Search

CN-120653238-B - Large-model intelligent agent dynamic tool arrangement method, system, medium and equipment

CN120653238BCN 120653238 BCN120653238 BCN 120653238BCN-120653238-B

Abstract

A method, a system, a medium and equipment for arranging dynamic tools of large-model agents relate to the technical field of large models. The method comprises the steps of obtaining interface metadata of an open platform, generating standardized large model tool description based on the interface metadata, registering the standardized large model tool description to a tool library, receiving a query request of a user, determining a target tool based on semantic similarity between the query request and each tool description in the tool library, mapping service parameters in the query request to interface call parameters of the target tool, executing interface call corresponding to the target tool based on the interface call parameters, preprocessing a call result, and displaying the preprocessed call result in a visual mode. By implementing the technical scheme provided by the application, the dynamic arrangement of the management tool can be realized, the adaptability of the system to the service scene change is improved, and the management efficiency of enterprises is further improved.

Inventors

  • GAO HAIFENG

Assignees

  • 苏州盖雅信息技术有限公司

Dates

Publication Date
20260508
Application Date
20250526

Claims (8)

  1. 1. A method for orchestrating dynamic tools of large model agents, the method comprising: acquiring interface metadata of an open platform, generating standardized large model tool description based on the interface metadata, and registering the standardized large model tool description to a tool library; receiving a query request of a user, and determining a target tool based on semantic similarity between the query request and each tool description in the tool library; mapping the service parameters in the query request into interface calling parameters of the target tool; executing the interface call corresponding to the target tool based on the interface call parameter, and preprocessing the call result; displaying the preprocessed calling result in a visual form; wherein the mapping the service parameters in the query request to interface call parameters of the target tool includes: Carrying out semantic analysis on a query request by utilizing a large model, firstly identifying business entities in a text through a named entity identification model, then determining the relation between the entities by using dependency syntactic analysis, extracting corresponding modifier words and qualifier words, determining the association relation between the entities and query operation, organizing the identified entities and the relation thereof into a structured parameter list, wherein each parameter item comprises a parameter value and a parameter type label; Carrying out semantic matching on a parameter list to be mapped based on the parameter definition of the target tool to obtain a parameter mapping relation; Converting the parameter list to be mapped into a target format parameter according to the parameter mapping relation and a preset mapping rule; Checking the data type check and the filling item of the target format parameter, and generating a final interface calling parameter after the missing parameter is automatically complemented; the executing the interface call corresponding to the target tool based on the interface call parameter includes: Generating a call request based on the call address of the target tool and the interface call parameter; obtaining a response result of the target tool to the call request, and recording performance indexes of a call process; analyzing the service data in the response result, and performing format conversion according to a preset data processing rule; and integrating the converted service data with the performance index to generate a standardized calling result.
  2. 2. The large model agent dynamic tool orchestration method according to claim 1, wherein the generating and registering standardized large model tool descriptions to a tool library based on the interface metadata comprises: analyzing the interface description in the interface metadata to generate tool information containing a tool name and a function description; Converting the request parameters in the interface metadata into standardized parameter definitions; combining the tool information and the parameter definition into a tool description in a JSON format, and registering the tool description to a tool library.
  3. 3. The large model agent dynamic tool orchestration method according to claim 1, wherein the determining a target tool based on semantic similarity of the query request and tool descriptions in the tool library comprises: inputting the query request into a large model to extract intention information and generate a corresponding query vector; calculating the vector similarity between the query vector and each tool description in the tool library, and selecting tools with similarity larger than a preset threshold as candidate tools; and calculating the availability score of each candidate tool based on the history call record, and selecting the tool with the highest score as a target tool.
  4. 4. The method for orchestrating dynamic tools of large model agents according to claim 1, wherein the preprocessing of the call results comprises: performing anomaly detection on the calling result to generate an anomaly detection report; judging whether retries are needed according to the abnormality detection report; if retrying is needed, returning to the step of executing the interface call corresponding to the target tool based on the interface call parameter; If retrying is not needed, filtering and completing the data of the calling result passing through the anomaly detection based on a preset data cleaning rule to obtain cleaned data; And performing index calculation and structuring treatment on the cleaned data to obtain a preprocessed calling result.
  5. 5. The method of claim 4, wherein performing anomaly detection on the call result to generate an anomaly detection report, comprising: Detecting a response status code, data integrity and data format of the calling result, and generating a corresponding basic abnormal identifier; verifying the validity of key service indexes in the calling result according to a preset service rule to generate a service abnormality identifier; analyzing the basic anomaly identification and the business anomaly identification based on a preset anomaly rule base, and determining anomaly types and anomaly levels; and integrating the anomaly type, the anomaly level and the corresponding processing suggestions to generate an anomaly detection report.
  6. 6. A large model agent dynamic tool orchestration system for performing the large model agent dynamic tool orchestration method according to claim 1, the system comprising: The tool library generation module is used for acquiring the interface metadata of the open platform, generating standardized large model tool description based on the interface metadata and registering the standardized large model tool description to the tool library; The target tool determining module is used for receiving a query request of a user and determining a target tool based on semantic similarity between the query request and each tool description in the tool library; An interface parameter determining module, configured to map a service parameter in the query request to an interface call parameter of the target tool; The interface calling execution module is used for executing the interface calling corresponding to the target tool based on the interface calling parameters and preprocessing the calling result; and the calling result display module is used for displaying the preprocessed calling result in a visual form.
  7. 7. A computer readable storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the method according to any one of claims 1-5.
  8. 8. An electronic device comprising a processor, a memory, a user interface, and a network interface, the memory to store instructions, the user interface and the network interface to communicate to other devices, the processor to execute the instructions stored in the memory to cause the electronic device to perform the method of any of claims 1-5.

Description

Large-model intelligent agent dynamic tool arrangement method, system, medium and equipment Technical Field The application relates to the technical field of large models, in particular to a method, a system, a medium and equipment for arranging dynamic tools of large-model intelligent bodies. Background With the rapid development of artificial intelligence technology, large-model intelligent agents are increasingly widely applied in the field of enterprise management. In a labor management scenario, an enterprise typically needs to invoke interfaces of multiple open platforms to accomplish tasks such as personnel scheduling, attendance management, performance assessment, and the like. To improve management efficiency, enterprise administrators often describe specific business needs in natural language, and it is desirable for the system to automatically invoke corresponding management tools to handle these needs. Currently, a mainstream enterprise management system generally adopts a preset tool call flow, and the system binds user requirements with a specific management tool according to a preset rule. This approach requires enterprise administrators to be familiar with the functions and methods of use of the various tools, and to reconfigure manually when management tools are added or replaced. Because the call flow is fixed, the system can not flexibly select and combine proper tools according to the change of the actual service scene, which results in low use efficiency of the system and influences the management efficiency of enterprises. Disclosure of Invention The application provides a large-model intelligent body dynamic tool arrangement method, a system, a medium and equipment, which can realize the dynamic arrangement of management tools, improve the adaptability of the system to service scene changes and further improve the management efficiency of enterprises. In a first aspect, the present application provides a method for orchestrating dynamic tools of large model agents, the method comprising: acquiring interface metadata of an open platform, generating standardized large model tool description based on the interface metadata, and registering the standardized large model tool description to a tool library; receiving a query request of a user, and determining a target tool based on semantic similarity between the query request and each tool description in the tool library; mapping the service parameters in the query request into interface calling parameters of the target tool; executing the interface call corresponding to the target tool based on the interface call parameter, and preprocessing the call result; And displaying the preprocessed calling result in a visual form. By adopting the technical scheme, the system can accurately understand the user requirements and automatically complete tool call by acquiring the interface metadata of the open platform and generating the standardized tool description registration to the tool library, when the query request of the user is received, the target tool is determined based on the semantic similarity of the query request and the tool descriptions in the tool library, the manual pre-configuration of the call rule is not needed, the flexibility of tool selection is improved, the service parameters in the query request are mapped to the interface call parameters of the target tool, the interface call and the result preprocessing are executed, and finally the user can conveniently know the processing result by visually displaying the call result. The dynamic arrangement of the management tool is realized, the adaptability of the system to service scene changes is improved, the manual configuration cost is reduced, and the management efficiency of enterprises is further improved. In a second aspect of the application, there is provided a large model agent dynamic tool orchestration system, the system comprising: The tool library generation module is used for acquiring the interface metadata of the open platform, generating standardized large model tool description based on the interface metadata and registering the standardized large model tool description to the tool library; The target tool determining module is used for receiving a query request of a user and determining a target tool based on semantic similarity between the query request and each tool description in the tool library; An interface parameter determining module, configured to map a service parameter in the query request to an interface call parameter of the target tool; The interface calling execution module is used for executing the interface calling corresponding to the target tool based on the interface calling parameters and preprocessing the calling result; and the calling result display module is used for displaying the preprocessed calling result in a visual form. In a third aspect the application provides a computer storage medium having stored thereon a plurality of instru