Service request processing method and device, equipment, medium and program product

By identifying and calling target tools through large models to process banking business requests, the problem of knowledge update for customer service managers was solved, efficient and personalized customer service was achieved, and the quality of banking services was improved.

CN120639860APending Publication Date: 2025-09-12INDUSTRIAL AND COMMERCIAL BANK OF CHINA
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202510817451.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-06-18
Publication Date
2025-09-12

AI Technical Summary

Technical Problem

Bank customer service managers find it difficult to keep abreast of the latest business policies and procedures, resulting in inefficient and poor service quality. Existing intelligent customer service functions are unable to handle complex business and lack personalized customer service support.

Method used

The target tool identifier and parameters required for the business request are identified through a large model, the target server is called for processing, and the processing results are fed back. Combined with intent recognition, slot filling and sentiment analysis operations, computing resources are allocated to improve processing efficiency.

Benefits of technology

It improves the business processing capabilities of intelligent customer service, reduces the human error rate, provides professional and personalized customer service, and improves customer satisfaction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120639860A_ABST
    Figure CN120639860A_ABST
Patent Text Reader

Abstract

The invention provides a service request processing method which can be applied to the technical field of artificial intelligence. The service request processing method comprises the following steps: in response to an obtained service request of a user, inputting the service request and a plurality of callable target tool identifiers into a large model so as to identify target tool identifiers and parameters required for handling the service request; in response to the obtained large model, outputting a target tool identifier and a parameter, sending the target tool identifier and the parameter to a target server, and instructing the target server to process the target data by using the target tool and the parameter so as to output a handling result; acquiring a handling result output by the target server, and feeding back to the user according to the handling result; wherein the target server is provided with a plurality of callable target tools. The invention further provides a service request processing device and equipment, a storage medium and a program product.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, specifically to processing business requests using a large artificial intelligence model, and more specifically to a business request processing method, device, equipment, medium and program product. Background Art

[0002] In the modern banking service system, customer service managers play a vital role. They not only need to have rich professional knowledge to deal with various inquiries from customers, but also need to be familiar with the specific processes of banking business in order to provide accurate services.

[0003] However, current challenges include: Due to frequent updates to business knowledge, customer service managers struggle to stay abreast of the latest policies, regulations, and service procedures. When handling complex or unusual cases, customer service managers may lack sufficient guidance, leading to inefficient service or errors. Customer service managers face heavy workloads, making it easy for fatigue or negligence to cause a decline in service quality. There is a lack of personalized customer service support, making it impossible to provide customized solutions for diverse customer needs. Existing intelligent customer service features are relatively simple, lacking in intelligence and automation, and are unable to handle complex cases. Summary of the Invention

[0004] In view of the above problems, the present application provides a business request processing method, device, equipment, medium and program product to improve the business processing capabilities of intelligent customer service.

[0005] According to the first aspect of the present application, a business request processing method is provided, including: in response to obtaining a business request from a user, inputting the business request and multiple callable target tool identifiers into a large model to identify the target tool identifier and parameters required to process the business request; in response to obtaining the target tool identifier and parameters output by the large model, sending the target tool identifier and parameters to the target server, instructing the target server to use the target tool and parameters to process the target data to output the processing result; obtaining the processing result output by the target server, and feeding back to the user based on the processing result; wherein, the target server is provided with multiple callable target tools.

[0006] According to an embodiment of the present application, a business request and multiple callable target tool identifiers are input into a big model, including: inputting a business request, multiple callable target tool identifiers and prompt words into the big model; wherein the prompt words include at least one of a task instruction, an output requirement, and an inference example; the task instruction is used to instruct the big model to parse the intention of the business request, match the target tool identifier according to the intention, and construct parameters that meet the target tool identifier requirements; the output requirement is used to instruct the big model to output the format of the target tool identifier and parameters; the inference example is used for the big model to reference in order to output the required target tool identifier and parameters.

[0007] According to an embodiment of the present application, a business request and multiple callable target tool identifiers are input into a large model to identify the tools and parameters required to handle the business request, including at least one of the following operations: an intent recognition operation, which is used to perform natural language processing on the business request to obtain the user's intent; a slot filling operation, which is used to match tools according to intent and construct parameters that meet the tool requirements; a sentiment analysis operation, which is used to perform sentiment analysis on the business request, and the results of the sentiment analysis are used in the business handling process.

[0008] According to an embodiment of the present application, the operation to be performed is determined between at least two of the intent recognition operation, the slot filling operation, and the sentiment analysis operation based on the embedded vector of the business request, and computing resources are allocated to the operation to be performed.

[0009] According to an embodiment of the present application, the target server is instructed to use the target tool and parameters to process the target data to output the processing results, including: in response to the target tool being a query tool, analyzing the query characteristics of the business request and expressing the parameters as a dense vector or a sparse vector, and the target server uses the query tool and the dense vector or the sparse vector to process the target data to output the processing results.

[0010] According to an embodiment of the present application, the method also includes: sending permission information to the target server in response to the permission requirements corresponding to the target tool identifier; wherein the permission requirements are obtained by the large model based on the input business request and the callable target tool identifier; the permission requirements indicate the permission requirements for processing the target data using the target tool and parameters; the permission information is used for the tool to verify the permissions before processing the data to meet the permission requirements.

[0011] According to an embodiment of the present application, the method further includes: in response to the large model failing to identify the target tool identifier required to process the business request, providing feedback to the user based on the processing result of the business request by the large model.

[0012] The second aspect of the present application provides a business request processing device, including: an identification module, which is used to input the business request and the callable tool into a large model in response to obtaining the user's business request, so as to identify the tools and parameters required to handle the business request; a calling module, which is used to send the tools and parameters to the target server in response to obtaining the output tools and parameters of the large model, instructing the target server to use the tools and parameters to process the target data to output the processing results; and an output module, which is used to obtain the processing results output by the target server and provide feedback to the user based on the processing results.

[0013] The third aspect of the present application provides an electronic device, comprising: one or more processors; a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0014] The fourth aspect of the present application further provides a computer-readable storage medium having a computer program or instructions stored thereon, which implements the steps of the above method when the computer program or instructions are executed by a processor.

[0015] The fifth aspect of the present application further provides a computer program product, comprising a computer program or instructions, which implement the steps of the above method when executed by a processor.

[0016] One or more of the above embodiments have the following beneficial effects: identifying the target tool to be invoked based on a user's service request, invoking the target tool from a target server to obtain the processing result, and providing feedback to the user. If the client itself is unable to process the user's service request, the relevant tool can be invoked through the target server, thereby improving the service processing capabilities of intelligent customer service. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above contents and other objects, features and advantages of the present application will become more apparent through the following description of the embodiments of the present application with reference to the accompanying drawings, in which:

[0018] Figure 1 Schematically illustrates an application scenario diagram of a service request processing method, apparatus, device, medium, and program product according to an embodiment of the present application;

[0019] Figure 2 The following schematically shows a flow chart of a first method for processing a service request according to an embodiment of the present application;

[0020] Figure 3 The following schematically shows a flow chart of a second method for processing a service request according to an embodiment of the present application;

[0021] Figure 4 The following schematically shows a flow chart of a third method for processing a service request according to an embodiment of the present application;

[0022] Figure 5 The following schematically shows a flowchart of a fourth method for processing a service request according to an embodiment of the present application;

[0023] Figure 6 The following schematically shows a flowchart of a fifth method for processing a service request according to an embodiment of the present application;

[0024] Figure 7A block diagram schematically illustrates a structure of a service request processing device according to an embodiment of the present application; and

[0025] Figure 8 A block diagram of an electronic device suitable for implementing a service request processing method according to an embodiment of the present application is schematically shown. DETAILED DESCRIPTION

[0026] Hereinafter, embodiments of the present application will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present application. In the detailed description below, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present application. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessarily confusing the concepts of the present application.

[0027] The terms used herein are only for describing specific embodiments and are not intended to limit this application. The terms "comprise," "include," etc. used herein indicate the presence of the features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0028] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0029] When expressions such as "at least one of A, B, and C, etc." are used, they should generally be interpreted in accordance with the meaning commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0030] In the technical solution of this application, the user information involved (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0031] In the scenario of using personal information for automated decision-making, the methods, devices, and systems provided in the embodiments of the present application all provide users with corresponding operation portals for users to choose to agree or reject the automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered. The expression "automated decision-making" here refers to the activity of automatically analyzing and evaluating an individual's behavioral habits, interests and hobbies, or economic, health, credit status, etc. through computer programs and making decisions. The expression "expert decision-making" here refers to the activity of making decisions by people who specialize in a certain field, have specialized experience, knowledge and skills, and have reached a certain level of professionalism.

[0032] Existing bank customer service management systems face significant challenges in adapting to rapidly changing business environments, ensuring service quality, and meeting personalized customer needs. Therefore, the embodiments of this application provide a method for processing business requests, and the intelligent customer service manager assistant solution proposed in this application is extremely practical. It not only effectively solves existing technical problems, but also improves the work efficiency and service level of customer service managers through intelligent and automated means, ensuring that customers receive a more professional and personalized service experience. Furthermore, the system helps reduce human error rates and improve customer satisfaction. The application of this patent will bring revolutionary changes to banks' customer service models, while also driving the entire industry towards a more efficient and intelligent direction.

[0033] Figure 1 The application scenario diagram of a service request processing method, apparatus, device, medium and program product according to an embodiment of the present application is schematically shown.

[0034] like Figure 1 As shown, the application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links or optical fiber cables.

[0035] A user may use a first terminal device 101, a second terminal device 102, or a third terminal device 103 to interact with a server 105 via a network 104 to receive or send messages, etc. Various communication client applications may be installed on the first terminal device 101, the second terminal device 102, or the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (for example only).

[0036] The first terminal device 101 , the second terminal device 102 , and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0037] Server 105 can be a server that provides various services, such as a backend management server (for example only) that supports websites browsed by users using first terminal device 101, second terminal device 102, and third terminal device 103. The backend management server can analyze and process received user requests and other data, and provide feedback (e.g., web pages, information, or data obtained or generated based on user requests) to the terminal devices. For example, server 105 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud computing, network services, and middleware services.

[0038] In applications such as client applications and web applications (APPs), the client (i.e., front-end) and the server (i.e., back-end) can communicate data through network messages. For example, in the APP client, the server assembles the parameters that need to be obtained from the client and calls the network request method to send them to the server.

[0039] It should be noted that the service request processing method provided in the embodiment of the present application can generally be executed by the server 105. Accordingly, the service request processing device provided in the embodiment of the present application can generally be set in the server 105. The service request processing method provided in the embodiment of the present application can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the service request processing device provided in the embodiment of the present application can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0040] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0041] The following will be based on Figure 1 The scene described by Figures 2 to 6 The service request processing method according to the embodiment of the present application is described in detail.

[0042] Figure 2The flowchart of the service request processing method according to an embodiment of the present application is schematically shown.

[0043] like Figure 2 As shown, the business request processing method of this embodiment includes operations S210 to S230. The transaction processing method can be executed in the following manner.

[0044] In operation S210, in response to obtaining a business request from a user, the business request and multiple target tool identifiers that can be called are input into a large model to identify the target tool identifiers and parameters required to handle the business request.

[0045] In operation S220, in response to obtaining the output target tool identifier and parameters of the large model, the target tool identifier and parameters are sent to the target server, instructing the target server to use the target tool and parameters to process the target data to output a processing result. The target server is provided with multiple target tools that can be called.

[0046] In operation S230, the processing result output by the target server is obtained, and feedback is given to the user based on the processing result.

[0047] In an embodiment of the present application, before obtaining the user's information, the user's consent or authorization may be obtained. For example, before operation S220, a request to obtain the user's information may be issued to the user. If the user agrees or authorizes the acquisition of the user's information, operation S210 is performed.

[0048] In operation S210, the business request and multiple target tool identifiers that can be called are input into the big model for processing / decision-making on the user information to identify the target tool identifiers and parameters required to handle the business request. In an embodiment of the present application, a corresponding operation portal can be provided for the user to choose to agree or reject the automated decision result. That is, before the user information is input into the big model for processing / decision-making, an instruction of the user to agree or reject the processing / decision-making input through the corresponding operation portal can be obtained. If the user agrees to process / decision-making, the business request is input into the big model for processing / decision-making on the user information, that is, step S210 is executed. If the user refuses to process / decision-making, the expert decision-making process is entered.

[0049] In this embodiment, a user's service request is obtained, and the user can submit the service request via voice and / or text. Specifically, the user can make the service request to the client through a dialogue with the client. The dialogue between the user and the client can be multi-round, so as to obtain more service request input in multiple rounds. In response to the user's service request, the service request and multiple target tool identifiers that can be called are input into the macro model to identify the target tool identifiers and parameters required to process the service request.

[0050] In this embodiment, the large model can be pre-trained based on the integration of business request samples with tool and parameter labels. Specifically, historical customer inquiry data can be collected: for example, 100,000 conversations between customers and customer service managers over a period of time. Each conversation is annotated with the corresponding customer need category (e.g., inquiry, business processing, complaint, etc.) and the tool to be invoked. For example, for the business request "I want to transfer 5,000 yuan to a friend," the "transfer tool" should be invoked with the parameter "5,000 yuan"; for the request "I lost my bank card, what should I do?", the "report loss tool" should be invoked with the parameter "0"; account balance inquiries should be associated with the account information inquiry tool; loan applications should be associated with the loan qualification assessment tool and loan application submission tool. The large model can use either a deep learning-based text classification model or a traditional machine learning model (e.g., an intent classification model). The processed data is divided into training, validation, and test sets, and the model is trained using these three sets.

[0051] In this embodiment, the business request and multiple target tool identifiers that can be called are input into the big model to identify the target tool identifier and parameters required to handle the business request. For example, the business request can be a business request such as "I want to transfer 5,000 yuan to a friend" or "I lost my bank card, what should I do?" The multiple target tool identifiers that can be called can include, for example, target tools such as "financial calculator", "database query", and "customer service manager assistant". The target tool identifier can be the name of the target tool or the number of the target tool, etc. The purpose of inputting the business request and multiple target tool identifiers that can be called into the big model is to match the target tool identifier required for the user's specific business request. Depending on the complexity of the business request, the target tool identifier matched from the multiple target tool identifiers may be one or more. For example, when a user wants to report the loss of a bank card and wants to apply for a new card, he needs to call the "loss report tool" and "new card application" tools.

[0052] In operation S220, in response to obtaining the target tool identifier and parameters output by the large model, the target tool identifier and parameters are sent to the target server, instructing the target server to process the target data using the target tool and parameters to output a processing result. In this embodiment, for example, for a user business request, "I would like to apply for a 500,000 yuan personal business loan with a term of 5 years," the large model outputs the tool identifier for "loan application" and the parameters "50" (500,000 yuan) and "60" (60 months). After obtaining the target tool identifier and parameters, the tool identifier for "loan application" and the parameters "50" and "60" are sent to the target server. The target server processes the target data (the corresponding loan interest rate) using the target tool (loan application) and the parameters ("50" and "60") to output a processing result. It should be noted that the target server may require additional user information to process the target data using the target tool and parameters. For example, in the aforementioned loan application example, income and credit information may also be required. This information can be further collected during the interaction between the target server and the user.

[0053] In operation S230, the processing result outputted by the target server is obtained and the processing result is fed back to the user. In this embodiment, the client may obtain the processing result outputted from the target server and feed back the processing result to the user in the form of graphics and / or voice.

[0054] The embodiments of the present application can identify the target tool to be called based on the user's service request, call the target tool from the target server to obtain the processing result, and provide feedback to the user. If the target server does not have the processing capability to process the user's service request, it can call the relevant target tool through the target server to process the relevant data, thereby improving the service processing capabilities of intelligent customer service.

[0055] In some embodiments of the present application, a business request and a callable tool are input into a big model, including: inputting a business request and multiple callable target tool identifiers into the big model; wherein the prompt words include at least one of a task instruction, an output requirement, and an inference example; the task instruction is used to instruct the big model to parse the intention of the business request, match the target tool identifier according to the intention, and construct parameters that meet the target tool identifier requirements; the output requirement is used to instruct the big model to output the format of the target tool identifier and parameters; the inference example is used for the big model to reference in order to output the required target tool identifier and parameters.

[0056] In this embodiment, the prompt words include task instructions, output requirements, and reasoning examples, which are described in detail below.

[0057] For example, a task instruction might include the following: "Please complete the following processing flow based on the customer's business processing request: 1. Intent analysis: Analyze the business processing intent of the keyword {keywords}; 2. Tool matching: Select the most appropriate target tool from multiple available target tools {available_tools}; 3. Parameter generation: Construct a parameter structure that meets the tool's requirements; 4. Anomaly detection: Verify parameter legitimacy and permission compliance."

[0058] Output requirements can limit the specific output format requirements of the large model. For example, they can include: "Please return strictly in JSON format", and can further limit the JSON format to include specific output formats such as tool name, confidence level (used to indicate the probability that this tool should be called for the current demand), calling method, parameters, permission requirements, and alternative solutions when the main tool is unavailable.

[0059] The reasoning example may provide a specific example of the target tool identifier and parameters required for outputting the user's business request.

[0060] For example, Example 1, a typical processing scenario. A customer inputs "cross-border transfer fee calculation." The response might include: {Tool selection: {Tool name: "Financial Calculator," Confidence level: 0.95}, Call method: {Method: (omitted), Parameters: {Transfer type: cross-border, Currency pair: USD / CNY, Amount: 5,000 yuan}, Permission requirement: Basic}, Alternative solution: Call the search tool to retrieve the latest fee table}.

[0061] Example 2: Complex business calculation scenario. A customer inputs "Calculate equal principal and interest repayments for a three-year corporate loan." The returned data is as follows: {Tool selection: {Tool name: "Financial Calculator," Confidence level: 0.97}, Calling method: {Method: (omitted), Parameters: {Principal: 1,000,000, Annual interest rate: 0.045, Number of installments: 36, Repayment method: Equal principal and interest}, Permission requirement: Credit business permission}}.

[0062] Example 3, multi-condition query scenario. Customer input: "Query our company's cross-border settlement records for Q3 2023." Response: {Tool selection: {Tool name: "XX Database Query", Confidence: 0.89}, Call method: {Method: (omitted), Parameters {Statement: Query from the cross-border transaction table, company ID is parameter C, transaction date is between 2023-07-01 and 2023-09-30, parameter: {parameter C}}}, Permission requirements: [Corporate customer permissions, Financial query permissions]}}.

[0063] Example 4, knowledge consultation scenario. Customer input: "What documents are required to open a public account?" Response: {Tool selection: {Tool name: "Search Tool", Confidence: 0.92}, Call method: {Method: (omitted), Parameters: {Index: Public Account, Query: {Content: List of documents required to open a public account}}}}}}}

[0064] The above examples can be specifically input into the large model in the form of computer language. In some embodiments, a three-layer structure consisting of the above-mentioned task instructions, output requirements, and reasoning examples can be used to guide the large model's reasoning, instructing the large model to identify business requests based on the content of the prompt words, ensuring that the large model's output meets expectations, thereby improving the efficiency and quality of model reasoning.

[0065] In some embodiments of the present application, a business request and multiple callable target tool identifiers are input into a large model to identify the tools and parameters required to handle the business request, including at least one of the following operations: an intent recognition operation, which is used to perform natural language processing on the business request to obtain the user's intent; a slot filling operation, which is used to match tools according to intent and construct parameters that meet the tool requirements; a sentiment analysis operation, which is used to perform sentiment analysis on the business request, and the results of the sentiment analysis are used in the business handling process. For example, the business request and the callable tool can be input into the large model for natural language processing to obtain the user's intent, and subsequent processing is performed according to the user's intent and pre-set parameters; the business request and the callable tool can be input into the large model to directly output the tools and parameters required for the business request; or the business request and the callable tool can be input into the large model to perform sentiment analysis on the business request, and the results of the sentiment analysis are used to provide feedback to the user in combination with the handling results, or the results of the sentiment analysis are used to improve the content of the language interaction with the customer, such as soothing the customer's emotions as much as possible when it is recognized that the customer is anxious.

[0066] For example, a user's service request, "I want to complain about a 500 yuan deduction for a cross-border transfer fee on July 3rd," can be broken down into three operations: intent recognition, slot filling, and sentiment analysis. This embodiment divides large models into distinct operations and breaks down complex tasks into standardized subtasks, improving the efficiency and accuracy of service request processing through collaborative work.

[0067] In some embodiments of the present application, the operation to be performed is determined between at least two of the intent recognition operation, the slot filling operation, and the sentiment analysis operation based on the embedded vector of the business request, and computing resources are allocated to the operation to be performed.

[0068] For example, it may be that when it is determined that a slot filling operation and a sentiment analysis operation need to be performed based on the embedding vector of the business request, computing resources are allocated between the slot filling operation and the sentiment analysis operation. In this embodiment, a hybrid expert architecture including three operations, namely, intent recognition operation, slot filling operation, and sentiment analysis operation, is used. In a certain user requirement, the intent recognition operation is a lightweight classification task, which is allocated 20% of the calculation examples; the slot filling operation is a sequence labeling task that requires precise entity extraction, which is allocated 50% of the computing power. The sentiment analysis operation is a regression / classification task, which is allocated 30% of the computing power. For example, in another example, the user's business request is to understand the net value trend of a fund in order to determine whether to increase the position. In this example, the intent recognition operation is allocated 20% of the computing power; the slot filling operation is allocated 60% of the computing power; and the sentiment analysis operation is allocated 20% of the computing power.

[0069] This embodiment improves the efficiency and accuracy of multi-tasking by allocating computing resources between different operations based on the characteristics of different business requests. It is suitable for scenarios in the financial field that are intensive in professional terminology, have high compliance requirements, and are highly real-time.

[0070] In some embodiments of the present disclosure, the target server is instructed to use tools and parameters to process target data to output processing results, including: in response to the tool being a query tool, analyzing the query characteristics of the business request to represent the parameters as a dense vector or a sparse vector, and the target server uses the query tool and the dense vector or the sparse vector to process the target data to output processing results.

[0071] The vector-based retrieval model in this embodiment uses machine learning and mathematical algorithms to convert text or other forms of data into numerical vectors and then matches queries with documents by calculating the similarity between these vectors. This approach is particularly useful when processing large datasets because it can quickly locate the most relevant items without having to examine the content of each document individually.

[0072] Figure 3 The flowchart of the second service request processing method according to the embodiment of the present application is schematically shown. The following is a detailed explanation of how to use vector-based retrieval technology to match user queries with documents in the knowledge base.

[0073] like Figure 3 As shown, in response to the tool being a query tool, a dynamic routing controller is used to divide user queries into complex semantic queries and keyword-driven queries according to user queries. Complex semantic queries use a dense retrieval layer, while keyword-driven queries use a sparse retrieval layer.

[0074] The basic concepts of vector retrieval include embedding, indexing, similarity measurement, retrieval, etc.

[0075] Embedding: Unstructured data such as text and images are mapped to points or vectors in a high-dimensional space. This process is typically accomplished by deep learning models, such as pre-trained language models, which can capture the semantic information of the input data.

[0076] Indexing: To speed up the retrieval process, vectors are organized into a special data structure called an index. The index allows for quick retrieval of nearest neighbors, i.e., other vectors that are closest to a given vector in the vector space.

[0077] Similarity Measurement: Commonly used methods include cosine similarity and Euclidean distance, which are used to measure the "distance" between two vectors and then determine their degree of similarity.

[0078] Retrieval: Given a query vector, search the index for the most similar document vectors and return the corresponding documents as results.

[0079] First, the user's query is received to obtain the user's business request. Input parsing: First, the user's natural language query is received and preliminarily processed, including text preprocessing operations such as word segmentation and stop word removal. The query text is converted into a vector representation. Intent recognition is performed, using natural language understanding (NLU) technology to analyze the query intent and context to determine the user's specific needs.

[0080] Use dynamic routing controller to divide user queries into complex semantic queries and keyword-dominated queries.

[0081] 1. A distributed search engine can be used in the sparse search layer. In addition to traditional full-text search, it also supports vector fields and vector-based similarity search. This embodiment has the following advantages: (1) Distributed architecture, which can be easily expanded to multiple nodes to process massive amounts of data. (2) Good integration, easy to integrate with other tools and services, and provides a rich API and plug-in ecosystem. (3) Real-time performance, supporting real-time index updates to ensure that the latest data is always searchable.

[0082] The workflow of the sparse search layer in this embodiment is as follows: (1) Create an index, define a mapping containing vector fields, and then add documents and their corresponding vectors to the index. (2) Execute a query. After the user submits a query, the query text is first converted into a vector. A k-NN query or other suitable similarity query method is used to find the closest document vector in the index. A sorted list of results is returned to the user.

[0083] 2. For complex semantic queries in document retrieval, a dense search layer can be used to retrieve documents from the knowledge base that are most relevant to the query vector. For example, this can be achieved using deep learning-based dense search technology, whose core principle is to improve search accuracy through fine-grained semantic matching and delayed interaction. The dense search layer can use knowledge distillation to compress the original model to one-quarter its size (while retaining 95% accuracy). Vector quantization: For example, a 768-dimensional vector can be compressed to 64 bytes. Caching: A least-repeated (LRU) cache is established for high-frequency queries, returning results directly upon a hit.

[0084] The following are its core concepts.

[0085] (1) Fine-grained Token-level encoding.

[0086] Limitations of Traditional Search: Traditional dense search encodes the entire query or document into a single vector, which can lead to loss of semantic details. (For example, "cross-border transfer limit" and "international remittance limit" may be encoded as similar vectors, but the key difference between "limit" and "amount" may be blurred.)

[0087] Deep learning-based dense retrieval technology can generate independent vector representations for each token (such as a word or subword) in a text. For example, each word in a query corresponds to a vector, and each word in a document corresponds to a vector (preserving the independent semantic information of each word and capturing more subtle semantic differences).

[0088] (2) Document understanding and summarization.

[0089] Deep Reading Comprehension: For retrieved relevant documents, we use large language models to perform in-depth reading comprehension. These models can not only extract factual information from documents, but also infer implicit meaning.

[0090] Key information extraction: Based on the user's query intent, the most relevant key information is extracted from the document. This step may involve tasks such as named entity recognition (NER) and relation extraction.

[0091] Summary generation: To simplify the content of long documents, text summarization technology can be applied to generate a concise summary, ensuring that the core points are retained while reducing redundant information.

[0092] Through the dense or sparse search layer, the search results are output as candidate sets, which are then fused and re-ranked, and the final search results are generated based on the ranking. In this embodiment, a large language model is used to understand the retrieved documents and generate answers. The retrieved documents are input into the large language model as context to generate the final answer.

[0093] In this embodiment, answer generation can adopt methods such as template matching, free text generation, and multi-round dialogue management. Template matching: In some cases, the answer framework can be quickly constructed based on predefined answer templates, especially when the query is a common question. Free text generation: More commonly, a large language model is directly used to generate natural and fluent answers. This type of model is based on the Transformer architecture and can create coherent and logical responses based on understanding the context. Multi-round dialogue management: If the initial answer does not fully solve the user's problem, it has the ability to guide multiple rounds of dialogue to gradually clarify the details and provide more accurate answers.

[0094] In this embodiment, answers are optimized through quality assessment, editing and polishing, and personalization. Quality Assessment: An internal evaluation mechanism is used to check whether the generated answers meet standards such as accuracy, relevance, and grammatical correctness. Editing and Polishing: When necessary, minor adjustments are made to the automatically generated answers, such as adding links to original materials, correcting potential errors, or enhancing the professionalism of the expression. Personalization: Taking into account the different needs of different user groups, the answer style and terminology selection are appropriately adjusted to make the output more suitable for specific audiences.

[0095] In this embodiment, the output results are presented to the user. The optimized answers can be displayed in a user-friendly format, which may be plain text, tables, charts, etc., depending on the application scenario. In this embodiment, users can be encouraged to provide feedback on their satisfaction with the answers, which is crucial for subsequent improvement of the generative model.

[0096] This embodiment analyzes the query characteristics of business requests and represents parameters as dense or sparse vectors. The target server then uses query tools and these dense or sparse vectors to process the target data and output processing results. This enables efficient processing of simple queries and in-depth understanding of complex queries in financial scenarios, balancing retrieval efficiency and semantic accuracy.

[0097] In some embodiments of the present application, permission information is sent to a target server in response to a permission requirement corresponding to a target tool identifier. The permission requirement is obtained by the large model based on an input business request and an invokable target tool identifier; the permission requirement indicates the permission required to process target data using the target tool and parameters; and the permission information is used by the tool to verify permission before processing the data to ensure compliance with the permission requirement.

[0098] For example, a business transaction requires corporate financial query permission. During the transaction process, information indicating that the client has corporate financial query permission must be sent to the target server. If the client does not have financial query permission, the client's request for the corporate financial query will be rejected to protect the company's financial information.

[0099] This embodiment can automatically process user service requests that include permission requirements, preventing customers from handling services with permission requirements without relevant authorization, thereby improving customer account security.

[0100] Figure 4 The flowchart of the third service request processing method according to the embodiment of the present application is schematically shown.

[0101] like Figure 4 As shown, in this embodiment, operations S410 and S420 are included.

[0102] In operation S410, in response to obtaining a business request from a user, the business request and multiple target tool identifiers that can be called are input into a large model to identify the target tool identifiers and parameters required to handle the business request.

[0103] In operation S420, in response to the large model failing to identify the target tool identifier required to process the business request, feedback is provided to the user based on the large model's processing results for the business request. For example, the large model fails to identify the target tool identifier required to process the business request when the user does not clearly express their specific business request, the business request is unclear, the user is only engaged in communication unrelated to the actual business, or when the large model can directly provide a clear result. In this case, feedback is provided to the user based on the large model's processing results for the business request. This embodiment can optimize the feedback effect when no tool is output, thereby improving the user experience.

[0104] Figure 5 The flowchart schematically shows a fourth service request processing method according to an embodiment of the present application.

[0105] like Figure 5 As shown, in this embodiment, operations S510 to S540 are included. In this embodiment, S510 to S530 can refer to the introduction of the relevant steps of S210 to S230 above. It should be noted that in this embodiment, in the S540 operation, the processing results are input into the big model, feedback results are obtained, and the feedback results are fed back to the user. For example, based on the customer's historical transaction records, preference settings and market trend analysis, a personalized service recommendation mechanism is constructed. Using data analysis and predictive modeling technology, customer needs are identified in advance and corresponding solutions are prepared. This embodiment can perform personalized processing based on the customer's past preferences, usage habits, etc. to enhance the user experience.

[0106] In some embodiments of the present application, feedback can be collected from each service interaction to continuously improve its own performance. Both online update and offline training modes are supported to ensure that a certain learning ability can be maintained even in network-constrained situations.

[0107] Figure 6The flowchart schematically shows a fifth service request processing method according to an embodiment of the present application.

[0108] When the client starts the application, it first performs initialization, including loading the latest knowledge base content, checking the current status, and establishing necessary network connections. Figure 6 As shown, in this embodiment, the user sends a message to the client, and the client receives the user's service request by receiving the message. The user can ask questions or request help from the client via voice, text, or other means. When it comes to specific service processing, the process navigator intervenes and provides step-by-step operation instructions to ensure that each step complies with established business rules and process requirements.

[0109] In response to receiving a user's business request, the client can analyze the user's intent using the intelligent question-and-answer module. Based on the analysis results, the client inputs the business request and multiple target tool identifiers into the large model. Specifically, this message containing the business request and multiple target tool identifiers is sent to the large language model (LLM). The LLM identifies the target tool identifier and parameters required to process the business request and sends the required tool identifier or parameters to the client. In response to receiving the target tool identifier and parameters output by the large model, the client sends a request for the target tool identifier and parameters to the MCP server (target server), instructing the MCP server to process the target data using the target tool and parameters to output a processing result. During this process, the MCP server can obtain the required data from the data source. The data source returns the data to the MCP server. The MCP server processes the target data using the target tool and parameters. After processing, the MCP server returns the processing result to the client. The client inputs the processing result into the large language model, which then outputs a final feedback result. The client displays the final result to the user in a user-friendly format based on the final feedback result.

[0110] In this embodiment, the MCP server can also return a tool identifier to the client, enabling the client to provide the large language model with a callable tool identifier or data. Alternatively, the MCP server can return the callable tool identifier to the client before receiving a user's service request, allowing the client and the large language model to obtain the latest callable tool.

[0111] In this embodiment, a personalized service engine may be provided to timely recommend products or services that may be of interest to the customer based on their specific circumstances, thereby increasing sales opportunities and improving customer satisfaction.

[0112] In this embodiment, after each service is completed, users can be invited to evaluate the service effect and any useful feedback can be collected. These feedbacks will be incorporated into the continuous learning mechanism as the basis for subsequent improvements.

[0113] In this embodiment, after completing all tasks, the client closes the current session and saves all important interaction records for future review or auditing, thereby facilitating tracing and improving the quality and efficiency of customer service.

[0114] Implementing the service request processing method of this embodiment is expected to achieve the following technical effects: Improved work efficiency: Through automated and intelligent support, customer service managers spend less time searching for information and confirming processes, thereby improving overall work efficiency. Reduced human error: With the help of a knowledge base and process guidance, the risk of human error is reduced and service accuracy is improved.

[0115] Based on the above service request processing method, this application also provides a service request processing device. Figure 7 The device is described in detail.

[0116] Figure 7 The structural block diagram of the service request processing device according to an embodiment of the present application is schematically shown.

[0117] like Figure 7 As shown, the service request processing device 700 of this embodiment includes an identification module 710 , a calling module 720 and an output module 730 .

[0118] The identification module 710 is used to input the business request and the callable tools into the big model in response to obtaining the business request from the user, so as to identify the tools and parameters required to handle the business request.

[0119] In one embodiment, the identification module 710 inputs a business request and multiple callable target tool identifiers into the big model, including: inputting the business request, multiple callable target tool identifiers, and prompt words into the big model; wherein the prompt words include at least one of a task instruction, an output requirement, and an inference example; the task instruction is used to instruct the big model to parse the intent of the business request, match the target tool identifier based on the intent, and construct parameters that meet the target tool identifier requirements; the output requirement is used to instruct the big model to output the target tool identifier and parameter format; and the inference example is used for the big model to refer to in order to output the required target tool identifier and parameters. The big model is trained by fusing business request samples with tool and data labels.

[0120] In one embodiment, the identification module 710 inputs the business request and multiple callable target tool identifiers into the big model to identify the tools and parameters required to handle the business request, including at least one of the following operations: intent recognition operation, used to perform natural language processing on the business request to obtain the user's intent; slot filling operation, used to match tools according to intent and construct parameters that meet the tool requirements; sentiment analysis operation, used to perform sentiment analysis on the business request, and the results of the sentiment analysis are used in the business handling process.

[0121] In one embodiment, the recognition module 710 determines the operation to be performed among at least two of the intent recognition operation, the slot filling operation, and the sentiment analysis operation based on the embedded vector of the business request, and allocates computing resources for the operation to be performed.

[0122] The calling module 720 is used to respond to the acquisition of the target tool identifier and parameters of the large model output, send the target tool identifier and parameters to the target server, and instruct the target server to use the target tool and parameters to process the target data to output the processing result. The target server is provided with multiple target tools that can be called.

[0123] In one embodiment, the calling module 720 responds to the target tool being a query tool, analyzes the query characteristics of the business request and represents the parameters as a dense vector or a sparse vector. The target server uses the query tool and the dense vector or the sparse vector to process the target data to output the processing result. In this embodiment, a dynamic knowledge base can be set up, and the query tool can retrieve relevant knowledge content from the dynamic knowledge base. The dynamic knowledge base can contain data such as the bank's internal policies, regulations, product information and business processes. Real-time synchronization with other business systems and related databases through the API interface ensures the timeliness and accuracy of the information. Knowledge distillation technology is used to perform lightweight semantic vector compression on the documents in the knowledge base to reduce storage and retrieval delays while maintaining a semantic retention rate of more than 90%.

[0124] In one embodiment, permission information is sent to the target server in response to the permission requirement corresponding to the target tool identifier; wherein the permission requirement is obtained by the large model based on the input business request and the callable target tool identifier; the permission requirement indicates the permission requirement for processing the target data using the target tool and parameters; the permission information is used for the tool to verify the permission before processing the data to meet the permission requirements.

[0125] Output module 730 is used to obtain the processing results output by the target server and provide feedback to the user based on the processing results. In one embodiment, in response to the macro model failing to identify the target tool identifier required to process the business request, output module 730 may provide feedback to the user based on the macro model's processing results for the business request. In one embodiment, output module 730 may input the processing results into the macro model, obtain feedback results, and provide feedback results to the user.

[0126] According to embodiments of the present application, any multiple modules among the identification module 710, the call module 720, and the output module 730 may be combined into a single module, or any one of these modules may be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules may be combined with at least part of the functionality of other modules and implemented in a single module. According to embodiments of the present application, at least one of the identification module 710, the call module 720, and the output module 730 may be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or may be implemented in hardware or firmware through any other reasonable means of circuit integration or packaging, or may be implemented in any one of the three implementation methods of software, hardware, and firmware, or any appropriate combination of any of these. Alternatively, at least one of the identification module 710, the call module 720, and the output module 730 may be at least partially implemented as a computer program module that, when executed, performs the corresponding functionality.

[0127] In some embodiments of the present application, an integrated development environment (IDE) may be used: providing developers with a platform for extending or customizing the functionality of the intelligent customer service manager assistant. A plug-in architecture is supported to facilitate the rapid addition of new service features or integration with third-party applications.

[0128] Figure 8 A block diagram of an electronic device suitable for implementing a service request processing method according to an embodiment of the present application is schematically shown.

[0129] like Figure 8As shown, an electronic device 800 according to an embodiment of the present application includes a processor 801, which can perform various appropriate actions and processes based on a program stored in a read-only memory (ROM) 802 or a program loaded from a storage unit 808 into a random access memory (RAM) 803. The processor 801 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a dedicated microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 801 may also include onboard memory for caching purposes. The processor 801 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present application.

[0130] Various programs and data required for the operation of the electronic device 800 are stored in the RAM 803. The processor 801, ROM 802, and RAM 803 are connected to each other via a bus 804. The processor 801 performs various operations of the method flow according to the embodiment of the present application by executing the programs in the ROM 802 and / or RAM 803. It should be noted that the programs may also be stored in one or more memories other than the ROM 802 and the RAM 803. The processor 801 may also perform various operations of the method flow according to the embodiment of the present application by executing the programs stored in the one or more memories.

[0131] According to an embodiment of the present application, electronic device 800 may further include an input / output (I / O) target 805, which is also connected to bus 804. Electronic device 800 may also include one or more of the following components connected to I / O target 805: an input section 806 including a keyboard, mouse, etc.; an output section 807 including devices such as a cathode ray tube (CRT), liquid crystal display (LCD), and speakers; a storage section 808 including a hard disk; and a communication section 809 including a network target card such as a LAN card or modem. Communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to I / O target 805 as needed. Removable media 811, such as a magnetic disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed in drive 810 as needed, so that computer programs read from the removable media can be installed into storage section 808 as needed.

[0132] This application also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments, or may exist independently and not be incorporated into the device / apparatus / system. The computer-readable storage medium carries one or more programs, and when the one or more programs are executed, the method according to the embodiments of this application is implemented.

[0133] According to an embodiment of the present application, a computer-readable storage medium may be a non-volatile computer-readable storage medium, and may include, for example, but not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present application, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present application, a computer-readable storage medium may include the ROM 802 and / or RAM 803 described above and / or one or more memories other than ROM 802 and RAM 803.

[0134] The embodiments of the present application also include a computer program product, which includes a computer program containing program code for executing the method shown in the flowchart. When the computer program product is executed in a computer system, the program code is used to enable the computer system to implement the service request processing method provided in the embodiments of the present application.

[0135] The computer program executes the above functions defined in the system / device of the embodiment of the present application when the processor 801 executes the computer program. According to the embodiment of the present application, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0136] In one embodiment, the computer program may be stored on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may be transmitted and distributed in the form of a signal on a network medium, downloaded and installed via the communication portion 809, and / or installed from a removable medium 811. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to wireless, wired, or any suitable combination thereof.

[0137] In such an embodiment, the computer program can be downloaded and installed from the network via the communication section 809, and / or installed from the removable medium 811. When the computer program is executed by the processor 801, the above-mentioned functions defined in the system of the embodiment of the present application are performed. According to the embodiment of the present application, the systems, devices, means, modules, units, etc. described above can be implemented by computer program modules.

[0138] According to an embodiment of the present application, the program code for executing the computer program provided by the embodiment of the present application can be written in any combination of one or more programming languages. Specifically, these computer programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C" or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect via the Internet).

[0139] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the above-mentioned module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of the boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0140] Those skilled in the art will appreciate that the features described in the various embodiments of this application may be combined and / or coupled in various ways, even if such combinations or couplings are not explicitly described in this application. In particular, the features described in the various embodiments of this application may be combined and / or coupled in various ways without departing from the spirit and teachings of this application. All such combinations and / or couplings fall within the scope of this application.

Claims

1. A method for processing a business request, characterized in that: include: In response to obtaining a user's business request, the business request and multiple target tool identifiers that can be called are input into the big model to identify the target tool identifiers and parameters required to handle the business request; In response to obtaining the target tool identifier and parameters output by the large model, the target tool identifier and parameters are sent to a target server, instructing the target server to process the target data using the target tool and parameters to output a processing result; Obtaining the processing result output by the target server, and providing feedback to the user based on the processing result; The target server is provided with the plurality of callable target tools.

2. The method according to claim 1, characterized in that The step of inputting the business request and multiple callable target tool identifiers into the big model includes: Inputting the business request, multiple callable target tool identifiers and prompt words into the macro model; The prompt words include at least one of task instructions, output requirements, and reasoning examples; The task instruction is used to instruct the large model to parse the intent of the business request, match the target tool identification according to the intent, and construct parameters that meet the requirements of the target tool identification; The output requirement is used to instruct the large model to output the target tool identifier and parameter format; The inference instance is used for the large model reference to output the required target tool identification and parameters.

3. The method according to claim 1, characterized in that Inputting the business request and multiple callable target tool identifiers into the big model to identify the tools and parameters required to handle the business request includes at least one of the following operations: Intent recognition operation, used to perform natural language processing on the service request to obtain the user's intent; A slot filling operation is used to match the tool according to the intent and construct parameters that meet the tool requirements; The sentiment analysis operation is used to perform sentiment analysis on the business request, and the result of the sentiment analysis is used in the business handling process.

4. The method according to claim 3, characterized in that Determine the operation that needs to be performed among at least two of the intent recognition operation, the slot filling operation, and the sentiment analysis operation according to the embedded vector of the business request, and allocate computing resources for the operation that needs to be performed.

5. The method according to claim 1, characterized in that The instructing the target server to use the target tool and parameters to process the target data to output a processing result includes: In response to the target tool being a query tool, the query characteristics of the business request are analyzed to represent the parameters as a dense vector or a sparse vector. The target server uses the query tool and the dense vector or the sparse vector to process the target data to output a processing result.

6. The method according to claim 1, characterized in that Also includes: In response to the target tool identifying a corresponding permission requirement, sending permission information to the target server; Among them, the permission requirements are obtained by the big model based on the input business request and the callable target tool identification; the permission requirements indicate the permission requirements for processing the target data using the target tool and parameters; the permission information is used by the tool to verify the permissions before processing the data to meet the permission requirements.

7. The method according to claim 1, characterized in that Also includes: In response to the large model failing to identify the target tool identifier required to process the business request, feedback is provided to the user based on the result of the large model's processing of the business request.

8. The method according to claim 1, characterized in that Providing feedback to the user based on the processing result includes: The processing result is input into the big model to obtain a feedback result, and the feedback result is fed back to the user.

9. A service request processing device, characterized in that: include: an identification module, configured to, in response to obtaining a user's service request, input the service request and the callable tools into the macro model to identify the tools and parameters required to process the service request; a calling module, configured to, in response to obtaining the tool and parameters outputted by the large model, send the tool and parameters to a target server, and instruct the target server to process target data using the tool and parameters to output a processing result; as well as An output module, configured to obtain the processing result output by the target server and provide feedback to the user based on the processing result; The target server is provided with the plurality of callable target tools.

10. An electronic device comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 8.

11. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

12. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Citation Information

Cited By

  • MCP function calling method and device, equipment, storage medium and computer program product

    CN121277648A

  • MCP calling method and device, equipment, storage medium and computer program product

    CN121277649A

  • Electric power market power calculation method based on artificial intelligence

    CN121684971A