CN-122022736-A - Multi-agent cooperation data processing method, device, equipment and storage medium

CN122022736ACN 122022736 ACN122022736 ACN 122022736ACN-122022736-A

Abstract

The disclosure provides a data processing method, device, equipment and storage medium for multi-agent collaboration, relates to the technical field of computers, and particularly relates to the fields of artificial intelligence, artificial Intelligence Generated Content (AIGC), self-media content creation and the like. The method comprises the steps of carrying out intent analysis conversion on input natural language demands through a main control agent to obtain a structured total task, disassembling the structured total task into a plurality of subtasks through the main control agent, distributing the subtasks to a plurality of execution agents through the main control agent, executing the corresponding subtasks through corresponding execution agents in the execution agents, and carrying out cooperative data processing between the main control agent and the execution agents through a shared message bus.

Inventors

GUO ZHUCHENG
HUANG XU
GAO XUEZHI
LI SHIWU
CAO FEIFAN
LI CE
ZHANG TINGTING

Assignees

北京百度网讯科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260209

Claims (13)

1. A data processing method for multi-agent collaboration, the multi-agent comprising a master agent and a plurality of execution agents, the method comprising: the method comprises the steps that intent analysis conversion is carried out on input natural language requirements through a main control intelligent agent, and a structured total task is obtained; the structured total task is disassembled into a plurality of subtasks through the main control intelligent agent; distributing the plurality of subtasks to the plurality of execution agents by the master agent; executing corresponding subtasks by corresponding execution agents in the plurality of execution agents; And carrying out cooperative data processing between the main control intelligent agent and the plurality of execution intelligent agents through a shared message bus.
2. The method of claim 1, wherein the converting the input natural language requirement for intent analysis by the master agent results in a structured overall task, comprising: Inquiring a workflow template knowledge graph through the main control agent; based on the semantic similarity perceived by context in the natural language requirement, a target workflow template matched with the natural language requirement is screened from the workflow template knowledge graph; generating a workflow conforming to a custom format according to the target workflow template; the workflow is taken as the structured overall task.
3. The method of claim 2, wherein the querying, by the master agent, a workflow template knowledge graph comprises: invoking a context-aware workflow engine by the master agent, wherein the context-aware workflow engine is a graph execution engine with state awareness; querying the workflow template knowledge graph through the context-aware workflow engine.
4. The method of claim 2, wherein the generating a workflow conforming to a custom format from the target workflow template comprises: Generating a workflow rule according to a workflow node list in the target workflow template, wherein the workflow node list comprises input and output of each node in the workflow, connection relation of each node in the workflow and parameter configuration of each node in the workflow; and obtaining the workflow in the custom format according to the workflow rule.
5. The method of any of claims 1-4, wherein the coordinated data processing between the master agent and the plurality of execution agents over a shared message bus comprises: Carrying out data encapsulation on the cooperatively processed data by adopting a uniform custom format to obtain encapsulated data; And transmitting the encapsulated data between the main control intelligent agent and the plurality of execution intelligent agents through a shared message bus by adopting a unified cross-mode fusion protocol.
6. The method of any of claims 1-5, further comprising: calling a dynamic scheduler perceived by a Graphic Processor (GPU) through the main control agent; and carrying out parallel scheduling decision on the coordinated data processing between the main control agent and the plurality of execution agents through the GPU-aware dynamic scheduler.
7. The method of any of claims 1-6, further comprising: Receiving user feedback through a man-machine cooperation interface in the process of carrying out the cooperative data processing; And providing the user feedback to the master agent to dynamically adjust the task allocation logic of the plurality of subtasks by the master agent.
8. The method of any of claims 1-6, further comprising: Collecting data processing results obtained after the collaborative data processing is performed through the main control intelligent agent; and automatically generating a new subtask according to the data processing result.
9. The method of any of claims 1-6, further comprising: Collecting data processing results obtained after the collaborative data processing is performed through the main control intelligent agent; according to the data processing result, automatically generating an optimization strategy for the current subtask; And updating the current subtask according to the optimization strategy to obtain an updated subtask.
10. A multi-agent collaborative data processing apparatus, the multi-agent comprising a master agent and a plurality of execution agents, the apparatus comprising: The intention analysis module is used for carrying out intention analysis conversion on the input natural language requirement through the main control intelligent agent to obtain a structured total task; the task disassembly module is used for disassembling the structured total task into a plurality of subtasks through the main control intelligent agent; the task segmentation module is used for distributing the plurality of subtasks to the plurality of execution agents through the main control agent; The task execution module is used for executing corresponding subtasks through the corresponding execution agent in the plurality of execution agents; and the task cooperative processing module is used for carrying out cooperative data processing between the main control intelligent agent and the plurality of execution intelligent agents through a shared message bus.
11. An electronic device, comprising: at least one processor, and A memory communicatively coupled to the at least one processor, wherein, The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-9.
12. A non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the method of any one of claims 1-9.
13. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any of claims 1-9.

Description

Multi-agent cooperation data processing method, device, equipment and storage medium Technical Field The present disclosure relates to the field of computer technology, and in particular, to the fields of artificial intelligence, artificial intelligence Content generation (AIGC, artificial Intelligence Generated Content), self-media Content authoring, and the like. Background Along with the rapid development of artificial intelligence technology, AIGC has shown great potential and application value in various fields. However, the existing AIGC system has at least the problem that the task execution lacks flexibility and intelligence, and is difficult to cope with complex authoring requirements, and a more flexible and intelligent solution is needed. Disclosure of Invention The disclosure provides a multi-agent collaborative data processing method, device, equipment and storage medium. According to an aspect of the present disclosure, there is provided a data processing method of multi-agent cooperation, the multi-agent including a main control agent and a plurality of execution agents, the method including: the method comprises the steps that intent analysis conversion is carried out on input natural language requirements through a main control intelligent agent, and a structured total task is obtained; the structured total task is disassembled into a plurality of subtasks through the main control intelligent agent; distributing the plurality of subtasks to the plurality of execution agents by the master agent; executing corresponding subtasks by corresponding execution agents in the plurality of execution agents; And carrying out cooperative data processing between the main control intelligent agent and the plurality of execution intelligent agents through a shared message bus. According to another aspect of the present disclosure, there is provided a data processing apparatus of multi-agent cooperation, the multi-agent including a main control agent and a plurality of execution agents, the apparatus comprising: The intention analysis module is used for carrying out intention analysis conversion on the input natural language requirement through the main control intelligent agent to obtain a structured total task; the task disassembly module is used for disassembling the structured total task into a plurality of subtasks through the main control intelligent agent; the task segmentation module is used for distributing the plurality of subtasks to the plurality of execution agents through the main control agent; The task execution module is used for executing corresponding subtasks through the corresponding execution agent in the plurality of execution agents; and the task cooperative processing module is used for carrying out cooperative data processing between the main control intelligent agent and the plurality of execution intelligent agents through a shared message bus. According to another aspect of the present disclosure, there is provided an electronic device including: at least one processor, and A memory communicatively coupled to the at least one processor, wherein, The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the methods provided by any one of the embodiments of the present disclosure. According to another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform a method provided according to any one of the embodiments of the present disclosure. According to another aspect of the present disclosure, there is provided a computer program product comprising a computer program which, when executed by a processor, implements a method provided according to any of the embodiments of the present disclosure. It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the disclosure, nor is it intended to be used to limit the scope of the disclosure. Other features of the present disclosure will become apparent from the following specification. Drawings The drawings are for a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein: FIG. 1 is a schematic diagram of a distributed cluster processing scenario according to an embodiment of the present disclosure; FIG. 2 is a flow diagram of a method of data processing for multi-agent collaboration according to an embodiment of the present disclosure; FIG. 3 is a flow diagram of another multi-agent collaborative data processing method according to an embodiment of the present disclosure; FIG. 4 is a flow diagram of another multi-agent collaborative data processing method according to an embodiment of the present disclosure; FIG. 5 is a flow diagram of another multi-agent collaborative data processing method according to an embodiment of the prese