Service processing method, device and equipment

Abstract

The invention provides a service processing method, device and equipment. The scheme comprises the following steps: acquiring business appeal information in a natural language form sent by a user terminal; then, the business appeal information is input into a trained big language model, a request method identifier output by the big language model is obtained, the request method identifier is used for representing an interface request method to be executed in response to the business appeal information, and the big language model is obtained based on business appeal information sample training; the business appeal information sample carries a request method identification label; then, a calling instruction used for calling an application interface is generated based on the request method identifier; and requesting a business application corresponding to the application interface to execute a business process corresponding to the business appeal information based on the calling instruction.

Description

Technical Field

[0001] The present application relates to the field of large model technology, and in particular to a business processing method, device and equipment. Background Art

[0002] In today's digital age, service providers are facing increasingly complex and huge data operation and maintenance needs. The traditional operation and maintenance model usually relies on users, business or operation personnel to put forward specific requirements, and then these requirements are passed to technical personnel, who manually call the database or perform related operations based on their experience to implement them. The traditional operation and maintenance model requires manual participation in multiple links, and it often takes a long time from the proposal of requirements to the completion of operation and maintenance tasks. The process is cumbersome and inefficient, which not only affects the continuity and stability of the business, but may also lead to a decline in user satisfaction. In addition, since each link requires manual participation and coordination, the risk of errors is increased, and the operation and maintenance costs are increased.

[0003] In view of this, it is necessary to improve the existing data operation and maintenance business to simplify the process, improve efficiency and reduce operation and maintenance costs. Summary of the invention

[0004] In view of this, embodiments of the present application provide a business processing method, apparatus and device for simplifying the processing flow of existing operation and maintenance business, improving efficiency and reducing costs.

[0005] According to a first aspect of an embodiment of the present application, a service processing method is provided, including: Obtaining business demand information in natural language form sent by the user terminal; The business demand information is input into the trained large language model to obtain a request method identifier output by the large language model; the request method identifier is used to indicate the interface request method to be executed in response to the business demand information; the large language model is trained based on a business demand information sample, and the business demand information sample carries a request method identifier tag; Based on the request method identifier, generating a calling instruction for calling an application interface; Based on the calling instruction, a business application corresponding to the application interface is requested to execute a business process corresponding to the business demand information.

[0006] According to a second aspect of an embodiment of the present application, a service processing device is provided, including: A business demand information acquisition module is used to acquire business demand information in natural language form sent by a user terminal; A large model module is used to input the business demand information into the trained large language model to obtain a request method identifier output by the large language model; the request method identifier is used to indicate an interface request method to be executed in response to the business demand information; the large language model is trained based on a business demand information sample, and the business demand information sample carries a request method identifier tag; A calling instruction generation module, used to generate a calling instruction for calling an application interface based on the request method identifier; The business execution module is used to request the business application corresponding to the application interface to execute the business process corresponding to the business demand information based on the calling instruction.

[0007] According to a third aspect of an embodiment of the present application, a service processing device is provided, including: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: Obtaining business demand information in natural language form sent by the user terminal; The business demand information is input into the trained large language model to obtain a request method identifier output by the large language model; the request method identifier is used to indicate the interface request method to be executed in response to the business demand information; the large language model is trained based on a business demand information sample, and the business demand information sample carries a request method identifier tag; Based on the request method identifier, generating a calling instruction for calling an application interface; Based on the calling instruction, a business application corresponding to the application interface is requested to execute a business process corresponding to the business demand information.

[0008] One embodiment of the present specification can achieve at least the following beneficial effects: a large language model is obtained by pre-training based on business demand information samples carrying request method identification tags, and then the business demand information in natural language form obtained from the user terminal is provided to the trained large language model to obtain the request method identification output by the large language model. After that, a call instruction for calling an application interface can be generated based on the request method identification, and then the business application corresponding to the application interface is requested to execute the business process corresponding to the business demand information, thereby completing the business processing. As a result, the user of the user terminal can trigger an intelligent data operation and maintenance process through natural language, without having to submit the requirements to technical personnel and then have the technical personnel manually operate to implement data operation and maintenance business, which simplifies the business process, improves efficiency, and reduces costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0010] Figure 1 A schematic diagram of an application scenario of an intelligent operation and maintenance solution provided in an embodiment of this specification; Figure 2 A flowchart of a business processing method provided in an embodiment of this specification; Figure 3 The embodiments of this specification provide corresponding to Figure 2 A structural schematic diagram of a business processing device; Figure 4 The embodiments of this specification provide corresponding to Figure 2 A structural diagram of a business processing device. DETAILED DESCRIPTION

[0011] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0012] The terms used in one or more embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of the present application. The singular forms of "a", "said" and "the" used in one or more embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in one or more embodiments of the present application refers to and includes any or all possible combinations of one or more associated listed items.

[0013] It should be understood that, although the terms first, second, etc. may be used to describe various information in one or more embodiments of the present application, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of the present application, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0014] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards in the relevant regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0015] In the traditional operation and maintenance process, first, the service provider's users, business personnel or operation personnel need to clarify their needs, which may involve operations such as querying, modifying, deleting or updating data, or may involve the presentation of abnormal business phenomena; then, users, business personnel or operation personnel can convey their needs to technical personnel in some form (such as email, phone, instant messaging tools, meetings, etc.); after receiving the needs, the technical personnel need to understand the needs and convert them into specific database operation statements, and then manually execute these statements to complete the operation and maintenance tasks. This process is not only cumbersome, but also inefficient because it involves the participation and communication of multiple roles, which increases the risk of errors and operation and maintenance costs. In addition, the traditional operation and maintenance model also has the problem of slow response speed. Because it requires the manual participation of multiple roles and relies heavily on the participation and experience of technical personnel, it often takes a long time from the proposal of needs to the completion of operation and maintenance tasks, which not only affects the continuity and stability of the business, but also may lead to a decline in user satisfaction.

[0016] In actual applications, service providers can often provide services to many customers at the same time, and in most cases, the demands raised by multiple customers are highly repetitive. For example, a service provider can provide data operation and maintenance management services to multiple customers of similar nature (for example, financial institutions). For example, institution A may have raised the demand to adjust the risk level of products, and institution B may also have raised the demand to adjust the risk level of products. From a technical point of view, the operation and maintenance operations corresponding to these business demands are very similar. For example, they all call the same application interface to operate, and the only difference is that the parameters input during the call are different. In this case, for technical personnel, there are repeated operations, which wastes manpower and is inefficient.

[0017] In view of this, in the embodiments of this specification, an intelligent method for providing data operation and maintenance business services is provided, which is efficient and saves labor costs. Specifically, by using a large model, the natural language requirements of the demander can be converted into the data required to call the application interface, and then the application program can be automatically called to implement operation and maintenance, which greatly reduces the repetitive workload of technicians, reduces operation and maintenance costs, and improves daily operation and maintenance efficiency.

[0018] The technical solutions provided by the embodiments of this specification are described in detail below in conjunction with the accompanying drawings.

[0019] Figure 1 A schematic diagram of an application scenario of an intelligent operation and maintenance solution provided in an embodiment of this specification.

[0020] like Figure 1 As shown, in this implementation scenario, a user terminal 100, a first server 200 and a second server 300 are included. Specifically, a user can send business demand information in natural language to the first server 200 through the user terminal 100; the first server 200 can call a pre-trained large language model to convert the business demand information in natural language into call information of an application interface; then, the first server 200 can send a call instruction to the second server 300 where the business application corresponding to the application interface is located based on the call information of the application interface; then, the second server 300 can execute the business process corresponding to the business demand information based on the call instruction.

[0021] Large Language Model (LLM) refers to a deep learning model with a large number of parameters and rich pre-training. They show superior understanding and generation capabilities when processing natural language tasks, and can perform complex language reasoning and creative text generation based on the context. They are widely used in dialogue systems, text summarization, translation and other language processing scenarios.

[0022] like Figure 1 The large language model shown in can be a large language model obtained by fine-tuning and training based on the basic large language model using manually annotated data, which is referred to as a trained large language model in the embodiments of this specification. The trained large language model has the ability to convert the user's natural language into interface call information, and more specifically, has the ability to convert business demand information in natural language into a request method identifier of an interface request method to be executed in response to the business demand information.

[0023] In the embodiments of this specification, users may include customers served by a service provider, or may include business personnel, operation personnel, etc. on the service provider side. Figure 1 The user terminal 100 shown in the figure may include but is not limited to any form of terminal equipment such as a smart phone, a tablet computer, a laptop computer, a PDA, a personal computer, etc.

[0024] like Figure 1The first server 200 shown in the figure may include but is not limited to any device, equipment, platform, equipment cluster, etc. with computing and processing capabilities. Specifically, the first server 200 may be used to: after receiving the business demand information in the form of natural language provided by the user and sent by the user terminal 100, firstly call the large language model to parse the natural language demand into the calling information of the application interface; then, based on the calling information of the application interface, call the second server 300 to perform data operation and maintenance business processing and obtain the business processing result; afterwards, the business processing result may also be fed back to the user terminal 100.

[0025] In the embodiment of the present specification, an application program running in the second server 200 is provided for receiving request information from the user terminal 100, calling the large language model for data processing and receiving information output by the large language model, calling the second server 300 for business processing and receiving the processing result fed back by the second server 300, and feeding back the processing result to the user terminal 100. The application program connects the entire business process in series and can be called a series application. In actual application, the user can send business demand information in natural language form to the series application running on the first server 200 through the terminal application (for example, instant messaging application, browser, etc.) in the user terminal 100.

[0026] Although for the purpose of clarity, Figure 1 The large language model is shown as being separated from the first server 200, but in actual application, the large language model can be run on the first server 200 or on other computing devices outside the first server 200. This application does not limit the running device of the large language model.

[0027] like Figure 1 The second server 300 shown in the figure may include but is not limited to any device, equipment, platform, equipment cluster, etc. with computing and processing capabilities. Specifically, the second server 300 may respond to the call request for the business application sent by the first server 200, execute the data operation and maintenance business requested by the call request through the business application running on the second server 300, that is, execute the data operation and maintenance business corresponding to the business demand information, and feed back the business processing result to the first server 200.

[0028] In actual applications, after the first server 200 receives the business execution result fed back by the second server 300, it can optionally feed back the business execution result directly to the user terminal 100 according to the content of the actual business demand information. For example, for some business demands of obtaining data, the first server 200 can directly return the data obtained from the second server 300 based on the interface call information to the user terminal 100; or optionally, the first server 200 can further process the business execution result to obtain a result that matches the user's business demand information and then feed it back to the user terminal 100. For example, for business demands of the reason inquiry type, the first server 200 can obtain the data from the second server 300 based on the interface call information, analyze the data to obtain the analysis conclusion corresponding to the user's business demand, and then return the analysis conclusion to the user terminal 100.

[0029] In such Figure 1 In the application scenario shown, the first server 200 may be connected to one or more user terminals 100 via a local area network connection, a wide area network connection, an Internet connection or other types of data networks.

[0030] Based on the solution of the embodiment of this specification, using Figure 1 In the system shown, users can describe their operation and maintenance requirements in natural language, and the system automatically performs corresponding operation and maintenance operations based on the user's requirements, thus realizing intelligent operation and maintenance. In view of this, the operation and maintenance costs are reduced and the operation and maintenance efficiency is improved.

[0031] In the present application, a business processing method is provided. The present application also relates to a business processing apparatus and a computing device, which are described in detail one by one in the following embodiments.

[0032] Figure 2 A flowchart of a business processing method provided in an embodiment of this specification.

[0033] From the program perspective, the execution subject of the process can be a program installed on the application server. It can be understood that the method can be executed by any device, equipment, platform, or device cluster with computing and processing capabilities.

[0034] like Figure 2 As shown, the process may include the following steps.

[0035] Step 202: Acquire the business demand information in natural language form sent by the user terminal.

[0036] Among them, the user terminal may include a terminal used by a user who makes a business demand. Among them, the user who makes a business demand may specifically include users who use services, operators, business personnel, etc., but is not limited thereto. For example, users who use services may be users who use services provided by service providers, for example, users who use Internet services such as e-commerce services, e-financial services, logistics services, big data cloud computing services, advertising services, and cross-border trade services. Operators may be personnel who perform project management, product operations, user operations, risk management, and the like in the process of providing Internet services by service providers. Business personnel may be personnel who perform product promotion, business contracts, and the like in the process of providing Internet services by service providers. It is understandable that the personnel who use user terminals to propose business demand information are not limited to the examples here, for example, they may also include product department personnel who perform product design and the like.

[0037] The business demand information may be information used to express the business demand of the user, and may be information in natural language form. In actual application, the business demand information in natural language form may be expressed in the form of text, voice, etc., but is not limited to these forms. The business demand information may be sent by the user to the terminal application running on the user terminal, such as Figure 2 The execution body of the method shown.

[0038] As an example, in a financial management business scenario, the content of the business demand information can be "Bank A's customer growth rate in 2024", "Deposit amount in the past three years", "User 20888****'s appointment account opening requests in the past three years", "Why user China Life's 30-day income has not been updated", etc.

[0039] Step 204: Input the business demand information into the trained large language model to obtain the request method identifier output by the large language model; the request method identifier is used to indicate the interface request method that should be executed in response to the business demand information; the large language model is trained based on business demand information samples, and the business demand information samples carry a request method identifier label.

[0040] Typically, the training of a large language model includes two stages: pre-training and fine-tuning.

[0041] Pre-training stage: The model is first trained on a huge text corpus to learn the language structure, semantic relationships, and contextual information of the text. The goal of pre-training is usually to allow the model to learn a wide range of language representations through language model tasks such as predicting the next word or filling in the gaps in a sentence. The key to this stage is to use massive amounts of data and powerful computing resources to enable the model to capture the deep laws and knowledge of the language.

[0042] Fine-tuning phase: After pre-training is complete, the model is fine-tuned to adapt to a specific downstream task. Fine-tuning is usually performed on a task-specific dataset, and the objective function is directly related to the downstream task. For example, in a text classification task, the model adjusts its parameters based on the classification label to improve classification accuracy. The fine-tuning process can significantly improve the performance of the model on a specific task because it allows the model to combine the general language representation obtained in pre-training with the needs of a specific task.

[0043] In the embodiments of this specification, the trained large language model used can be obtained by fine-tuning the basic large language model. In the process of fine-tuning training, business demand information samples carrying request method labels can be used for training, so that the trained large model has the ability to determine the interface request method to be executed in response to the business demand information based on the business demand information.

[0044] In the embodiments of this specification, the interface may specifically refer to an application programming interface (Application Programming Interface, referred to as API), which is a set of predefined functions that aims to provide application programs and developers with the ability to access a set of routines based on certain software or hardware without having to access the source code or understand the details of the internal working mechanism. In the following, the application programming interface is also referred to as an application interface.

[0045] In a common data operation and maintenance scenario, a technician needs to write a call instruction for calling the application interface of a business application so as to call the business application to perform the data operation and maintenance operation corresponding to the business demand. Figure 2 The execution subject of the method shown replaces the technical personnel. When receiving business demand information in natural language form, it uses a large language model to determine the interface request method that should be executed in response to the business demand information, and then automatically calls the corresponding application interface to execute the business process, thereby realizing intelligent operation and maintenance.

[0046] Step 206: Based on the request method identifier, generate a calling instruction for calling the application interface.

[0047] The request method identifier may represent an identifier of a method for calling an application program interface. In actual application, the identifier may specifically be a method name, for example, an API method name.

[0048] In actual application, the execution subject can query and generate the necessary metadata required for the call instruction from the database for storing interface metadata based on the request method identifier, for example, the request service identifier, which can be a service name or an interface package name, for example, an API service name. Then, based on the request method identifier and the necessary metadata such as the request service identifier, a call instruction for calling the application interface can be generated.

[0049] Usually, the API service name refers to the name of a complete API interface, which represents the functionality or service scope of the entire interface. An API service name corresponds to a specific set of functions and operations. The service name of an API service is unique and represents the entire API interface. The API method name refers to the name of the function or command used to perform specific operations within the API service. An API method name corresponds to a specific operation or functional implementation, such as GET, POST, PUT, DELETE, etc. An API service usually contains multiple method names, each of which corresponds to a different specific operation.

[0050] Step 208: Based on the calling instruction, request the business application corresponding to the application interface to execute the business process corresponding to the business demand information.

[0051] The business application may represent an application program running on the server. The business process corresponding to the business demand information may represent a business process to be executed in order to complete the demand expressed by the business demand information. The specific business process may vary from business to business.

[0052] Based on the solutions of the embodiments of this specification, users can conveniently call business applications to perform specific business operations by providing business demands in natural language, thereby realizing business processing without the intervention of technical personnel, reducing labor costs and improving business processing efficiency.

[0053] It should be understood that in the methods described in one or more embodiments of this specification, the order of some steps can be adjusted according to actual needs, or some steps can be omitted.

[0054] Figure 2The method in the invention obtains a large language model by pre-training based on business demand information samples carrying request method identification tags, and then provides the business demand information in natural language form obtained from the user terminal to the trained large language model to obtain the request method identification output by the large language model. After that, a call instruction for calling the application interface can be generated based on the request method identification, and then the business application corresponding to the application interface is requested to execute the business process corresponding to the business demand information, so as to complete the business processing. Therefore, the user of the user terminal can trigger the intelligent data operation and maintenance process through natural language, without submitting the requirements to the technical personnel first and then manually implementing the data operation and maintenance business by the technical personnel, which simplifies the business process, improves efficiency and reduces costs.

[0055] based on Figure 2 The present specification also provides some improved implementation methods of the method, which are described below.

[0056] In one or more embodiments of the present specification, in order to enable the large model to more accurately obtain the business results that the user wants to obtain, prompt word engineering can be applied to optimize the business demand information in the natural language form input by the user.

[0057] Specifically, the step of inputting the business demand information into the trained large language model to obtain the request method identifier output by the large language model may include: converting the business demand information into model input information that conforms to the prompt word template according to a prompt word template; and inputting the model input information into the trained large language model to obtain the request method identifier output by the large language model.

[0058] In actual application, a prompt word template may be pre-set, and the business demand information input by the user may be rewritten into a form that conforms to the prompt word template before being input into the large language model.

[0059] The prompt word template may reflect the function of the large language model, or in other words, the prompt word template may guide the large language model to implement the preset function. Considering that the function expected to be implemented by the large model in the embodiment of this specification is based on the business demand information in natural language form, and the interface request method required to implement the demand expressed by the business demand information is determined, then, as an example, the prompt template may be in the format of "converting user questions into standard JSON for calling the API:".

[0060] Assuming that the business demand information in natural language form provided by the user is "Bank A's customer growth rate in 2024", then after rewriting the business demand information into a form that conforms to the prompt word template, the resulting content may be "Convert the user's question into standard JSON for calling the API: Bank A's customer growth rate in 2024".

[0061] In one or more embodiments of the present specification, the large language model used to generate the request method identification based on the business demand information is pre-trained, and more specifically, is obtained by fine-tuning the basic large model in advance. During the fine-tuning training process, business demand information samples carrying identification tags can be used for training. A specific method for fine-tuning training is provided herein.

[0062] Specifically, before inputting the business demand information into the trained large language model to obtain the request method identifier output by the large language model, it can also include: obtaining a business demand information sample carrying a request method identifier label; and using the business demand information sample to fine-tune the basic large language model to obtain the trained large language model.

[0063] The business demand information sample carries a request method identification tag, which means that the business demand sample carries a tag, and the carried tag is used to indicate the interface request method that should be executed in response to the business demand information sample. Therefore, by fine-tuning the basic large language model using business demand information samples carrying request method identification tags, the large model can learn the ability to convert business demand information in natural language form into request method identification.

[0064] For example, a sample of business demand information may be "Check the account opening record of user 2088xxxx at Bank A last week", and the label of the interface request method to be executed corresponding to the sample may be marked as "apiMethod-1". For another example, a sample of business demand information may be "Please help me find the account entry flow of 2099xxxx", and the label of the interface request method to be executed corresponding to the sample may be marked as "apiMethod-2".

[0065] In one or more embodiments of the present specification, in order to improve the training effect at a lower training cost, the corpus for fine-tuning training can be expanded. Specifically, on the basis of obtaining real historical appeal information, not only can the real historical appeal information be used as the basic fine-tuning training sample, but also a large number of derived fine-tuning training samples can be generated by replacing keywords in the real historical appeal information.

[0066] Optionally, the acquiring of the business demand information sample carrying the request method identification tag may specifically include: acquiring historical demand information in a natural language format; and modifying the value of a replaceable parameter in the historical demand information to obtain multiple business demand information samples.

[0067] As an example, the historical demand information in natural language form may be, "Check the account opening record of user 208811 at Bank A last week." In this example, it can be identified from the historical demand information that the values ​​of replaceable parameters may include the user ID "208811", the time range of the business to be checked "last week", the institution ID to be checked "Bank A", and the business type to be checked "account opening". Therefore, multiple business demand information samples can be obtained by modifying the values ​​of at least some of the aforementioned multiple replaceable parameters.

[0068] For example, the time range of the business to be checked "last week" can be changed to "last month", and the identification of the institution to be checked "Bank A" can be changed to "Bank B", etc. For another example, the user identification "208811", the time range of the business to be checked "last week", the identification of the institution to be checked "Bank A", and the type of business to be checked "opening an account" can all be replaced.

[0069] In actual application, in a historical demand information, one, two or more replaceable parameters that meet the preset parameter type can be identified. In this case, the values ​​of one, two or more replaceable parameters among the identified replaceable parameters can be replaced.

[0070] Optionally, modifying the value of the replaceable parameter in the historical claim information to obtain multiple business claim information samples may include: identifying replaceable parameters of a preset parameter type from the historical claim information; replacing the parameter value of the replaceable parameter with the parameter identifier of the replaceable parameter to obtain a claim information template corresponding to the historical claim information; obtaining a list of replaceable values ​​for the replaceable parameter; and then, replacing the parameter identifier in the claim information template with the replaceable value in the list of replaceable values ​​to obtain a business claim information sample.

[0071] The preset parameter type of the replaceable parameter may be pre-defined, and in actual application, the preset parameter type of the replaceable parameter may correspond to the business scenario.

[0072] Optionally, the replaceable parameters may include at least one of a user ID, a time range of the service to be checked, an institution ID to be checked, and a type of service to be checked. The time range of the service to be checked indicates the time range of the service to be queried.

[0073] In actual application, the claim information template corresponding to the historical claim information can be obtained based on the identification of all replaceable parameters. Assuming that the parameter identifier of the user ID is {userId}, the parameter identifier of the time range of the business to be checked is {time}, the parameter identifier of the institution ID to be checked is {instId}, and the parameter identifier of the business type to be checked is {orderType}, then the claim information template corresponding to the aforementioned historical claim information "Check the account opening record of user 208811 at Bank A last week" can be obtained as "Check the {instId}{orderType} record of user {userId}{time}".

[0074] Furthermore, a list of replaceable values ​​of replaceable parameters of the aforementioned preset parameter types can be obtained. When there are multiple replaceable parameters of the preset parameter types, a list of replaceable values ​​corresponding to each replaceable parameter can be obtained. Using the above example, a list of replaceable values ​​of the user identifier, a list of replaceable values ​​of the time range of the business to be checked, a list of replaceable values ​​of the institution identifier to be checked, and a list of replaceable values ​​of the business type to be checked can be obtained.

[0075] In one or more embodiments of the present specification, in order to ensure that the features of the expanded training samples do not deviate from the actual business scenarios, in actual applications, the replaceable values ​​in the replaceable value list of the preset parameter type may specifically include a set of values ​​of replaceable parameters of the preset parameter type extracted from a large amount of historical demand information under the same business scenario. For example, for an insurance business scenario, a set of user identifiers, a set of institution identifiers to be checked, a set of business types to be checked, and a set of business time ranges to be checked covered in the business scenario may be summarized based on a large amount of historical business demands of users. Since the training samples are expanded based on parameter values ​​extracted from real business scenarios as a list of replaceable values, the accuracy of the training samples obtained thereby is higher and the model fine-tuning training effect is better.

[0076] Furthermore, when the parameter identifier in the demand information template is replaced with a replaceable value in the replaceable value list, for example, by inserting a new parameter value, for the template "check the {instId}{orderType} record of user {userId}{time}", a business demand information sample such as "check the account binding record of user 208822 in Bank A last month" can be obtained.

[0077] In one or more embodiments of the present specification, the same demand information template may correspond to the same interface request method. The business demand information samples corresponding to the same demand information template may carry the same request method identification tag. In actual application, the business demand information generated based on the target demand information template and the historical business demand information that generates the target demand information template may be marked with the same request method identification tag. In view of the fact that a large number of business demand information samples are obtained based on the expansion of historical demand information and carry the same request method identification tag, the workload of marking training samples can be greatly reduced. For example, the business demand expressed by the demand information template "check the {instId}{orderType} record of user {userId}{time}" needs to be implemented by calling the interface request method "apiMethod1". Then, a large number of business demand information samples obtained based on the demand information template "check the {instId}{orderType} record of user {userId}{time}" can be marked as carrying the "apiMethod1" tag.

[0078] Optionally, different demand information templates may correspond to different interface request methods. For example, the business demand expressed by the demand information template "Please help me find out the {orderType} flow of {userId}" needs to be implemented by calling the interface request method "apiMethod2". Then, a large number of business demand information samples obtained based on the demand information template "Please help me find out the {orderType} flow of {userId}" can be marked as carrying the "apiMethod2" tag.

[0079] Optionally, different demand information templates may correspond to the same interface request method. This is because the demand information template is generated based on historical demand information in natural language form, and given that different natural language expressions may express the same business demand, there may be different demand information templates corresponding to the same interface request method. For example, the business demand expressed by the demand information template "Check the {instId}{orderType} records of user {userId}{time}" may be consistent with the business demand expressed by the demand information template "All {instId}{orderType} records of user {userId}" and both need to be implemented by calling the interface request method "apiMethod1". Then, a large number of business demand information samples obtained based on the demand information template "Check the {instId}{orderType} records of user {userId}{time}" and a large number of business demand information samples obtained based on the demand information template "All {instId}{orderType} records of user {userId}" can be marked as carrying the "apiMethod1" tag.

[0080] In one or more embodiments of this specification, when calling an application interface, some input parameters may need to be provided for carrying in the interface call request. These input parameters may be provided by the user in the business demand information, or may be extracted from the business demand information provided by the user. To this end, the embodiments of this specification also provide a technical solution for determining the request parameters for carrying in the interface call instruction from the business demand information in the natural language form provided by the user based on a trained large language model.

[0081] In an optional embodiment, the business demand information sample further carries a request parameter tag; the request parameter tag is used to indicate parameter type information of each user request parameter in at least one user request parameter carried in the business demand information sample.

[0082] Therefore, the basic large language model is fine-tuned and trained using the business demand information sample, and the trained large language model can not only identify the interface request method corresponding to the business demand information in natural language form, but also determine the input parameters carried in the business demand information in natural language form, so that these input parameters can be carried in the calling instruction when generating the calling instruction later.

[0083] Specifically, the step of inputting the business demand information into the trained large language model to obtain the request method identifier output by the large language model may specifically include: inputting the business demand information into the trained large language model to obtain the request method identifier and request parameters output by the large language model; the request parameters are used to represent the input parameters of the interface request method corresponding to the request method identifier. Then, the step of generating a call instruction for calling an application interface based on the request method identifier may specifically include: generating a call instruction for calling an application interface based on the request method identifier and the request parameters.

[0084] The request parameters may be carried in a call instruction issued to the application interface when requesting to execute a business process corresponding to the business demand information. In actual practice, the request service identifier corresponding to the request method identifier may be queried first; then, based on the request service identifier, the request method identifier and the request parameters, a call instruction for calling the application interface may be generated.

[0085] In an embodiment of the present specification, the request parameter identified based on the business demand information may be consistent with the replaceable parameter in the business demand information. In actual application, the request parameter may include at least one of a user identifier, a time range of the business to be checked, an institution identifier to be checked, and a business type to be checked.

[0086] Regarding the request parameters, optionally, in some business scenarios, the request parameters may include a user identifier. That is, the request parameters may include a user identifier, and the call request may carry a user identifier so that the called business application knows which user's corresponding data to query or process, for example, querying the account opening record of the user indicated by the user identifier, etc. Optionally, in some business scenarios, the request parameters may not include a user identifier, that is, the call request may not carry a user identifier, then the called business application may query or process information not associated with the user identity, or may query or process information associated with all users, and the specific processing method may be executed according to the business agreement, for example, querying the number of account opening personnel of a financial institution, etc. It can be seen that in actual application, it can be determined whether to carry a user identifier in the request parameters according to actual business needs.

[0087] In the embodiments of this specification, the user identifier refers to the identifier of the user targeted by the service proposed by the service demand information. In actual application, the user targeted by the service demand information and the terminal user using the user terminal to send the service demand information can be the same user or a different user. When the execution subject of the above method obtains the service demand information, it can usually obtain the identifier of the terminal user who sends the service demand information at the same time, and the service demand information may or may not carry the user identifier of the user targeted by the service demand information.

[0088] Regarding the request parameters, optionally, in some business scenarios, the request parameters may include a time range for the business to be checked. In this case, the business to be processed is related to time, and the call request may carry the time range for the business to be checked, so that the called business application knows which time range to query or process the business data. For example, the user's account flow within a certain time period can be queried. Optionally, in some business scenarios, the request parameters may not contain the time range for the business to be checked. In this case, the business to be processed is not related to time, and the call request may not carry the time range for the business to be checked. Then the called business application can query or process information that is not related to time. The specific processing method can be executed according to the business agreement. For example, a predetermined operation can be performed on a certain type of data. It can be seen that in actual application, it can be determined whether to carry the time range for the business to be checked in the request parameters according to actual business needs.

[0089] In an embodiment of the present specification, the time range of the business to be checked output by the large model can be time range information directly converted from the natural language information in the business demand information. Specifically, the information used to indicate the time range of the business to be checked is information that can be obtained without combining with the current time for calculation.

[0090] In actual application, optionally, the format of the time range of the business to be checked output by the large language model can conform to a preset first time expression. Specifically, the first time expression can use a time offset unit, an offset start time, and an offset end time to represent the time range; wherein the time offset unit includes a day, a week, a month, a quarter, or a year. The offset start time can be specifically used to represent the offset start time compared to the current time. The offset end time can be specifically used to represent the offset end time compared to the current time.

[0091] For example, the form of the preset first time expression can be: UI (begin, end). "UI" represents the time offset unit. In actual application, the value of the time offset unit can be d (day), w (week), m (month), s (season), y (year), etc., but is not limited thereto. "begin" can be used to represent the offset start time, for example, it can be -1, -3, etc. "end" can be used to represent the offset start time, for example, it can be 0, -1, etc.

[0092] Specifically, the generating of a calling instruction for calling an application interface based on the request method identifier may include: determining the start time and end time of the time range of the business to be checked according to the time offset unit, the offset start time and the offset end time; generating a calling instruction for calling an application interface based on the request method identifier and the start time and the end time.

[0093] As an example, if the business demand information raised by the user is, "Check the account binding records of user 2088****** in Bank A for the past three months", then the big model can output the time range of the business to be checked in the form of a first time expression m(-3,0); then, if it is assumed that the current time is 20241001, then m(-3,0) can be converted into the second time expression form (20240710,20241010).

[0094] In actual application, optionally, the format of the time range of the business to be checked output by the large language model conforms to a preset second time expression. Specifically, the second time expression uses a start time and an end time to represent the time range; wherein the start time is used to represent the start time of the business to be checked. The end time is used to represent the end time of the business to be checked. Specifically, the second time expression represents the operation and maintenance processing of the data after the start time and before the end time.

[0095] As an example, if the business demand information proposed by the user is "query the customer growth rate of Bank A in 2023", the big model can output the time range of the business to be queried in the form of a second time expression (20230101, 20231231).

[0096] Regarding the request parameters, optionally, in some business scenarios, the request parameters may also include information such as the identification of the organization to be checked and the type of business to be checked. The specific types of request parameters carried are related to the application interface to be called.

[0097] To facilitate understanding, the method provided in the embodiments of this specification is now described with examples.

[0098] Assume that the business demand information input by the user in natural language form is: "Check the account binding records of user 2088****** in Bank A for the past three months."

[0099] After receiving the business demand information in natural language form, it can be converted according to the prompt word template, for example: "Convert the user's question into a standard JSON for calling the API: Check the account binding records of user 2088****** at Bank A for the past three months."

[0100] After that, the converted information is provided to the trained large language model to obtain the output of the large language model. As required by the prompt word, the output of the large language model can be in JSON format. An example is as follows: { "apiMethod": "com.pay.service.facade.ai.ApiFacade#queryAccountOrder", "timeFormula": "m(-3,0)", "userId": "2088******", "instId": "CIB", "orderType": "bindSpAccount" }, in this example, "com.pay.service.facade.ai.ApiFacade#queryAccountOrder" corresponding to apiMethod is used to indicate the API method name of the interface request method that should be executed in response to the business demand information; "m(-3,0)" corresponding to timeFormula is used to indicate the time range of the business to be checked; "2088******" corresponding to userId is used to indicate the user ID; "CIB" corresponding to instId is used to indicate Bank A; "bindSpAccount" corresponding to orderType is used to indicate the type of business to be checked.

[0101] The above is only an example of model output. In actual application, optionally, the format of the model's output content may vary according to the content of the prompt word template, for example, it may be a format other than JSON format. Optionally, the items included in the model's output content may be different from this example. On the one hand, the items included in the model's output content are related to the request parameter labels carried by the training samples during model fine-tuning training. It can be understood that the set of items included in the model's output content may be a subset of the request parameter label set carried by the training samples during model fine-tuning training; on the other hand, the items included in the model's output content are related to the specific business demand information input by the user. It can be understood that the set of items included in the model's output content may be a subset of the request parameter set carried in the business demand information input by the user.

[0102] Based on the content output by the large model, if it contains information about the time range of the business to be checked and meets the preset first time expression, the format of the time range of the business to be checked can be converted first, and then a calling instruction for calling the application interface can be generated.

[0103] Using the above example, if the business time range m(-3,0) output by the large model meets the preset first time expression, further conversion is required. If the current time is assumed to be 20241001, m(-3,0) can be converted to the second time expression form (20240710,20241010). After that, a call instruction can be generated based on the business time range (20230101,20231231) in the second time expression form.

[0104] In actual application, when generating a call instruction, you can also first query the corresponding API service name from the metadata based on the API method name "com.pay.service.facade.ai.ApiFacade#queryAccountOrder", and then generate and send a call request to the application interface corresponding to the determined API service name and API method name.

[0105] In one or more embodiments of the present specification, a specific method for generating a calling instruction for calling an application interface based on a request method identifier is also provided.

[0106] Specifically, generating a calling instruction for calling an application interface based on the request method identifier may include: querying a request service identifier corresponding to the request method identifier; and generating a calling instruction for calling an application interface based on the request service identifier and the request method identifier.

[0107] Optionally, the query for the request service identifier corresponding to the request method identifier may specifically include: obtaining interface metadata of at least one available application interface from a database; the interface metadata including the available request service identifier and the available request method identifier of each available application interface in the at least one available application interface; based on the request method identifier, querying the request service identifier corresponding to the request method identifier from the interface metadata.

[0108] The interface metadata may include data used to describe the application interface. Optionally, the interface metadata may include a request method identifier (e.g., an API method name) and a request service identifier (e.g., an API service name). Optionally, the interface metadata may also include information such as an input parameter identifier, annotation, and function description of the request method.

[0109] Return to reference Figure 1 The first server 200 may also access a database for storing interface metadata ( Figure 1 In actual application, the database for storing interface metadata may be a database of the first server 200, which is directly accessed by the first server 200, or may be a database of another server other than the first server 200, and the first server 200 may access the database via another server.

[0110] In actual application, the database may store interface metadata corresponding to the available application interface. Specifically, the available application interface may be a pre-set application interface that can be intelligently accessed through the solution provided by the embodiment of this specification. Optionally, the available application interface may be an application interface corresponding to the request method identifier represented by the label information used to mark the training sample during the fine-tuning training of the large language model.

[0111] Optionally, based on the request method identifier, querying the request service identifier corresponding to the request method identifier from the interface metadata may specifically include: determining whether the request method identifier is included in the interface metadata, and obtaining a first judgment result; if the first judgment result indicates that the request method identifier is included in the interface metadata, then obtaining the request service identifier corresponding to the request method identifier from the interface metadata.

[0112] Optionally, if the first judgment result indicates that the request method identifier is not included in the interface metadata, an error prompt message is generated; the error prompt message is used to indicate that the large language model has not determined a request method identifier that complies with regulations.

[0113] Optionally, a first prompt message corresponding to the error prompt message may be sent to the user terminal. The first prompt message may be used to prompt the user of the user terminal to provide adjusted business demand information. In actual application, if the large language model fails to determine a request method identifier that complies with regulations because the business demand information input by the user is not clear or accurate enough, then the user may adjust the business demand information input in natural language form, and may re-execute the business processing flow by obtaining the adjusted business demand information provided by the user, in the hope that the large language model can determine a request method identifier that complies with regulations and execute the actual business flow when calling the application interface.

[0114] Alternatively, a second prompt message corresponding to the error prompt message may be sent to the technician's terminal. The second prompt message may be used to prompt the technician to determine the application interface to be called based on business experience and to execute a method call to execute the actual business process. In actual application, as a fallback solution, after the execution subject of the business processing method outputs the second prompt message, the technician may respond to the business demand information provided by the user to execute the corresponding business process.

[0115] In one or more embodiments of the present specification, when calling an application interface, in addition to the request method identifier (or, in addition to the request method identifier and the request service identifier), signature information may also need to be carried in the calling instruction according to the requirements of the pre-agreed communication protocol.

[0116] Specifically, generating a calling instruction for calling an application interface based on the request method identifier may include: querying the signature information corresponding to the request method identifier; generating a calling instruction for calling an application interface based on the request service identifier, the request method identifier and the signature information.

[0117] The signature information may be generated by the execution subject of the method of the embodiment of this specification before sending the call instruction. Optionally, the signature information may be used to reflect the identity of the device that initiates the call instruction.

[0118] In one or more embodiments of the present specification, a specific method for requesting a business application to execute a subsequent business process based on a calling instruction is also provided.

[0119] Optionally, based on the calling instruction, requesting the business application corresponding to the application interface to execute the business process corresponding to the business demand information may specifically include: based on the calling instruction, calling the business application corresponding to the application interface to obtain the first business data returned by the business application; and sending the first business data to the user terminal.

[0120] For example, for some business scenarios, the business demands raised by users of the user terminal may include data query demands. In this case, the data that the user wants to query can be obtained from the business database by calling the application, and then the obtained data can be fed back to the user terminal. During this period, there is no need to calculate and process the data returned by the calling instruction.

[0121] Optionally, based on the calling instruction, requesting the business application corresponding to the application interface to execute the business process corresponding to the business demand information may specifically include: based on the calling instruction, calling the business application corresponding to the application interface to obtain second business data returned by the business application; processing the second business data to obtain third business data; and sending the third business data to the user terminal.

[0122] For example, for some business scenarios, the business demands raised by users of the user terminal may include fault consultation demands. In this case, the basic data used to determine the fault diagnosis results can be queried from the business database by calling the business application. Then, the fault diagnosis results can be determined based on the basic data and fed back to the user terminal.

[0123] In actual application, the process of determining the information used to feed back to the user terminal based on the data returned by the calling instruction may specifically include calling the trained large language model, processing the second business data, and obtaining the third business data.

[0124] Optionally, based on the calling instruction, requesting the business application corresponding to the application interface to execute the business process corresponding to the business demand information may specifically include: based on the calling instruction, calling the business application corresponding to the application interface, performing an operation and maintenance operation on the business database, and obtaining an operation and maintenance operation result; the operation and maintenance operation result includes a first operation and maintenance operation result or a second operation and maintenance operation result; the first operation and maintenance operation result indicates that the operation and maintenance operation is successful; the second operation and maintenance operation result indicates that the operation and maintenance operation fails; and sending the operation and maintenance operation result to the user terminal.

[0125] For example, for some business scenarios, the business demands raised by users of the user terminal may include data processing demands. In this case, database operations can be performed on the data in the business database by calling the business application, and then the operation results can be fed back to the user terminal.

[0126] In addition, in the embodiments of the present specification, in order to enable users to obtain a better data exchange experience, the first server 200 can call a large language model to rewrite the information to be fed back before feeding the information back to the user terminal 100, so that the information sent to the user terminal 100 is more in line with the user's natural language expression habits, thereby improving the user experience.

[0127] Based on one or more solutions of the embodiments of this specification, most business requests that require repeated data operation and maintenance operations can be implemented without the need for in-depth involvement of technical personnel, which can improve business execution efficiency and reduce labor costs.

[0128] Moreover, the technical solution has good scalability. Specifically, when a new scenario needs to be added, the model can be fine-tuned using training sample data with new labels in the new scenario, so that the large model has the ability to recognize the request method identifier corresponding to the business demand information in the natural language form of the user in the scenario.

[0129] The various technical features in the above embodiments can be combined arbitrarily as long as there is no conflict or contradiction between the combinations of features. However, due to space limitations, they are not described one by one. Therefore, any combination of the various technical features in the above embodiments also falls within the scope of this specification.

[0130] Based on the same idea, the embodiments of this specification also provide a device corresponding to the above method.

[0131] Figure 3 The embodiments of this specification provide corresponding to Figure 2 A structural diagram of a business processing device.

[0132] like Figure 3 As shown, the device may include: The business demand information acquisition module 302 is used to acquire the business demand information in natural language form sent by the user terminal; The large model module 304 is used to input the business demand information into the trained large language model to obtain the request method identifier output by the large language model; the request method identifier is used to indicate the interface request method to be executed in response to the business demand information; the large language model is trained based on the business demand information sample, and the business demand information sample carries the request method identifier tag; A calling instruction generating module 306, configured to generate a calling instruction for calling an application interface based on the request method identifier; The business execution module 308 is used to request the business application corresponding to the application interface to execute the business process corresponding to the business demand information based on the calling instruction.

[0133] based on Figure 3 The present specification also provides some specific implementation schemes of the method, which are described below.

[0134] Optionally, the large model module 304 can be specifically used to: convert the business demand information into model input information that conforms to the prompt word template according to the prompt word template; input the model input information into the trained large language model to obtain the request method identifier output by the large language model.

[0135] Optionally, the device also includes a large model training module, which includes: a sample acquisition unit, used to obtain business demand information samples carrying request method identification tags; a fine-tuning training unit, in which a user uses the business demand information samples to fine-tune the basic large language model to obtain the trained large language model.

[0136] Optionally, the sample acquisition unit may be specifically configured to: acquire historical demand information in a natural language format; and modify values ​​of replaceable parameters in the historical demand information to obtain a plurality of business demand information samples.

[0137] Optionally, the modification of the value of the replaceable parameter in the historical claim information to obtain multiple business claim information samples may specifically include: identifying replaceable parameters of a preset parameter type from the historical claim information; replacing the parameter value of the replaceable parameter with the parameter identifier of the replaceable parameter to obtain a claim information template corresponding to the historical claim information; obtaining a list of replaceable values ​​for the replaceable parameter; and replacing the parameter identifier in the claim information template with the replaceable value in the list of replaceable values ​​to obtain a business claim information sample.

[0138] Optionally, the replaceable parameters include at least one of a user identifier, a time range of the business to be checked, an institution identifier to be checked, and a business type to be checked.

[0139] Optionally, the calling instruction generation module 306 may specifically include: a query unit, used to query the request service identifier corresponding to the request method identifier; and a calling instruction generation unit, used to generate a calling instruction for calling an application interface based on the request service identifier and the request method identifier.

[0140] Optionally, the query unit can be specifically used to: obtain interface metadata of at least one available application interface from a database; the interface metadata includes an available request service identifier and an available request method identifier of each available application interface in the at least one available application interface; based on the request method identifier, query the request service identifier corresponding to the request method identifier from the interface metadata.

[0141] Optionally, based on the request method identifier, querying the request service identifier corresponding to the request method identifier from the interface metadata may specifically include: determining whether the request method identifier is included in the interface metadata, and obtaining a first judgment result; if the first judgment result indicates that the request method identifier is included in the interface metadata, then obtaining the request service identifier corresponding to the request method identifier from the interface metadata.

[0142] Optionally, if the first judgment result indicates that the request method identifier is not included in the interface metadata, an error prompt message is generated; the error prompt message is used to indicate that the large language model has not determined a request method identifier that complies with regulations.

[0143] Optionally, generating a calling instruction for calling an application interface based on the request method identifier may specifically include: querying signature information corresponding to the request method identifier; generating a calling instruction for calling an application interface based on the request service identifier, the request method identifier and the signature information.

[0144] Optionally, the business demand information sample further carries a request parameter tag; the request parameter tag is used to indicate parameter type information of each user request parameter in at least one user request parameter carried in the business demand information sample.

[0145] Optionally, the large model module 304 is specifically used to: input the business demand information into the trained large language model to obtain the request method identifier and request parameters output by the large language model; the request parameters are used to represent the input parameters of the interface request method corresponding to the request method identifier; the call instruction generation module 306 is specifically used to: generate a call instruction for calling the application interface based on the request method identifier and the request parameters.

[0146] Optionally, the request parameter includes a time range of the service to be checked.

[0147] Optionally, the format of the time range of the business to be checked output by the large language model conforms to a preset first time expression; the first time expression uses a time offset unit, an offset start time, and an offset end time to represent the time range; wherein the time offset unit includes a day, a week, a month, a quarter, or a year.

[0148] Optionally, the calling instruction generation module 306 is specifically used to: determine the start time and end time of the business time range to be checked according to the time offset unit, the offset start time and the offset end time; and generate a calling instruction for calling the application interface based on the request method identifier and the start time and the end time.

[0149] Optionally, the service execution module 308 is specifically configured to: based on the calling instruction, call the service application corresponding to the application interface to obtain first service data returned by the service application; and send the first service data to the user terminal.

[0150] Optionally, the business execution module 308 is specifically used to: based on the calling instruction, call the business application corresponding to the application interface to obtain second business data returned by the business application; process the second business data to obtain third business data; and send the third business data to the user terminal.

[0151] Optionally, the business execution module 308 is specifically used to: based on the calling instruction, call the business application corresponding to the application interface, perform operation and maintenance operations on the business database, and obtain operation and maintenance operation results; the operation and maintenance operation results include a first operation and maintenance operation result or a second operation and maintenance operation result; the first operation and maintenance operation result indicates that the operation and maintenance operation is successful; the second operation and maintenance operation result indicates that the operation and maintenance operation fails; and send the operation and maintenance operation result to the user terminal.

[0152] It is understood that the above modules refer to computer programs or program segments for executing one or more specific functions. In addition, the distinction between the above modules does not mean that the actual program codes must also be separated.

[0153] The above is a schematic scheme of a service processing device of this embodiment. It should be noted that the technical scheme of the service processing device and the technical scheme of the service processing method described above are of the same concept, and the details not described in detail in the technical scheme of the service processing device can be found in the description of the technical scheme of the service processing method described above.

[0154] Based on the same idea, the embodiments of this specification also provide a device corresponding to the above method.

[0155] Figure 4 The embodiments of this specification provide corresponding to Figure 2 A structural diagram of a business processing device.

[0156] like Figure 4 As shown, the device 400 may include: at least one processor 410; and, A memory 430 is communicatively connected to the at least one processor; wherein, The memory 430 stores instructions 420 executable by the at least one processor 410. The instructions are executed by the at least one processor 410 to enable the at least one processor 410 to: Obtaining business demand information in natural language form sent by the user terminal; The business demand information is input into the trained large language model to obtain a request method identifier output by the large language model; the request method identifier is used to indicate the interface request method to be executed in response to the business demand information; the large language model is trained based on a business demand information sample, and the business demand information sample carries a request method identifier tag; Based on the request method identifier, generating a calling instruction for calling an application interface; Based on the calling instruction, a business application corresponding to the application interface is requested to execute a business process corresponding to the business demand information.

[0157] The above is a schematic scheme of a computing device of this embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the above-mentioned business processing method belong to the same concept, and the details not described in detail in the technical scheme of the computing device can be referred to the description of the technical scheme of the above-mentioned business processing method.

[0158] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the device and equipment embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments. The devices, equipment and methods provided in the embodiments of this specification correspond to each other, so the devices and equipment also have beneficial technical effects similar to the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the corresponding devices and equipment will not be repeated here.

[0159] The above describes specific embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0160] In the 1990s, it was very clear whether the improvement of a technology was hardware improvement (for example, improvement of the circuit structure of diodes, transistors, switches, etc.) or software improvement (improvement of the method flow). However, with the development of technology, many improvements of the method flow today can be regarded as direct improvements of the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that the improvement of a method flow cannot be implemented with a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming themselves, without having to ask chip manufacturers to design and make dedicated integrated circuit chips. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one kind of HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0161] The controller may be implemented in any suitable manner, for example, the controller may take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (e.g., software or firmware) executable by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320, and the memory controller may also be implemented as part of the control logic of the memory. It is also known to those skilled in the art that, in addition to implementing the controller in a purely computer-readable program code manner, the controller may be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, such a controller may be considered as a hardware component, and the devices for implementing various functions included therein may also be considered as structures within the hardware component. Or even, the devices for implementing various functions may be considered as both software modules for implementing the method and structures within the hardware component.

[0162] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0163] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0164] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0165] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0166] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0167] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0168] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0169] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0170] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined in this article, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0171] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0172] The present application may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0173] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A business processing method, comprising: Obtaining business demand information in natural language form sent by the user terminal; Input the business demand information into the trained large language model to obtain a request method identifier output by the large language model; the request method identifier is used to indicate an interface request method to be executed in response to the business demand information; The large language model is trained based on a business demand information sample, and the business demand information sample carries a request method identification tag; Based on the request method identifier, generating a calling instruction for calling an application interface; Based on the calling instruction, a business application corresponding to the application interface is requested to execute a business process corresponding to the business demand information.

2. The method according to claim 1, wherein inputting the business demand information into a trained large language model to obtain a request method identifier output by the large language model specifically comprises: According to the prompt word template, converting the business demand information into model input information that conforms to the prompt word template; The model input information is input into the trained large language model to obtain a request method identifier output by the large language model.

3. The method according to claim 1, before inputting the business demand information into the trained large language model and obtaining the request method identifier output by the large language model, further comprising: Obtain a sample of business demand information carrying a request method identification tag; The business demand information sample is used to perform fine-tuning training on the basic large language model to obtain the trained large language model.

4. The method according to claim 3, wherein obtaining a sample of the business demand information carrying a request method identification tag specifically comprises: Obtain historical appeal information in natural language form; The values ​​of replaceable parameters in the historical demand information are modified to obtain multiple business demand information samples.

5. The method according to claim 4, wherein the step of modifying the value of the replaceable parameter in the historical demand information to obtain multiple business demand information samples comprises: identifying replaceable parameters of a preset parameter type from the historical demand information; Replacing the parameter value of the replaceable parameter with the parameter identifier of the replaceable parameter to obtain a demand information template corresponding to the historical demand information; Obtain a list of replaceable values ​​for the replaceable parameter; The parameter identifier in the demand information template is replaced with a replaceable value in the replaceable value list to obtain a business demand information sample.

6. The method of claim 4, wherein: The replaceable parameters include at least one of a user identifier, a time range of the business to be checked, an institution identifier to be checked, and a business type to be checked.

7. The method of claim 3, wherein: The business demand information sample also carries a request parameter tag; the request parameter tag is used to indicate parameter type information of each user request parameter in at least one user request parameter carried in the business demand information sample.

8. The method according to claim 7, wherein inputting the business demand information into a trained large language model to obtain a request method identifier output by the large language model specifically comprises: Inputting the business demand information into the trained large language model to obtain the request method identifier and request parameters output by the large language model; The request parameter is used to represent the input parameter of the interface request method corresponding to the request method identifier; The generating, based on the request method identifier, a calling instruction for calling the application interface specifically includes: Based on the request method identifier and the request parameters, a calling instruction for calling an application interface is generated.

9. The method of claim 8, wherein: The request parameters include the time range of the service to be checked.

10. The method of claim 9, wherein: The format of the time range of the business to be checked output by the large language model conforms to a preset first time expression; the first time expression uses a time offset unit, an offset start time, and an offset end time to represent the time range; wherein the time offset unit includes a day, a week, a month, a quarter, or a year.

11. The method according to claim 10, wherein generating a calling instruction for calling an application interface based on the request method identifier specifically comprises: Determine the start time and end time of the service time range to be checked according to the time offset unit, the offset start time and the offset end time; A calling instruction for calling an application interface is generated based on the request method identifier, the start time, and the end time.

12. The method according to claim 1, wherein generating a calling instruction for calling an application interface based on the request method identifier specifically comprises: Querying the request service identifier corresponding to the request method identifier; Based on the request service identifier and the request method identifier, a calling instruction for calling an application interface is generated.

13. The method according to claim 12, wherein the querying of the request service identifier corresponding to the request method identifier specifically comprises: Obtaining interface metadata of at least one available application interface from a database; The interface metadata includes an available request service identifier and an available request method identifier of each available application interface in the at least one available application interface; Based on the request method identifier, a request service identifier corresponding to the request method identifier is queried from the interface metadata.

14. The method according to claim 13, wherein the querying of the request service identifier corresponding to the request method identifier from the interface metadata based on the request method identifier specifically comprises: Determine whether the request method identifier is included in the interface metadata to obtain a first determination result; If the first judgment result indicates that the request method identifier is included in the interface metadata, a request service identifier corresponding to the request method identifier is obtained from the interface metadata.

15. The method according to claim 14, wherein after the determining whether the request method identifier is included in the interface metadata and obtaining a first determination result, the method further comprises: If the first judgment result indicates that the request method identifier is not included in the interface metadata, an error prompt message is generated; The error prompt information is used to indicate that the large language model fails to determine a request method identifier that complies with regulations.

16. The method according to claim 12, wherein generating a calling instruction for calling an application interface based on the request method identifier specifically comprises: Query the signature information corresponding to the request method identifier; Based on the request service identifier, the request method identifier and the signature information, a calling instruction for calling an application interface is generated.

17. The method according to claim 1, wherein based on the calling instruction, requesting the business application corresponding to the application interface to execute the business process corresponding to the business demand information specifically comprises: Based on the calling instruction, calling the business application corresponding to the application interface, and obtaining the first business data returned by the business application; Sending the first service data to the user terminal.

18. The method according to claim 1, wherein based on the calling instruction, requesting the business application corresponding to the application interface to execute the business process corresponding to the business demand information specifically comprises: Based on the calling instruction, calling the business application corresponding to the application interface, and obtaining the second business data returned by the business application; Processing the second service data to obtain third service data; Sending the third service data to the user terminal.

19. The method according to claim 1, wherein based on the calling instruction, requesting the business application corresponding to the application interface to execute the business process corresponding to the business demand information specifically comprises: Based on the calling instruction, calling the business application corresponding to the application interface, performing operation and maintenance operations on the business database, and obtaining operation and maintenance operation results; The operation and maintenance operation result includes a first operation and maintenance operation result or a second operation and maintenance operation result; the first operation and maintenance operation result indicates that the operation and maintenance operation is successful; the second operation and maintenance operation result indicates that the operation and maintenance operation fails; Send the operation and maintenance result to the user terminal.

20. A service processing device, comprising: A business demand information acquisition module is used to acquire business demand information in natural language form sent by a user terminal; A large model module, used to input the business demand information into the trained large language model to obtain a request method identifier output by the large language model; the request method identifier is used to indicate an interface request method to be executed in response to the business demand information; The large language model is trained based on a business demand information sample, and the business demand information sample carries a request method identification tag; A calling instruction generation module, used to generate a calling instruction for calling an application interface based on the request method identifier; The business execution module is used to request the business application corresponding to the application interface to execute the business process corresponding to the business demand information based on the calling instruction.

21. A service processing device, comprising: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: Obtaining business demand information in natural language form sent by the user terminal; The business demand information is input into the trained large language model to obtain a request method identifier output by the large language model; the request method identifier is used to indicate the interface request method to be executed in response to the business demand information; the large language model is trained based on a business demand information sample, and the business demand information sample carries a request method identifier tag; Based on the request method identifier, generating a calling instruction for calling an application interface; Based on the calling instruction, a business application corresponding to the application interface is requested to execute a business process corresponding to the business demand information.

Images