Search

CN-122019640-A - Report generation method, system, equipment and storage medium based on multiple agents

CN122019640ACN 122019640 ACN122019640 ACN 122019640ACN-122019640-A

Abstract

The embodiment of the invention discloses a report generation method, a system, equipment and a storage medium based on multiple agents. The method comprises the steps of receiving a research task, configuring at least one agent corresponding to the research task, driving the agent to execute the research task, carrying out online real-time calibration on at least one configuration parameter of the agent according to a collaboration log and evidence pool content generated by the agent to execute the research task, and returning to execute the step of driving the agent to execute the research task, wherein the configuration parameter comprises at least one of a prompt template, tool permission, evidence weight and output structure constraint, and generating and displaying a research report generated by the agent to execute the research task. By means of the method, the configuration parameters of the intelligent agents are calibrated in real time on line, so that the real-time self-adaptive tuning of the role parameters of the intelligent agents is realized, and the configuration parameters of the intelligent agents can be timely adjusted and corrected.

Inventors

  • LIU LIZHAO

Assignees

  • 深圳市海云天科技股份有限公司

Dates

Publication Date
20260512
Application Date
20260116

Claims (10)

  1. 1. A multi-agent based report generation method, the method comprising: receiving a research mission; configuring at least one agent corresponding to the research mission; Driving the agent to execute the research task; Performing online real-time calibration on at least one configuration parameter of the intelligent agent according to the collaboration log and evidence pool content generated by the intelligent agent for executing the research task, and returning to the step of executing the driving intelligent agent for executing the research task, wherein the configuration parameter comprises at least one of a prompt template, tool permission, evidence weight and output structure constraint; and generating and displaying a research report generated by the agent executing the research task.
  2. 2. The method of claim 1, wherein the online real-time calibration of at least one configuration parameter of the agent based on collaboration log and evidence pool content generated by the agent performing the research mission comprises: calculating a calibration index according to a collaboration log and evidence pool content generated by the agent executing the research task; And when the calibration index meets a preset triggering condition, carrying out online real-time calibration on at least one configuration parameter of the intelligent agent.
  3. 3. The method of claim 2, wherein the calibration index comprises at least one of a degree of compliance, a frequency of conflict, and a severity of conflict.
  4. 4. The method of claim 1, wherein prior to said driving said agent to perform said research mission, said method further comprises: generating an execution constraint for executing the research task for each agent, wherein the execution constraint comprises subtasks corresponding to each agent, and execution sequence, dependency relationship and ending condition of each subtask; after the driving the agent to perform the research mission, the method further comprises: pushing the whole process event stream of the execution process of the research task to a front-end visual interface, and receiving an intervention instruction input by a user in the process of pushing the whole process event stream; Analyzing the intervention instruction into task arrangement constraint and/or role modification request; And/or calibrating configuration parameters of the corresponding agent on line based on the role modification request.
  5. 5. The method of claim 1, wherein said configuring at least one agent corresponding to said research mission comprises: receiving character intention information input by a user; At least one agent configuration is generated from the character intent information, the agent configuration including a configuration of at least one of a character name, a role description, a background story, a hint template parameter, a tool right, an evidence weight, an output structure constraint.
  6. 6. The method of claim 1, further comprising, upon detecting that the perspectives of different ones of the agents on the same problem generation conflict with each other, performing the following: Triggering conflict views to perform multiple rounds of dialectics on the conflict views by corresponding agents to obtain conflict processing results, or Calculating the weight of the evidence corresponding to each conflict viewpoint, selecting the evidence corresponding viewpoint of the weight meeting the preset condition as the adoption viewpoint to obtain a conflict processing result, wherein the weight is calculated at least based on one of the timeliness of the evidence, the source credibility of the evidence and the demonstration integrity of the evidence, or Triggering the conflict viewpoint corresponding agent to supplement evidence aiming at the conflict viewpoint to obtain a conflict processing result; And sending the conflict processing result to the front end in the form of an event message, and writing arbitration reasons, adopted views, overrule views and reference evidence numbers corresponding to the conflict processing result into the evidence pool.
  7. 7. A multi-agent based report generating system, the system comprising: The receiving module is used for receiving the research task; the agent configuration module is used for configuring at least one agent corresponding to the research task; The agent cooperation engine is used for driving the agent to execute the research task; The role configuration calibration module is used for carrying out online real-time calibration on at least one configuration parameter of the intelligent agent according to the collaboration log and the evidence pool content generated by the intelligent agent executing the research task, wherein the configuration parameter comprises at least one of a prompt template, tool permission, evidence weight and output structure constraint; and the report generation and display module is used for generating and displaying a research report generated by the agent executing the research task.
  8. 8. The system of claim 7, wherein the role configuration calibration module is further configured to calculate a calibration index according to a collaboration log and evidence pool content generated by the agent executing the research mission, and perform online calibration on configuration parameters of the agent when the calibration index meets a preset trigger condition.
  9. 9. A multi-agent based report generating device comprising a processor and a memory coupled to each other, the memory having stored therein a computer program for executing the computer program to implement the steps of the method of any of claims 1-6.
  10. 10. A computer readable storage medium, characterized in that the computer readable storage medium stores program data, which when executed by a processor, implements the steps of the method according to any of claims 1-6.

Description

Report generation method, system, equipment and storage medium based on multiple agents Technical Field The invention relates to the technical field of intelligent collaborative decision support, in particular to a report generation method, a system, equipment and a storage medium based on multiple agents. Background With the development of artificial intelligence large model technology, tools for automatically generating research reports are continuously emerging. The prior art is broadly classified into the following categories: The system takes a single large model as a core, generates a report around a preset template, lacks a role division and cooperation mechanism, is difficult to cover multiple professional views, connects different subtasks in series through a static pipeline, has limited information exchange among the subtasks, cannot dynamically adjust the tasks and roles according to discussion progress, and has the advantages that a plurality of intelligent agents are introduced into part of the system, but the cooperation process is invisible to a user, the user only can see a final result, a conclusion forming path is difficult to understand, and the correction is difficult to timely correct. Furthermore, existing solutions do not enable real-time calibration of agent configuration parameters. Disclosure of Invention The invention aims to provide a report generation method, a system, equipment and a storage medium based on multiple agents, which aim to solve the problem that the configuration parameters of the agents cannot be calibrated in real time in the prior art. In order to solve the technical problems, the first aspect of the application provides a report generation method based on multiple agents, which comprises the steps of receiving a research task, configuring at least one agent corresponding to the research task, driving the agent to execute the research task, carrying out online real-time calibration on at least one configuration parameter of the agent according to a collaboration log and evidence pool content generated by the agent to execute the research task, and returning to execute the step of driving the agent to execute the research task, wherein the configuration parameter comprises at least one of a prompt template, tool authority, evidence weight and output structure constraint, and generating and displaying a research report generated by the agent to execute the research task. In order to solve the technical problems, the second aspect of the application provides a report generating system based on multiple agents, which comprises a receiving module, an agent configuration module, an agent cooperation engine, a role configuration calibration module and a role configuration calibration module, wherein the receiving module is used for receiving a research task, the agent configuration module is used for configuring at least one agent corresponding to the research task, the agent cooperation engine is used for driving the agent to execute the research task, the role configuration calibration module is used for carrying out online real-time calibration on at least one configuration parameter of the agent according to a cooperation log and evidence pool content generated by the agent to execute the research task, and the configuration parameter comprises at least one of a prompt template, tool rights, evidence weights and output structure constraints, and the report generation and display module is used for generating and displaying a research report generated by the agent to execute the research task. To solve the above technical problem, a third aspect of the present application provides a multi-agent based report generating device, which includes a processor and a memory coupled to each other, the memory storing a computer program therein, the processor being configured to execute the computer program to implement the steps of the method as provided in the first aspect. To solve the above technical problem, a fourth aspect of the present application provides a computer-readable storage medium storing program data which, when executed by a processor, implements the steps of the method as provided in the first aspect. The embodiment of the application has the advantages that the method is different from the situation in the prior art, the method drives the agent to execute the research task by configuring at least one agent corresponding to the research task, carries out online real-time calibration on at least one configuration parameter of the agent according to the collaboration log and the evidence pool content generated by the agent to execute the research task, and returns to execute and drive the agent to execute the research task, wherein the configuration parameter comprises at least one of a prompt template, tool authority, evidence weight and output structure constraint, and generates and displays a research report generated by the agent to execute the research