and efficiency of obtaining business feature information and solving the technical problem of low accuracy and efficiency in obtaining business feature information in related technologies.

[0062] Optionally, in the method for obtaining business feature information provided in the embodiments of this application, inputting the business data and target feature set of the target business into the target large language model includes: obtaining a prompt word template associated with a preset business operation, and adding the business data and target feature set to the prompt word template to obtain an initial prompt word; obtaining the business version of the target business, and querying the preset requirements under the business version, wherein the preset requirements are used to limit the requirements for obtaining the feature values ​​of each target feature; updating the initial prompt word according to the preset requirements to obtain the target prompt word, and inputting the target prompt word into the target large language model.

[0063] It should be noted that the prompt word template refers to a predefined structured framework used to standardize the input format of the target large language model. The initial prompt word refers to the intermediate text generated after directly embedding the original business data and target feature set into the prompt word template. The business version refers to the specific version identifier of the current target business, used to distinguish target businesses from different periods or specifications. Preset requirements refer to the set of business rules defined for a specific business version, used to limit the requirements for obtaining feature values ​​for each target feature, which may include data range and precision standards. The target prompt word refers to the final prompt word after being corrected according to the preset requirements.

[0064] For example, when inputting the business data and target feature set of the target business into the target large language model, the system first needs to retrieve a prompt word template matching the current preset business operation from the database. This template defines the basic reasoning logic and output specifications that the target large language model must follow. Subsequently, the system fills the received business data and target feature set into the corresponding positions of the template to generate initial prompt words. At this time, the initial prompt words only contain basic data content and general task descriptions, lacking fine-grained constraints for specific business versions.

[0065] Furthermore, it is necessary to identify the current business version of the target service and query the corresponding preset requirements accordingly. These preset requirements specify how the feature values ​​of each target characteristic should be obtained under the current business version, including specific calculation methods, data filtering conditions, and precision limitations. Based on the retrieved preset requirements, the system dynamically updates the initial prompts. This update process may include inserting specific feature acquisition constraints into the prompts, adjusting task instructions to match the characteristics of the business version, and adding handling guidelines for abnormal situations, thereby generating target prompts containing complete business rules.

[0066] Finally, the system inputs the optimized target prompt words into the target large language model. Since the target prompt words have been integrated with specific business rules and constraints, the target large language model can more accurately understand the specific acquisition requirements of each target feature, thereby generating preset tool information and instruction information that conform to the specifications.

[0067] This embodiment obtains a prompt word template, processes the prompt word template to obtain a target prompt word corresponding to the target feature set, and inputs the target prompt word into the target large language model, thereby achieving the technical effect of improving the accuracy of the preset tool information and instruction information output by the target large language model.

[0068] Optionally, in the method for obtaining business feature information provided in the embodiments of this application, the instruction information is a code instruction. After obtaining the instruction information, the method further includes: for any instruction information, obtaining the code metadata of the instruction information; determining test data for testing the instruction information based on the code metadata; using the test data to test the instruction information and obtaining test results; and generating a first prompt message if the test results indicate an abnormality, wherein the first prompt message indicates that the instruction information is abnormal.

[0069] It should be noted that code instructions refer to the executable program code or scripts generated by the target large language model to drive the preset tools to perform specific computational logic. Code metadata refers to structured information describing the code instruction structure, dependencies, variable definitions, and execution environment characteristics, used to guide the generation and matching of test data. Test data refers to input samples constructed based on the code metadata to verify the correctness, robustness, and boundary conditions of code instructions, including normal input, abnormal input, and boundary values. Test results refer to the status feedback or output generated after executing the test data, used to characterize whether the code instructions work as expected. The first prompt information refers to the alarm or correction signal generated when an abnormal test result is detected, used to indicate that the current instruction information has defects and trigger subsequent repair or retry mechanisms.

[0070] For example, after receiving instruction information generated by the target large language model, it is necessary to verify the validity of the instruction information in order to verify its usability.

[0071] First, it is necessary to extract the code metadata of the code instruction. The code metadata covers the code's syntax structure, the library dependencies it introduces, the definitions of input and output parameters, and the characteristics of logical branches.

[0072] Furthermore, test data needs to be dynamically generated based on the acquired code metadata. This can be achieved by analyzing the variable types and data ranges in the code metadata. The test data should not only include data that conforms to normal business logic, but also include abnormal data that may cause errors (such as null values, type mismatches, out-of-range values) and boundary value data to comprehensively test the robustness of the code instructions.

[0073] Furthermore, the generated test data is used to test the code instructions. The code instructions can be run in an isolated sandbox environment with the test data, and the execution process can be monitored, capturing the final output or any exceptions. After the test, the system analyzes the execution feedback to obtain the test results, which record whether the code ran successfully, whether the output meets the expected format, and whether any exceptions or logical errors were thrown.

[0074] Finally, the system judges the test results. If the test results are normal, that is, the code executes without error and the output meets expectations, the process continues. If the test results are abnormal, that is, the code execution fails, an uncaught exception is thrown, or the output result is incorrect, the system generates the first prompt message, indicating that there is an anomaly in the current instruction information. This message can be fed back to the target large language model to trigger the code correction mechanism, or notify the coordination component to intervene manually or adjust the strategy, thereby preventing the erroneous code from entering the subsequent feature value calculation stage.

[0075] This embodiment improves the accuracy and usability of code instructions generated by the target large language model by introducing an automated testing mechanism based on code metadata.

[0076] Optionally, in the method for obtaining business feature information provided in the embodiments of this application, calling the corresponding preset tool according to the instruction information and preset tool information of each target feature includes: dividing the instruction information according to function to obtain multiple instruction segments; determining the preset tool information corresponding to each instruction segment; and calling the preset tool corresponding to each instruction segment in sequence according to the execution order of the instruction segments in the instruction information.

[0077] It should be noted that an instruction segment refers to a sub-part obtained by breaking down a complete instruction message according to its functional module or logical unit of execution. The execution order is the order in which the instruction segments are arranged in the original instruction message.

[0078] For example, when invoking a preset tool, since the instruction information may contain multiple types of operations (such as data cleaning, aggregation calculation, real-time stream processing, etc.), directly invoking a preset tool as a whole cannot cover all needs or will overload the tool. Therefore, the instruction information can be parsed first, and divided according to its internal functional logic. For example, it can be identified which parts belong to data preprocessing, which parts belong to statistical analysis, and which parts belong to real-time calculation, thereby splitting the complete instruction information into multiple instruction segments, each corresponding to a specific functional unit.

[0079] Furthermore, for each of the divided instruction segments, the system analyzes its functional characteristics, data scale, and real-time requirements to determine the corresponding preset tool information. For example, if an instruction segment involves the aggregation and statistics of large-scale historical data, the system labels its corresponding preset tool information as tool H; if an instruction segment involves simple numerical transformation or small sample data processing, the system labels its corresponding preset tool information as tool P. Through this fine-grained mapping, the execution tool for each instruction segment can be determined.

[0080] Finally, following the execution order of the instruction segments in the original instruction information, the system sequentially calls the preset tools corresponding to each instruction segment based on the determined preset tool information, maintaining the coherence of the data processing logic and the correctness of dependencies. For example, if instruction segment A is data cleaning and instruction segment B is feature statistics, and the result of A is the input of B, the system will first call the preset tool corresponding to A to perform cleaning. After A has finished executing and output the result, the result will then be passed to the preset tool corresponding to B to perform statistics, thereby obtaining the final output feature values.

[0081] This embodiment achieves modular execution of complex feature engineering tasks through a step-by-step, orderly invocation mechanism, thus achieving the technical effect of accurately determining preset tools.

[0082] Optionally, in the method for obtaining business feature information provided in this application embodiment, generating feature values ​​of target features through a preset tool includes: for any target feature, inputting the instruction information of the target feature into the preset tool and obtaining the output result of the preset tool, wherein the output result is obtained by running the instruction information of the preset tool; preprocessing the output result to obtain an initial value and obtaining feature requirement information of the target feature; determining whether the initial value meets the feature requirement information based on the feature requirement information; outputting a second prompt information if the initial value does not meet the feature requirement information, wherein the second prompt information indicates that the initial value is abnormal; and determining the initial value as the feature value of the target feature if the initial value meets the feature requirement information.

[0083] For example, when determining feature values, for any target feature, the corresponding instruction information needs to be input into a specified preset tool first. The preset tool then runs the instruction information according to the preset tool information, performs the corresponding data calculation or transformation operation, and returns the output result.

[0084] At this point, the output may contain raw data formats, non-standard values, or noisy data. Directly using these as feature values ​​could lead to errors in subsequent model training or analysis. Therefore, the output can be preprocessed to obtain initial values. Preprocessing operations can include data cleaning, type conversion (such as converting strings to floating-point numbers), missing value handling, and outlier filtering. After preprocessing, the data is normalized into numerical values ​​of a uniform format, i.e., the initial values.

[0085] At the same time, in order to check the initial value, it is also necessary to obtain the feature requirement information corresponding to the target feature from the memory component or knowledge base. The feature requirement information specifies in detail the attributes that the feature should have, such as the value must be positive, the precision must be retained to two decimal places, or it must fall within a specific business range.

[0086] After obtaining the feature requirement information, the initial value can be verified based on it. If the initial value meets all the requirements contained in the feature requirement information, it is determined that there is no need to modify the initial value, and the initial value can be directly determined as the final feature value of the target feature.

[0087] If the initial value does not meet any of the requirements contained in the feature requirement information, it is determined that the initial value does not meet the feature requirement information, and a second prompt message is output. The second prompt message indicates that the initial value is abnormal and triggers the system's error handling mechanism, such as recording an error log, triggering manual review, or requesting the target large language model to regenerate the instruction information, thereby improving the accuracy of the feature value.

[0088] This embodiment achieves the technical effect of improving the accuracy and compliance of feature values ​​by preprocessing and verifying the initial values ​​output by the tool.

[0089] Optionally, in the method for obtaining business feature information provided in this application embodiment, determining M target features and corresponding M feature values ​​as business feature information for a preset business operation includes: obtaining the input conditions of the target model and determining whether the M target features and M feature values ​​meet the input conditions; if the M target features and / or M feature values ​​do not meet the input conditions, processing the M target features and / or M feature values ​​according to the input conditions to obtain input data; and determining the input data as business feature information.

[0090] For example, when determining business feature information, it is also necessary to obtain the input conditions of the target model. These input conditions define the specific specifications for the data received by the target model, such as the length of the feature vector, the numerical type of each feature (e.g., floating-point number, integer), the value range (e.g., the normalized 0-1 interval), and the rules for handling missing values.

[0091] Furthermore, the M target features and M feature values ​​need to be compared with the acquired input conditions to determine whether each target feature and its corresponding feature value meets the input conditions. The judgment process may include checking whether the number of feature values ​​is equal to the dimension required by the model, checking whether the type and range of each feature value are within the allowed range, and checking whether there is any abnormal data that does not meet the preprocessing requirements.

[0092] If the judgment result is that all M target features and M feature values ​​meet the input conditions, then the M target features and M feature values ​​can be directly identified as business feature information and thus directly used as input data for the target model. If the judgment result is that any target feature or feature value does not meet the input conditions, then adaptation processing is performed, that is, according to the specific constraints of the input conditions, the target features or feature values ​​are processed accordingly. For example, if the input conditions require data normalization, then the feature values ​​are normalized; if there are missing values, then they are filled; if the dimensions do not match, then they are pruned or filled; if the data types do not match, then they are converted. After the above processing, input data that meets the input conditions is obtained.

[0093] Finally, the system uses the target features and corresponding feature values ​​in the generated input data as key-value pairs, and determines multiple key-value pairs as the final business feature information, thereby completing the business feature information acquisition operation.

[0094] This embodiment improves the accuracy and effectiveness of the final business feature information by introducing an input condition verification and adaptive processing mechanism to process the obtained M target features and their corresponding M feature values.

[0095] Optionally, in the method for obtaining business feature information provided in the embodiments of this application, the target large language model is trained in the following manner: obtaining multiple sample features and a flowchart for obtaining each sample feature, and determining the tool information and instruction information of each node in the flowchart; determining each sample feature and its corresponding flowchart, as well as the tool information and instruction information of each node in the flowchart, as a set of sample data to obtain multiple sets of sample data; using the multiple sets of sample data to train the language model to obtain the target large language model.

[0096] It should be noted that the flowchart refers to a structural diagram that records the task planning, tool call order, and dependencies required to generate sample features, reflecting the logical path of feature construction.

[0097] For example, when training a target large language model, it is first necessary to obtain multiple sample features and a flowchart for obtaining each sample feature. The flowchart details the complete construction process from raw data to final feature values, including data preprocessing, feature extraction, selection, and derivation. Simultaneously, the system parses each node in the flowchart to determine the tool and instruction information corresponding to each node.

[0098] Furthermore, the system combines each sample feature, its corresponding acquisition flowchart, and the tool and instruction information of each node in the flowchart to determine a set of sample data, thereby establishing a mapping relationship between the target feature entity and the complete logical path to achieve that feature. By processing all sample features, the system constructs multiple sets of sample data, which constitute the basic dataset for training the language model, covering feature construction paradigms under different business scenarios.

[0099] Finally, the system uses these multiple sets of sample data to train the language model. During training, the model learns how to generate reasonable acquisition flowcharts, tool selection strategies, and specific instruction information based on the input business requirements by analyzing the feature distribution patterns, logical dependencies, and tool-instruction matching patterns in the sample data. The model iteratively processes multiple sets of sample data, continuously adjusting its internal parameters to minimize the difference between the predicted results and the actual sample data. After training, the resulting model is the target large language model. This model has the ability to automatically generate feature engineering task plans containing tool and instruction information based on the user's input business requirements.

[0100] This embodiment achieves the technical effect of intelligently constructing a target large language model by using historical features to build experience to train the language model.

[0101] 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.

[0102] Example 2

[0103] This application also provides a device for acquiring business feature information. It should be noted that the device for acquiring business feature information in this application can be used to execute the method for acquiring business feature information provided in the above embodiments. The following describes the device for acquiring business feature information provided in this application.

[0104] According to an embodiment of this application, an apparatus for implementing the above-described method for obtaining business feature information is also provided. Figure 3This is a schematic diagram of a device for acquiring business feature information according to Embodiment 2 of this application, as shown below. Figure 3 As shown, the device includes:

[0105] The first processing unit 31 is used to input the business data and target feature set of the target business into the target large language model, and process it to obtain the preset tool information and instruction information of each target feature. The target feature set includes M target features, which are the features required to perform the preset business operations of the target business, and M is a positive integer.

[0106] The calling unit 32 is used to call the corresponding preset tool according to the instruction information and preset tool information of each target feature, and generate the feature value of the target feature through the preset tool to obtain M feature values.

[0107] The second processing unit 33 is used to determine the M target features and the corresponding M feature values ​​as business feature information for preset business operations, wherein the business feature information is used to support the target model to perform preset business operations.

[0108] The business feature information acquisition device provided in this application embodiment inputs the business data and target feature set of the target business into the target large language model through the first processing unit 31, and processes it to obtain the preset tool information and instruction information of each target feature. The target feature set includes M target features, which are features required to perform preset business operations of the target business, where M is a positive integer. The calling unit 32 calls the corresponding preset tool according to the instruction information and preset tool information of each target feature, and generates feature values ​​of the target feature through the preset tool, obtaining M feature values. The second processing unit 33 determines the M target features and the corresponding M feature values ​​as a pre-defined set. The present invention establishes business characteristic information for business operations, which supports the target model in executing preset business operations. By parsing business data using a target large language model, the invention determines the instruction information and preset tool information corresponding to each business characteristic. The preset tool executes the instruction information to obtain the feature value. Finally, the target feature and its corresponding feature value are determined as the business characteristic information. This achieves the goal of determining the feature value of each target feature through a large model and obtaining the specific feature value by calling the tool. This improves the accuracy and efficiency of obtaining business characteristic information and solves the technical problem of low accuracy and efficiency in obtaining business characteristic information in related technologies.

[0109] Optionally, in the business feature information acquisition device provided in this application embodiment, the first processing unit 31 includes: a first acquisition module, used to acquire a prompt word template associated with a preset business operation, and add business data and a target feature set to the prompt word template to obtain an initial prompt word; a second acquisition module, used to acquire the business version of the target business, and query the preset requirements under the business version, wherein the preset requirements are used to limit the acquisition requirements of the feature values ​​of each target feature; and an update module, used to update the initial prompt word according to the preset requirements to obtain the target prompt word, and input the target prompt word into the target large language model.

[0110] Optionally, in the business feature information acquisition device provided in this application embodiment, the instruction information is a code instruction. After obtaining the instruction information, the device further includes: a first acquisition unit, used to acquire the code metadata of any instruction information; a first determination unit, used to determine test data for testing the instruction information based on the code metadata; a testing unit, used to test the instruction information using the test data and obtain test results; and a generation unit, used to generate a first prompt message when the test result indicates an abnormality, wherein the first prompt message indicates that the instruction information is abnormal.

[0111] Optionally, in the business feature information acquisition device provided in the embodiments of this application, the calling unit 32 includes: a division module, used to divide the instruction information according to function to obtain multiple instruction segments; a first determination module, used to determine the preset tool information corresponding to each instruction segment respectively; and a calling module, used to call the preset tool corresponding to each instruction segment in sequence according to the execution order of the instruction segments in the instruction information.

[0112] Optionally, in the business feature information acquisition device provided in this application embodiment, the calling unit 32 includes: a third acquisition module, used to input the instruction information of any target feature into a preset tool for any target feature, and acquire the output result of the preset tool, wherein the output result is obtained through the preset tool information running instruction information; a first processing module, used to preprocess the output result to obtain an initial value, and acquire the feature requirement information of the target feature; a first judgment module, used to judge whether the initial value meets the feature requirement information according to the feature requirement information; an output module, used to output a second prompt information when the initial value does not meet the feature requirement information, wherein the second prompt information indicates that the initial value is abnormal; and a second determination module, used to determine the initial value as the feature value of the target feature when the initial value meets the feature requirement information.

[0113] Optionally, in the business feature information acquisition device provided in the embodiments of this application, the second processing unit 33 includes: a second judgment module, used to acquire the input conditions of the target model and judge whether the M target features and M feature values ​​meet the input conditions; a second processing module, used to process the M target features and / or M feature values ​​according to the input conditions to obtain input data when the M target features and / or M feature values ​​do not meet the input conditions; and a third determination module, used to determine the input data as business feature information.

[0114] Optionally, in the business feature information acquisition device provided in this application embodiment, the target large language model is trained by the following method: a second acquisition unit is used to acquire multiple sample features and an acquisition flowchart for each sample feature, and to determine the tool information and instruction information of each node in the acquisition flowchart; a second determination unit is used to determine each sample feature and its corresponding acquisition flowchart, as well as the tool information and instruction information of each node in the acquisition flowchart, as a set of sample data to obtain multiple sets of sample data; and a training unit is used to train the language model using multiple sets of sample data to obtain the target large language model.

[0115] It should be noted that the first processing unit 31, the calling unit 32, and the second processing unit 33 mentioned above correspond to steps S201 to S203 in Embodiment 1. The instances and application scenarios implemented by each of the above units and the corresponding steps are the same, but are not limited to the content disclosed in Embodiment 1. It should be noted that the above modules or 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 modules can also be part of a device and can run in the computer terminal 10 provided in Embodiment 1.

[0116] Example 3

[0117] Embodiments of this application may provide an electronic device. Figure 4 This is a structural block diagram of an electronic device according to an embodiment of this application. Figure 4 As shown, the electronic device may include: one or more ( Figure 4 (Only one is shown) processor 1002, memory 1004, memory controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0118] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the methods and apparatus in the embodiments of this application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby implementing the above-described methods. 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.

[0119] Those skilled in the art will understand that Figure 4 The structure shown is for illustrative purposes only. Electronic devices can also be smartphones, tablets, handheld computers, mobile internet devices (MIDs), PADs, and other terminal devices. Figure 4 This does not limit the structure of the aforementioned electronic device. For example, electronic devices may also include components that are more... Figure 4 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 4 The different configurations shown.

[0120] 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.

[0121] Example 4

[0122] 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 obtaining business feature information provided in Embodiment 1.

[0123] 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.

[0124] Embodiments of this application also provide a computer program product, which, when executed on a data processing device, is a program adapted to perform the steps of a method for obtaining business characteristic information.

[0125] Embodiments of this application also provide a computer-readable storage medium, which includes a stored executable program, wherein the executable program controls the device where the computer-readable storage medium is located to execute the above-described method for obtaining business feature information when it runs.

[0126] 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.

[0127] 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.

[0128] 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.

[0129] 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.

[0130] 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.

[0131] 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.

[0132] 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 obtaining business characteristic information, characterized in that, include: The business data and target feature set of the target business are input into the target large language model, and the preset tool information and instruction information of each target feature are obtained. The target feature set includes M target features, which are the features required to perform the preset business operations of the target business, and M is a positive integer. Each target feature is assigned a corresponding preset tool based on its instruction information and preset tool information, and the preset tool is used to generate the feature value of the target feature, resulting in M ​​feature values. The M target features and their corresponding M feature values ​​are determined as the business feature information of the preset business operation, wherein the business feature information is used to support the target model in executing the preset business operation.

2. The method according to claim 1, characterized in that, Inputting the target business data and target feature set into the target large language model includes: Obtain a prompt word template associated with the preset business operation, and add the business data and the target feature set to the prompt word template to obtain the initial prompt word; Obtain the service version of the target service and query the preset requirements under the service version, wherein the preset requirements are used to limit the requirements for obtaining the feature values ​​of each target feature; The initial prompt word is updated according to the preset requirements to obtain the target prompt word, and the target prompt word is input into the target large language model.

3. The method according to claim 1, characterized in that, The instruction information is a code instruction. After obtaining the instruction information, the method further includes: For any instruction information, obtain the code metadata of the instruction information; The test data for testing the instruction information is determined based on the code metadata; The instruction information is tested using the test data to obtain test results; If the test result indicates an anomaly, a first prompt message is generated, wherein the first prompt message indicates that the instruction information is abnormal.

4. The method according to claim 1, characterized in that, The corresponding preset tools are invoked based on the instruction information and preset tool information for each target feature, including: The instruction information is divided according to function to obtain multiple instruction segments; Determine the preset tool information corresponding to each instruction segment; According to the execution order of the instruction segments in the instruction information, the preset tools corresponding to each instruction segment are called sequentially according to the preset tool information.

5. The method according to claim 1, characterized in that, Generating the feature values ​​of the target feature using the preset tool includes: For any target feature, the instruction information of the target feature is input into the preset tool, and the output result of the preset tool is obtained, wherein the output result is obtained by running the instruction information through the preset tool information; The output results are preprocessed to obtain initial values, and the feature requirement information of the target feature is obtained. Determine whether the initial value meets the feature requirement information based on the feature requirement information; If the initial value does not meet the feature requirement information, a second prompt message is output, wherein the second prompt message indicates that the initial value is abnormal; If the initial value meets the feature requirement information, the initial value is determined as the feature value of the target feature.

6. The method according to claim 1, characterized in that, The business feature information for determining the M target features and their corresponding M feature values ​​as the preset business operation includes: Obtain the input conditions of the target model, and determine whether the M target features and the M feature values ​​meet the input conditions; If the M target features and / or the M feature values ​​do not meet the input conditions, the M target features and / or the M feature values ​​are processed according to the input conditions to obtain input data; The input data is identified as the business characteristic information.

7. The method according to claim 1, characterized in that, The target large language model is trained in the following way: Acquire multiple sample features and a flowchart for acquiring each sample feature, and determine the tool information and instruction information for each node in the acquisition flowchart; Each sample feature, its corresponding acquisition flowchart, and the tool and instruction information of each node in the acquisition flowchart are determined as a set of sample data, resulting in multiple sets of sample data. The language model is trained using the multiple sets of sample data to obtain the target large language model.

8. A device for acquiring business characteristic information, characterized in that, include: The first processing unit is used to input the business data and target feature set of the target business into the target large language model, and process it to obtain the preset tool information and instruction information of each target feature. The target feature set includes M target features, which are features required to perform preset business operations of the target business, and M is a positive integer. The calling unit is used to call the corresponding preset tool according to the instruction information and preset tool information of each target feature, and generate the feature value of the target feature through the preset tool to obtain M feature values; The second processing unit is used to determine the M target features and the corresponding M feature values ​​as the business feature information of the preset business operation, wherein the business feature information is used to support the target model to execute the preset business operation.

9. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the method for obtaining business feature information as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program executes the method for obtaining business feature information according to any one of claims 1 to 7 when it runs.

Images