CN-122019712-A - Multi-agent cooperative scientific hypothesis automatic generation method

CN122019712ACN 122019712 ACN122019712 ACN 122019712ACN-122019712-A

Abstract

The application discloses a multi-agent collaborative scientific hypothesis automatic generation method, which belongs to the field of creative scientific research engineering and comprises the steps of inputting a to-be-researched problem and research background information of the to-be-researched problem into creative generation agents to obtain a plurality of creative hypothesis texts of the to-be-researched problem, respectively inputting the plurality of creative hypothesis texts into hypothesis test agents to obtain evaluation data of each creative hypothesis text under a plurality of evaluation dimensions, and determining target creative hypothesis texts corresponding to the to-be-researched problem from all creative hypothesis texts according to the evaluation data of each creative hypothesis text under the plurality of evaluation dimensions. According to the application, through the cooperative work of creative generation agent and hypothesis test agent, the problems of insufficient innovation, one-sided evaluation result and incomplete reasoning chain in the automatic hypothesis process are solved, and the overall working efficiency and decision quality of scientific researchers in complex scientific research tasks are improved.

Inventors

LI YUN
YANG GUOLI
ZHENG QIBIN
YUE AIZHEN
HAN HONGWEI

Assignees

北京大数据先进技术研究院

Dates

Publication Date: 20260512
Application Date: 20260121

Claims (10)

1. The method for automatically generating the scientific hypothesis of multi-agent cooperation is characterized by comprising the following steps of: Inputting a to-be-researched problem and research background information of the to-be-researched problem into a creative generating agent to obtain a plurality of creative hypothesis texts of the to-be-researched problem; inputting a plurality of creative hypothesis texts into hypothesis testing agents respectively to obtain evaluation data of each creative hypothesis text under a plurality of evaluation dimensions respectively; And determining target creative hypothesis texts corresponding to the problems to be researched from all the creative hypothesis texts according to the evaluation data of each creative hypothesis text under a plurality of evaluation dimensions.
2. The method for automatically generating scientific hypotheses for a multi-agent collaboration of claim 1, wherein inputting a problem to be studied and research background information of the problem to be studied into a creative generation agent to obtain a plurality of creative hypothesis texts for the problem to be studied includes: carrying out knowledge graph association analysis on the problems to be researched and research background information of the problems to be researched so as to obtain structured domain knowledge of the problems to be researched; Determining a candidate innovation point data set of the problem to be researched from the structured domain knowledge according to the problem to be researched, wherein the candidate innovation point data set comprises at least one innovation point description text of the problem to be researched; One of the creative hypothesis texts is generated based on each of the creative description texts in the creative data set.
3. The method of claim 1, wherein the inputting the plurality of creative hypothesis texts into hypothesis test agents, respectively, to obtain the evaluation data for each of the creative hypothesis texts under a plurality of evaluation dimensions, respectively, comprises: Carrying out knowledge graph reasoning analysis on each creative assumption text based on the structured domain knowledge of the creative assumption text to obtain a reasoning analysis result of each creative assumption text, wherein the reasoning analysis result is used for representing semantic association degree and/or logic support degree between each creative assumption text and the structured domain knowledge; Respectively carrying out forensic arguments among forensic agent agents under a plurality of different expert identities on each creative hypothesis text to obtain a forensic arguments result of each creative hypothesis text, wherein the forensic arguments result is used for representing at least one of feasibility conclusion, reasonability conclusion, potential risk conclusion, innovation conclusion and credibility conclusion of the creative hypothesis text; and determining the evaluation data of each creative assumption text under a plurality of evaluation dimensions according to the respective reasoning analysis results and dialect demonstration results of each creative assumption text.
4. The multi-agent collaborative scientific hypothesis automatic generation method of claim 3, wherein said respectively performing forensic arguments among forensic agent agents under a plurality of different expert identities for each of said creative hypothesis texts to obtain forensic arguments for each of said creative hypothesis texts comprises: Inputting the creative hypothesis text into a dialect agent under at least one first expert identity to obtain at least one first argumentation result of the creative hypothesis text, wherein each first argumentation result is used for representing data of supporting attributes of the creative hypothesis text in the structured domain knowledge; Inputting the creative hypothesis text into a dialect agent under at least one second expert identity to obtain at least one second argumentation result of the creative hypothesis text, wherein each second argumentation result is used for representing data of anti-attribute of the creative hypothesis text in the structured domain knowledge; And inputting the first argumentation result and the second argumentation result into a dialectical agent under a third expert identity together to obtain a dialectical argumentation result of the creative hypothesis text.
5. The multi-agent collaborative scientific hypothesis automatic generation method of claim 4, the method is characterized in that the method for automatically generating the scientific hypothesis of multi-agent cooperation further comprises the following steps: Updating the creative hypothesis text entered into the forensic agent under the first expert identity by means of at least one of the second demonstration results and returning to the forensic agent under the at least one first expert identity for entering the creative hypothesis text to obtain at least one first demonstration result of the creative hypothesis text; Updating the creative hypothesis text entered into the forensic agent under the second expert identity by means of at least one of the first demonstration results and returning to the step of entering the creative hypothesis text into the forensic agent under the at least one second expert identity to obtain at least one second demonstration result of the creative hypothesis text.
6. The multi-agent collaborative scientific hypothesis automatic generation method of claim 1, wherein the determining target creative hypothesis text corresponding to the problem under study from all of the creative hypothesis text based on the rating data for each of the creative hypothesis texts under a plurality of rating dimensions, respectively, comprises: comparing the evaluation data of each two creative hypothesis texts under different evaluation dimensions in pairs, determining a comparison evaluation result between each two creative hypothesis texts, and determining a relative evaluation value between each two creative hypothesis texts according to the comparison evaluation result between each two creative hypothesis texts; sorting all of the creative hypothesis texts based on the relative evaluation values between each two of the creative hypothesis texts, and determining the target creative hypothesis text from the sorted creative hypothesis texts.
7. The multi-agent collaborative scientific hypothesis automatic generation method of claim 6, wherein the determining a relative evaluation value between each two creative hypothesis texts based on a comparison evaluation result between each two creative hypothesis texts comprises: according to the initial embedded evaluation values of the two creative hypothesis texts, characterizing the contrast evaluation results through preset learning rate parameters, and characterizing the difference of the two creative hypothesis texts under different evaluation dimensions to obtain the difference characteristics between the two creative hypothesis texts; And respectively adjusting the embedded evaluation values of the two creative hypothesis texts according to the differentiation characteristics, and determining the difference value between the two adjusted embedded evaluation values as the relative evaluation value between the two creative hypothesis texts.
8. The utility model provides a scientific assumption automatic generation device that many agents cooperated which characterized in that includes: the creative generation module is used for inputting the to-be-researched problem and research background information of the to-be-researched problem into a creative generation agent so as to obtain a plurality of creative hypothesis texts of the to-be-researched problem; The creative evaluation module is used for inputting a plurality of creative hypothesis texts into a hypothesis test agent respectively so as to obtain evaluation data of each creative hypothesis text under a plurality of evaluation dimensions respectively; And the creative screening module is used for determining target creative hypothesis texts corresponding to the problems to be researched from all the creative hypothesis texts according to the evaluation data of each creative hypothesis text under a plurality of evaluation dimensions.
9. A computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, which when executed by a processor, implements the multi-agent collaborative scientific hypothesis automatic generation method of any one of claims 1-7.
10. An electronic device comprising a processor, a memory and a computer program stored on the memory and executable on the processor, the computer program implementing the steps of the multi-agent collaborative scientific hypothesis automatic generation method of any one of claims 1-7 when executed by the processor.

Description

Multi-agent cooperative scientific hypothesis automatic generation method Technical Field The application belongs to the field of creative scientific research engineering, and particularly relates to a multi-agent cooperative scientific hypothesis automatic generation method, device, equipment and storage medium. Background In scientific research, researchers often need to bring innovative scientific assumptions around specific questions to be studied, and make comprehensive evaluations of these assumptions in terms of feasibility, rationality, and potential risk to determine the most valuable target assumptions to be studied. With the continuous increase of the complexity of scientific research problems, researchers often need to process a large amount of interdisciplinary background information and generate and screen a plurality of candidate hypotheses in a limited time, so that higher demands are put on intelligent and systematic support of hypothesis generation and verification processes. In the prior art, rule-based automated methods are often employed to assist researchers in generating scientific assumptions. For example, some methods assist in hypothesis generation by inputting research questions into a language model to generate several candidate hypotheses, or by performing a correlation analysis on the input questions using knowledge maps, while techniques exist to initially screen candidate hypotheses by performing simple text similarity calculations, rule-based logic checks, or scoring of a single evaluation index on the generated hypotheses. However, the technical proposal is easy to generate the problems of insufficient innovation of the hypothesis, one-sided evaluation result, incomplete inference chain and the like when the complex scientific research problem is processed, and the target creative hypothesis text which meets the research requirement is difficult to be effectively determined. Therefore, scientific researchers still need to put a great deal of manpower to screen and judge, and the efficiency is low. Disclosure of Invention The application aims to provide a multi-agent cooperative scientific hypothesis automatic generation method, device, equipment and storage medium, which at least solve the problems of low efficiency of an automatic scientific hypothesis generation and screening process caused by insufficient innovation of an automatic hypothesis generation process, one-sided hypothesis evaluation results and incomplete reasoning chains. In a first aspect, an embodiment of the present application discloses a method for automatically generating scientific assumptions for multi-agent collaboration, including: Inputting a to-be-researched problem and research background information of the to-be-researched problem into a creative generating agent to obtain a plurality of creative hypothesis texts of the to-be-researched problem; inputting a plurality of creative hypothesis texts into hypothesis testing agents respectively to obtain evaluation data of each creative hypothesis text under a plurality of evaluation dimensions respectively; And determining target creative hypothesis texts corresponding to the problems to be researched from all the creative hypothesis texts according to the evaluation data of each creative hypothesis text under a plurality of evaluation dimensions. In a second aspect, the embodiment of the application also discloses a device for automatically generating scientific assumptions by multi-agent cooperation, which comprises the following steps: the creative generation module is used for inputting the to-be-researched problem and research background information of the to-be-researched problem into a creative generation agent so as to obtain a plurality of creative hypothesis texts of the to-be-researched problem; The creative evaluation module is used for inputting a plurality of creative hypothesis texts into a hypothesis test agent respectively so as to obtain evaluation data of each creative hypothesis text under a plurality of evaluation dimensions respectively; And the creative screening module is used for determining target creative hypothesis texts corresponding to the problems to be researched from all the creative hypothesis texts according to the evaluation data of each creative hypothesis text under a plurality of evaluation dimensions. In a third aspect, an embodiment of the present application further discloses an electronic device, including a processor and a memory, where the memory stores a program or instructions executable on the processor, where the program or instructions implement the steps of the method according to the first aspect when executed by the processor. In a fourth aspect, embodiments of the present application also disclose a readable storage medium having stored thereon a program or instructions which, when executed by a processor, implement the steps of the method as described in the first aspect. In summary, in the embod