CN-121981260-A - Answer decision method, device, computer equipment and storage medium

Abstract

The application belongs to the technical field of artificial intelligence, and relates to an answer decision method which comprises the steps of obtaining contradictory answers output by a plurality of intelligent agents, determining conflict levels of the contradictory answers according to a preset conflict level triggering rule, and determining a conflict decision scheme of the contradictory answers according to the conflict levels. The application also provides an answer decision device, computer equipment and a storage medium. The application can be applied to business management program systems such as intelligent medical treatment, financial science and technology, and the like, can solve the conflict problem of the contradiction answers of multi-professional intelligent body, and improves the accuracy and reliability of the decision of the contradiction answers.

Inventors

• WANG JIANZONG
• DENG YUWEI

Assignees

• 平安科技（深圳）有限公司

Dates

Publication Date

20260505

Application Date

20260107

Claims (10)

1. 1. An answer decision method, said method comprising the steps of: Obtaining a plurality of contradictory answers output by the intelligent agents, wherein the number of the contradictory answers is at least two; according to a preset conflict level triggering rule, determining the conflict level of the conflict answer; and determining a conflict decision scheme of the conflict answer according to the conflict level.
2. 2. The answer decision method according to claim 1, wherein the step of determining the conflict level of the contradictory answer according to a preset conflict level triggering rule comprises: acquiring knowledge sources and answer confidence degrees of all the contradictory answers; Calculating answer similarity according to the contradictory answers, calculating source similarity according to the knowledge source, and calculating confidence difference according to the answer confidence; determining a conflict level score based on the answer similarity, the source similarity, and the confidence difference; and acquiring a preset threshold interval, and determining the conflict level of the conflict answer according to the conflict level score and the threshold interval.
3. 3. The answer decision method according to claim 1, wherein said step of determining a conflict decision scheme of said contradictory answers according to said conflict level comprises: Extracting key information from each contradiction answer, and matching the key information with a knowledge source base constructed in advance to obtain associated knowledge clauses; Acquiring a timeliness coefficient and an authority coefficient of the associated knowledge clause; Calculating the credibility score of each contradiction answer according to the timeliness coefficient and the authority coefficient; sorting the contradictory answers according to the credibility score, and screening to obtain a first target answer; and if the conflict level is a low-level conflict, determining the first target answer as an optimal answer.
4. 4. The answer decision method of claim 3 in which, after said step of screening the first target answer based on said confidence score of said contradictory answer, further comprising: extracting an reasoning path of each contradictory answer from the intelligent agent, wherein the reasoning logic comprises a precondition hypothesis, logic steps and a conclusion; generating a visual flow chart according to the premise hypothesis, the logic step and the conclusion; comparing a pre-constructed logic error recognition library with the visual flow chart to obtain a logic contradiction point; Generating an inference difference analysis report according to the logic contradiction points; determining a second target answer according to the reliability scores of the contradictory answers and the reasoning difference analysis report; And if the conflict level is a medium-level conflict, determining the second target answer as an optimal answer.
5. 5. The answer decision method of claim 4 in which prior to said step of extracting the inference path for each of said answers to contradictions from said agent, further comprising: Acquiring an inference rule of the agent; Decomposing the reasoning rule into a sufficient condition and a necessary condition; mapping the sufficient conditions and the necessary conditions to a general logic framework to generate a logic error recognition library.
6. 6. The answer decision method of claim 4 in which, after said step of determining a second target answer from said report of inferential variance analysis and said confidence score for each said contradictory answer, further comprising: Classifying the logic contradiction points in the reasoning difference analysis report to obtain contradiction categories; If the conflict category is a quantifiable conflict point, generating a virtual verification instruction according to a conflict answer corresponding to the quantifiable conflict point; a virtual simulation interface is called to generate verification experiment parameters according to the virtual verification instruction, and the coincidence degree of the verification experiment parameters and the quantifiable contradiction points is determined; sequencing all the contradictory answers according to the fitness to determine a third target answer; And if the conflict level is a high-level conflict, determining the third target answer as an optimal answer.
7. 7. The answer decision method of claim 6 in which, after said step of classifying logical contradictory points in said inference difference analysis report to obtain contradictory categories, further comprising: If the contradiction category is an unquantifiable contradiction point, generating a clarification question according to the unquantifiable contradiction point, wherein the clarification question is used for acquiring the user supplementary information; Determining a correctness probability of the unquantifiable contradiction point according to the user supplementary information; And sequencing all the contradictory answers according to the correctness probability, and determining a third target answer.
8. 8. An answer decision device comprising: the answer acquisition module is used for acquiring a plurality of contradictory answers output by the intelligent agents, wherein the number of the contradictory answers is at least two; the conflict determination module is used for determining the conflict level of the conflict answer according to a preset conflict level triggering rule; And the answer decision module is used for determining a conflict decision scheme of the contradictory answers according to the conflict level.
9. 9. A computer device comprising a memory and a processor, the memory having stored therein computer readable instructions which when executed by the processor implement the steps of the answer decision method of any one of claims 1 to 7.
10. 10. A computer readable storage medium having stored thereon computer readable instructions which when executed by a processor implement the steps of the answer decision method according to any one of claims 1 to 7.

Description

Answer decision method, device, computer equipment and storage medium Technical Field The application relates to the technical field of artificial intelligence, and is applied to online processing business scenes of intelligent medical science, financial science and technology, in particular to an answer decision method, an answer decision device, computer equipment and a storage medium. Background In the fields of medical treatment, finance and the like with extremely high requirements on decision accuracy, the application of multi-professional intelligent agents is increasingly wide. In medical scenes, different specialized agents, such as image diagnosis agents, pathology analysis agents and clinical experience agents, may give contradictory diagnosis suggestions for the same condition, and in the financial field, macro economic analysis agents, industry research agents and quantization investment agents, may output conflicting decision schemes for the same investment target. However, there is currently a lack of conflict resolution technology for multi-specialty agents to output answers to contradictions. The technologies related to agents on the market focus on answer generation and optimization of a single agent and collaborative task allocation of multiple agents, but do not involve systematic conflict resolution when multiple professional agents output contradictory answers to the same question. Therefore, the existing conflict resolution mechanism cannot fundamentally process conflicts among answers, and only can perform surface result screening, so that answer correctness is reduced. Therefore, an answer decision scheme is needed to solve the contradictory answer conflict problem of multi-professional agents. Disclosure of Invention The embodiment of the application aims to provide an answer decision method, an answer decision device, computer equipment and a storage medium, which aim to solve the conflict problem of the contradiction answers of multi-professional intelligent agents. In a first aspect, an answer decision method is provided, which adopts the following technical scheme: Obtaining a plurality of contradictory answers output by the intelligent agents, wherein the number of the contradictory answers is at least two; according to a preset conflict level triggering rule, determining the conflict level of the conflict answer; and determining a conflict decision scheme of the conflict answer according to the conflict level. In a second aspect, an answer decision device is provided, which adopts the following technical scheme: the answer acquisition module is used for acquiring a plurality of contradictory answers output by the intelligent agents, wherein the number of the contradictory answers is at least two; the conflict determination module is used for determining the conflict level of the conflict answer according to a preset conflict level triggering rule; And the answer decision module is used for determining a conflict decision scheme of the contradictory answers according to the conflict level. In a third aspect, a computer device is provided, comprising a memory having stored therein computer readable instructions, and a processor, which when executed, performs the steps of the answer decision method described above. In a fourth aspect, a computer readable storage medium having computer readable instructions stored thereon which, when executed by a processor, implement the steps of the answer decision method described above is provided. Compared with the prior art, the embodiment of the application has the following main beneficial effects: According to the embodiment of the application, the conflict level of the conflict answers is determined according to the preset conflict level triggering rule by acquiring the conflict answers outputted by a plurality of intelligent agents, so that the conflict degree and the influence result of the conflict answers can be accurately identified, and the conflict decision scheme of the conflict answers is determined according to the conflict level of the conflict answers, so that the differentiation and refinement of decisions are realized. Different conflict levels correspond to different decision requirements and risk tolerance levels. For the case of lower conflict level, a simpler, more convenient and quicker coordination decision scheme can be adopted, and for the case of higher conflict level which possibly affects important decisions, a more strict and comprehensive decision scheme is adopted. The method for determining the decision scheme according to the conflict level can fundamentally process the conflict among answers, not just surface result screening, and can improve the accuracy and reliability of the contradictory answer decision. Drawings In order to more clearly illustrate the solution of the present application, a brief description will be given below of the drawings required for the description of the embodiments of the pres