An intelligent decision-making system based on large models

By introducing decision scenario creation, information acquisition and preprocessing modules into the intelligent decision-making system, combining large models and machine learning algorithms, the problems of data error and analysis singularity in traditional systems are solved, and more accurate and personalized decision-making support is achieved. .

CN118780373BActive Publication Date: 2024-12-06FUJIAN BIG DATA PRIMARY DEV CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411241402.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-05
Publication Date
2024-12-06
Estimated Expiration
2044-09-05

AI Technical Summary

Technical Problem

Traditional intelligent decision-making systems have errors or deviations in data collection and processing, and the analysis scenarios are too single, so they cannot deeply understand user needs and provide personalized decision-making support.

Method used

An intelligent decision-making system based on large models is designed, including decision scenario creation module, decision information parameter acquisition module, decision information preprocessing module, decision information analysis module and human-computer interaction module. By creating decision scenarios in the system interface, users can obtain and preprocess decision information, establish a comprehensive decision evaluation model, calculate the comprehensive decision evaluation index, judge the executability of decisions, and issue early warning signals.

Benefits of technology

It realizes intelligent processing of complex decision-making problems, provides comprehensive suggestions, improves the accuracy and personalized support of decision-making results, and enhances the system's data collection, processing and analysis capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118780373B_ABST
    Figure CN118780373B_ABST
Patent Text Reader

Abstract

The present invention discloses an intelligent decision-making system based on a large model, which specifically relates to the field of intelligent decision-making, including a decision scenario creation module, a decision information parameter acquisition module, a decision information preprocessing module, a decision information analysis module, a decision judgment module and a human-computer interaction module. The basic decision information, including decision keyword information, decision accuracy parameters and expert qualification parameters, is acquired through the decision information parameter acquisition module, and the decision information preprocessing module and the decision information analysis module are used to calculate the decision comprehensive evaluation index with the acquired data, so as to provide convenience for users in the decision-making process. The present invention realizes intelligent processing of complex decision-making problems by integrating multiple modules such as data collection, processing, analysis and decision execution, and provides comprehensive suggestions for users' decision-making by establishing a large model and combining algorithms such as machine learning.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of intelligent decision-making technology, and more specifically, to an intelligent decision-making system based on a large model. Background Art

[0002] With the rapid development of artificial intelligence technologies such as deep learning and reinforcement learning, the performance and intelligence of intelligent decision-making systems have also been greatly improved. By introducing different machine learning algorithms and applying them to multiple fields such as finance and education, the intensified market competition and the diversification of people's needs have provided new development prospects for intelligent decision-making systems. Intelligent decision-making systems based on large models provide scientific decision-making support for enterprises and governments through analysis and mining of massive data, helping them to achieve intelligent transformation.

[0003] At present, the traditional decision-making system includes a scenario creation module, a decision information parameter collection module, a decision information analysis module and a human-computer interaction module. The scenario creation module is used for users to create decision scenarios and provide a suitable decision-making environment for decision-making; the decision information parameter collection module collects decision information parameters by training a large model; the decision information analysis module is used to establish a decision information analysis model and perform integrated analysis on the collected data; the human-computer interaction module visualizes the decision information through an intelligent decision-making system, and users provide suggestions for the feasibility of the decision through the decision information.

[0004] However, there are still some shortcomings in its actual use. For example, errors or deviations may occur in the data collection and processing process of the intelligent decision-making system. For example, incomplete data collection or excessive influence of historical decision data may affect the accuracy of decision results. It is necessary to introduce preprocessing technology to improve data quality. The system needs to collect different parameters and analyze different decisions according to different decisions. The analysis scenario of the traditional intelligent decision-making system is too single. More personalized decision-making support is needed to have a deeper understanding of user needs and provide a decision-making system that better meets user needs. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an intelligent decision-making system based on a large model, which solves the problems raised in the above-mentioned background technology through the following scheme.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The decision scenario creation module is used to create a scenario for decision making. The user operates in the system user interface, and the user selects a desired scenario template from the preset scenario templates, and matches the business in the selected scenario template with the user's goal or standard, and the user's goal or standard includes the solution direction of the corresponding decision problem;

[0008] The decision information parameter acquisition module is used to acquire information of the text to be decided, including a decision keyword parameter acquisition unit, a decision accuracy parameter acquisition unit and an expert qualification parameter acquisition unit, and transmits the collected data to the decision information preprocessing module;

[0009] The decision information preprocessing module includes a decision keyword preprocessing unit, a decision accuracy preprocessing unit and an expert qualification preprocessing unit, which are used to preprocess the data transmitted by the decision information parameter acquisition module and transmit the processing results to the decision information analysis module;

[0010] The decision information analysis module is used to analyze the data of the decision information preprocessing module, establish a decision comprehensive evaluation model, import relevant decision accuracy coefficients and decision accuracy coefficients into the decision comprehensive evaluation model, and calculate the decision comprehensive evaluation index;

[0011] The decision judgment module is used to establish a preset value of the decision comprehensive evaluation index, make a judgment based on the preset value of the decision comprehensive evaluation index and the decision comprehensive evaluation index, and issue a warning signal according to the judgment result;

[0012] The human-computer interaction module is used to send signals for unexecutable decisions. The intelligent decision-making system transmits the judgment results and warning signals in the decision judgment module to the management control panel through IDSS, and transmits unexecutable decision information to the management through the control panel.

[0013] Preferably, the decision keyword parameter acquisition unit extracts keywords from the information to be decided by training models and data, the information to be decided includes decision text, images, and videos, etc., obtains each key word in the information to be decided, forms a key word set of the information to be decided, extracts the key words and their corresponding historical keyword sets, analyzes the matching degree of the key words to be decided and their corresponding historical keyword sets, and the key word set is recorded as p u i ={p u 1,p u 2,...,p u i ,...,p u n}, where p u i It is represented as the i-th keyword in the u-th decision-making information, where u=1,2,…,z, i=1,2,…,n, and the historical decision key word set corresponding to the decision-making information stored in the decision storage database is extracted, and the corresponding historical decision key word set is recorded as q f v ={q f 1,q f 2,...,qf v ,...,q f h}, where q f v It is represented as the vth key word in the fth corresponding historical decision information, where f = 1, 2, ..., t, v = 1, 2, ..., h.

[0014] Preferably, the decision accuracy parameter acquisition unit analyzes the decision accuracy parameters corresponding to historical decisions based on the successfully trained designated model, the decision accuracy parameters including the number of successful decisions, the expected number of decisions and the total number of decisions. According to the required solutions for various decisions, historical decisions are evaluated, the situations after the implementation of various historical decisions are recorded, and successful decision results are marked as "1", and failed decision results are marked as "0". The number of successful decisions and the total number of decisions are counted, which are denoted as Vz and Cz respectively. Whether the decision meets expectations is judged according to pre-set goals or standards, and the expected number of decisions is counted, which is denoted as Qz.

[0015] Preferably, the expert qualification parameter acquisition unit acquires expert qualification parameters by training a model to simulate human experts solving problems in professional fields. The expert qualification parameters include the capacity of the expert knowledge base and the number of expert recalls. The capacity of the expert knowledge base includes rules, facts, experiences and cases provided by experts in various fields, denoted as Cw, and the number of expert recalls determines whether it is a correct decision based on the defined correct standards, and counts the number of correct decisions, denoted as Rw.

[0016] Preferably, the decision keyword preprocessing unit is used to establish a decision keyword preprocessing model, import the keyword parameters transmitted by the decision information parameter acquisition module into the decision keyword preprocessing module model, and calculate the keyword matching degree. The specific analysis formula is: , obtain the matching degree between the u-th key word of the information to be decided and its corresponding historical key word, according to the matching degree between the key word of the information to be decided and its corresponding historical key word, if the matching degree between a key word of the information to be decided and its corresponding historical key word is less than the preset keyword matching degree threshold, it indicates that the keyword information to be decided is an irrelevant keyword; if the matching degree between a key word of the information to be decided and its corresponding historical key word is greater than or equal to the preset keyword matching degree threshold, it indicates that the keyword information to be decided is a relevant keyword, and the historical decision information corresponding to the relevant key words to be decided is screened and counted, and recorded as relevant historical decision information.

[0017] Preferably, the relevant historical decision information includes the corresponding number of historical decision information, the number of historical successful decisions, and the number of historical decisions that meet expectations, which are respectively denoted as Ch, Vh, and Qh. The accuracy and precision of the relevant historical decisions are calculated. The calculation formula for the accuracy of the relevant historical decisions is: , the calculation formula of relevant historical decision accuracy is: ; Import the accuracy and precision of relevant historical decisions into the specific analysis formula of the relevant decision accuracy coefficient: ,in represents the accuracy of historical decisions related to the standard, represents the accuracy of the standard relevant historical decision, and α1 and α2 represent the influencing factors of the accuracy and precision of the relevant historical decision, respectively.

[0018] Preferably, the decision accuracy preprocessing unit is used to establish a decision accuracy preprocessing model, import the decision accuracy parameters transmitted by the decision information parameter acquisition module into the decision accuracy preprocessing model, and calculate the decision accuracy coefficient, which includes historical decision accuracy and precision. The historical decision accuracy calculation formula is: , where Vz m represents the number of successful decisions corresponding to the mth decision, Cz represents the total number of decisions, and the calculation formula for historical decision accuracy is: , where Qz m Indicates the expected number of decisions corresponding to the mth decision; historical decision accuracy and historical decision precision are introduced into the specific analysis formula of the decision accuracy coefficient: , where ω m Expressed as the decision accuracy coefficient corresponding to the mth decision, ω 1m and ω 2m They represent historical decision accuracy and historical decision precision, ω1´ and ω2´ represent standard decision accuracy and standard decision precision, β1 and β2 represent the influencing factors of historical decision accuracy and precision, respectively.

[0019] Preferably, the expert qualification preprocessing unit is used to establish an expert qualification preprocessing model, import the expert qualification parameters transmitted by the decision information parameter acquisition module into the expert qualification preprocessing model, and calculate the expert qualification coefficient. The expert qualification coefficient includes a probability index and an expert decision recall rate. The calculation formula of the probability index is:

[0020] , where P m represents the probability that the system finds relevant knowledge in the expert knowledge base in the mth decision, Cw m represents the capacity of the expert knowledge base in the mth decision, PI represents the knowledge capacity required for decision making, and P M represents the standard probability of finding relevant knowledge, Z m represents the probability index of the success of the mth decision, k represents the total number of various decisions, and the calculation formula of the expert decision recall rate is: , where Rw represents the number of expert recalls, and the probability index and expert decision recall rate are introduced into the expert qualification coefficient. The specific analysis formula is:

[0021] , where ζ´ represents the standard expert recall rate and ζ represents the expert seniority coefficient.

[0022] Preferably, the specific analysis formula of the decision comprehensive evaluation index is: , where ω m It is represented as the decision accuracy coefficient corresponding to the m-th decision, ζ represents the expert qualification coefficient, It represents the relevant decision accuracy coefficient. The higher the decision accuracy, the higher the probability that the system can find relevant knowledge in the knowledge base, and the greater the comprehensive decision evaluation index.

[0023] Preferably, the preset value of the decision comprehensive evaluation index is marked as φ Def , when φ Def <φ, it means that the decision comprehensive evaluation index is greater than the preset value of the decision comprehensive evaluation index, indicating that this decision is executable. Def >φ, it means that the decision comprehensive evaluation index is less than the preset value of the decision comprehensive evaluation index, indicating that this decision is an unexecutable decision. The judgment result is transmitted to the human-computer interaction module, and a warning signal is issued for the unexecutable decision through the decision judgment module.

[0024] Preferably, the IDSS is an intelligent decision-making system, which combines artificial intelligence technology, data mining, machine learning, expert systems, natural language processing and other advanced technologies to provide more intelligent and automated decision support. For executable decisions, the intelligent decision system provides managers with data for better understanding and analysis. For unexecutable decisions, the intelligent decision system provides managers with unexecutable decision information, where the unexecutable decision information includes the potential risks of the decision identified by the intelligent decision system when analyzing the data and predicted future results, and provides decision recommendations based on these analyses.

[0025] Technical effects and advantages of the present invention:

[0026] 1. The present invention obtains basic decision information, including decision keyword information, decision accuracy parameters and expert qualification parameters, through a decision information parameter acquisition module, and imports the acquired data into a preprocessing model through a decision information preprocessing module, and obtains the decision keyword matching degree and the corresponding decision accuracy coefficient, the historical decision accuracy coefficient and the expert qualification coefficient through processing. These data can intuitively reflect the decision information and system capabilities;

[0027] 2. Combine the decision keyword matching degree and the corresponding decision accuracy coefficient, the historical decision accuracy coefficient and the expert qualification coefficient, and transmit the data to the decision information analysis module for further analysis to obtain the decision comprehensive evaluation index. By judging the decision comprehensive evaluation index, it is evaluated whether the decision to be executed is an executable decision, which provides convenience for users in the decision-making process. The present invention integrates multiple modules such as data collection, processing, analysis and decision execution to realize intelligent processing of complex decision-making problems, and uses the establishment of a large model combined with machine learning and other algorithms to provide comprehensive suggestions for user decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a schematic diagram of the overall structure of the present invention. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] As attached Figure 1 An intelligent decision-making system based on a large model is shown, comprising a decision scenario creation module, a decision information parameter acquisition module, a decision information preprocessing module, a decision information analysis module, a decision judgment module and a human-computer interaction module.

[0031] The output end of the decision scenario creation module is connected to the input end of the decision information parameter acquisition module, the output end of the decision information parameter acquisition module is connected to the input end of the decision information preprocessing module, the output end of the decision information preprocessing module is connected to the input end of the decision information analysis module, the output end of the decision information analysis module is connected to the input end of the decision judgment module, and the output end of the decision judgment module is connected to the input end of the human-computer interaction module.

[0032] The decision scenario creation module is used to create scenarios for decision-making. The user operates in the system user interface, selects the desired scenario template from the preset scenario templates, and matches the business in the selected scenario template with the user's goals or standards. The user's goals or standards include the solution direction of the corresponding decision problem, such as financial direction, foreign trade direction, etc.

[0033] The decision information parameter acquisition module is used to acquire information of the text to be decided, including a decision keyword parameter acquisition unit, a decision accuracy parameter acquisition unit and an expert qualification parameter acquisition unit, and transmits the collected data to the decision information preprocessing module.

[0034] In this embodiment, it should be specifically explained that the decision keyword parameter acquisition unit extracts keywords from the decision information through training models and data, and the decision information includes decision text, images, and videos, etc., obtains each key word in the decision information, constitutes a key word set of the decision information, and extracts the key words and their corresponding historical keyword sets, analyzes the matching degree of the decision key words and their corresponding historical keyword sets, and the key word set is recorded as p u i ={p u 1,p u 2,...,p u i ,...,p u n}, where p u i It is represented as the i-th keyword in the u-th decision-making information, where u=1,2,…,z, i=1,2,…,n, and the historical decision key word set corresponding to the decision-making information stored in the decision storage database is extracted, and the corresponding historical decision key word set is recorded as q f v ={q f 1,q f 2,...,q f v ,...,q f h}, where q f v It is represented as the vth key word in the fth corresponding historical decision information, where f = 1, 2, ..., t, v = 1, 2, ..., h.

[0035] In this embodiment, it is necessary to specifically explain that the decision accuracy parameter acquisition unit analyzes the decision accuracy parameters corresponding to historical decisions based on the successfully trained designated model. The decision accuracy parameters include the number of successful decisions, the expected number of decisions and the total number of decisions. According to the required solutions for various decisions, historical decisions are evaluated, the situations after the implementation of various historical decisions are recorded, and the successful decision results are marked as "1", and the failed decision results are marked as "0". The number of successful decisions and the total number of decisions are counted, which are denoted as Vz and Cz respectively; whether the decision meets expectations is judged according to the pre-set goals or standards, and the expected number of decisions is counted, which is denoted as Qz.

[0036] In this embodiment, it should be specifically explained that the expert qualification parameter acquisition unit obtains the expert qualification parameters by training the model to simulate human experts solving problems in professional fields. The expert qualification parameters include the capacity of the expert knowledge base and the number of expert recalls. The capacity of the expert knowledge base includes the rules, facts, experiences and cases provided by experts in various fields, denoted as Cw, and the number of expert recalls determines whether it is a correct decision based on the defined correct standards, and counts the number of correct decisions, denoted as Rw.

[0037] The decision information preprocessing module includes a decision keyword preprocessing unit, a decision accuracy preprocessing unit and an expert qualification preprocessing unit, which are used to preprocess the data transmitted by the decision information parameter acquisition module and transmit the processing results to the decision information analysis module.

[0038] In this embodiment, it should be specifically explained that the decision keyword preprocessing unit is used to establish a decision keyword preprocessing model, import the keyword parameters transmitted by the decision information parameter acquisition module into the decision keyword preprocessing module model, and calculate the keyword matching degree. The specific analysis formula is: , get the matching degree between the uth key word of the decision information and its corresponding historical key word, denoted as ζ u , where p u i It is represented by the i-th keyword in the u-th decision-making information, where u=1,2,…,z, i=1,2,…,n,q f v It is expressed as the vth key word in the fth corresponding historical decision information, where f=1, 2, ..., t, v=1, 2, ..., h, according to the matching degree between the key words of the information to be decided and its corresponding historical key words, if the matching degree between a key word of the information to be decided and its corresponding historical key words is less than the preset keyword matching degree threshold, it indicates that the keyword information to be decided is an irrelevant keyword; if the matching degree between a key word to be decided and its corresponding historical key words is greater than or equal to the preset keyword matching degree threshold, it indicates that the keyword information to be decided is a relevant keyword, and the historical decision information corresponding to the relevant key words to be decided is screened and counted, and recorded as relevant historical decision information.

[0039] Furthermore, the relevant historical decision information includes the number of historical decision information corresponding thereto, the number of historical successful decisions, and the number of historical decisions that meet expectations, which are respectively denoted as Ch, Vh, and Qh. The accuracy and precision of the relevant historical decisions are calculated. The calculation formula for the accuracy of the relevant historical decisions is: , the calculation formula of relevant historical decision accuracy is: ; Import the accuracy and precision of relevant historical decisions into the specific analysis formula of the relevant decision accuracy coefficient: ,in represents the accuracy of historical decisions related to the standard, It represents the accuracy of standard relevant historical decisions, α1 and α2 represent the influencing factors of the accuracy and precision of relevant historical decisions, such as the impact of changes in market environment and industry trends on relevant decisions, and transmit the processing results to the decision information analysis module.

[0040] In this embodiment, it should be specifically explained that the decision accuracy preprocessing unit is used to establish a decision accuracy preprocessing model, import the decision accuracy parameters transmitted by the decision information parameter acquisition module into the decision accuracy preprocessing model, and calculate the decision accuracy coefficient. The decision accuracy coefficient includes historical decision accuracy and precision. The historical decision accuracy calculation formula is: , where Vz m represents the number of successful decisions corresponding to the mth decision, Cz represents the total number of decisions, and the calculation formula for historical decision accuracy is: , where Qz m Indicates the expected number of decisions corresponding to the mth decision; historical decision accuracy and historical decision precision are introduced into the specific analysis formula of the decision accuracy coefficient: , where ω m Expressed as the decision accuracy coefficient corresponding to the mth decision, ω 1m and ω 2m They represent the historical decision accuracy and historical decision precision respectively, ω1´ and ω2´ represent the standard decision accuracy and standard decision precision respectively, β1 and β2 represent the influencing factors of historical decision accuracy and precision respectively, such as the changes in the political, economic and technological aspects of historical decisions, and the processing results are transmitted to the decision information analysis module.

[0041] In this embodiment, it should be specifically explained that the expert qualification preprocessing unit is used to establish an expert qualification preprocessing model, import the expert qualification parameters transmitted by the decision information parameter acquisition module into the expert qualification preprocessing model, and calculate the expert qualification coefficient. The expert qualification coefficient includes a probability index and an expert decision recall rate. The calculation formula of the probability index is:

[0042] , where P m represents the probability that the system finds relevant knowledge in the expert knowledge base in the mth decision, Cw m represents the capacity of the expert knowledge base in the mth decision, PI represents the knowledge capacity required for decision making, and P M represents the standard probability of finding relevant knowledge, Z m represents the probability index of the success of the mth decision, k represents the total number of various decisions, and the calculation formula of the expert decision recall rate is: , where Rw represents the number of expert recalls, and the probability index and expert decision recall rate are introduced into the expert qualification coefficient. The specific analysis formula is:

[0043] , where ζ´ represents the standard expert recall rate and ζ represents the expert seniority coefficient.

[0044] The decision information analysis module is used to analyze the data of the decision information preprocessing module, establish a decision comprehensive evaluation model, import relevant decision accuracy coefficients and decision accuracy coefficients into the decision comprehensive evaluation model, and calculate the decision comprehensive evaluation index.

[0045] In this embodiment, it should be specifically explained that the specific analysis formula of the decision comprehensive evaluation index is: , where ω m It is represented as the decision accuracy coefficient corresponding to the m-th decision, ζ represents the expert qualification coefficient, It represents the relevant decision accuracy coefficient. The higher the decision accuracy, the higher the probability that the system can find relevant knowledge in the knowledge base, and the greater the comprehensive decision evaluation index.

[0046] The decision judgment module is used to establish a preset value of the decision comprehensive evaluation index, make a judgment based on the preset value of the decision comprehensive evaluation index and the decision comprehensive evaluation index, and issue a warning signal according to the judgment result.

[0047] In this embodiment, it should be specifically noted that the preset value of the decision comprehensive evaluation index is marked as φ Def , when φ Def <φ, it means that the decision comprehensive evaluation index is greater than the preset value of the decision comprehensive evaluation index, indicating that this decision is executable. Def >φ, it means that the decision comprehensive evaluation index is less than the preset value of the decision comprehensive evaluation index, indicating that this decision is an unexecutable decision. The judgment result is transmitted to the human-computer interaction module, and a warning signal is issued for the unexecutable decision through the decision judgment module.

[0048] The human-computer interaction module is used to send signals for unexecutable decisions. The intelligent decision-making system transmits the judgment results and warning signals in the decision judgment module to the management control panel through IDSS, and transmits unexecutable decision information to the management through the control panel.

[0049] In this embodiment, it should be specifically explained that the IDSS is an intelligent decision-making system, which combines artificial intelligence technology, data mining, machine learning, expert systems, natural language processing and other advanced technologies to provide more intelligent and automated decision support. For executable decisions, the intelligent decision system provides managers with data for better understanding and analysis. For unexecutable decisions, the intelligent decision system provides managers with unexecutable decision information, where the unexecutable decision information includes the potential risks of the decision identified by the intelligent decision system when analyzing the data and predicted future results, and provides decision recommendations based on these analyses.

[0050] The present invention obtains basic decision information, including decision keyword information, decision accuracy parameters and expert qualification parameters, through a decision information parameter acquisition module, and imports the acquired data into a preprocessing model through a decision information preprocessing module, and obtains the decision keyword matching degree and the corresponding decision accuracy coefficient, the historical decision accuracy coefficient and the expert qualification coefficient through processing, and these data can intuitively reflect the decision information and system capabilities; combined with the decision keyword matching degree and the corresponding decision accuracy coefficient, the historical decision accuracy coefficient and the expert qualification coefficient, the data is transmitted to the decision information analysis module, and further analyzed to obtain the decision comprehensive evaluation index, and the decision comprehensive evaluation index is judged to evaluate whether the decision to be executed is an executable decision, so as to provide convenience for users in the decision-making process. The present invention realizes intelligent processing of complex decision-making problems by integrating multiple modules such as data collection, processing, analysis and decision execution, and provides comprehensive suggestions for user decision-making by establishing a large model and combining algorithms such as machine learning.

[0051] Secondly: In the drawings of the embodiments disclosed in the present invention, only the structures related to the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other;

[0052] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent decision-making system based on a large model, comprising a decision scenario creation module, a decision information parameter acquisition module, a decision information preprocessing module, a decision information analysis module, a decision judgment module and a human-computer interaction module, characterized in that: The decision scenario creation module is used to create a scenario for decision making. The user operates in the system user interface, and the user selects a desired scenario template from the preset scenario templates, and matches the business in the selected scenario template with the user's goal or standard, and the user's goal or standard includes the solution direction of the corresponding decision problem; The decision information parameter acquisition module is used to acquire information of the text to be decided, including a decision keyword parameter acquisition unit, a decision accuracy parameter acquisition unit and an expert qualification parameter acquisition unit, and transmits the collected data to the decision information preprocessing module; The decision information preprocessing module includes a decision keyword preprocessing unit, a decision accuracy preprocessing unit and an expert qualification preprocessing unit, which are used to preprocess the data transmitted by the decision information parameter acquisition module and transmit the processing results to the decision information analysis module; The decision keyword preprocessing unit is used to establish a decision keyword preprocessing model, import the keyword parameters transmitted by the decision information parameter acquisition module into the decision keyword preprocessing module model, and calculate the keyword matching degree. The specific analysis formula is: , get the matching degree between the uth key word of the decision information and its corresponding historical key word, denoted as ζ u , where p u i It is represented by the i-th keyword in the u-th decision-making information, where u=1,2,…,z, i=1,2,…,n,q f v It is represented as the vth key word in the fth corresponding historical decision information, where f = 1, 2, ..., t, v = 1, 2, ..., h. According to the matching degree between the key words of the information to be decided and its corresponding historical key words, if the matching degree between a key word of the information to be decided and its corresponding historical key words is less than the preset keyword matching degree threshold, it indicates that the keyword information to be decided is an irrelevant keyword; If the matching degree of a key word to be decided and its corresponding historical key word is greater than or equal to the preset keyword matching degree threshold, it indicates that the key word information to be decided is a relevant keyword, and the historical decision information corresponding to the relevant key word to be decided is filtered and counted, and recorded as relevant historical decision information; The relevant historical decision information includes the number of historical decision information corresponding to it, the number of historical successful decisions, and the number of historical decisions that meet expectations, which are recorded as Ch, Vh, and Qh, respectively. The accuracy and precision of the relevant historical decisions are calculated. The calculation formula for the accuracy of the relevant historical decisions is: , the calculation formula of relevant historical decision accuracy is: ; Import the accuracy and precision of relevant historical decisions into the specific analysis formula of the relevant decision accuracy coefficient: ,in represents the accuracy of historical decisions related to the standard, represents the accuracy of the standard relevant historical decision, α1 and α2 represent the influencing factors of the accuracy and precision of the relevant historical decision, respectively; The decision accuracy preprocessing unit is used to establish a decision accuracy preprocessing model, import the decision accuracy parameters transmitted by the decision information parameter acquisition module into the decision accuracy preprocessing model, and calculate the decision accuracy coefficient. The decision accuracy coefficient includes historical decision accuracy and precision. The historical decision accuracy calculation formula is: , where Vz m represents the number of successful decisions corresponding to the mth decision, Cz represents the total number of decisions, and the calculation formula for historical decision accuracy is: , where Qz m represents the expected number of decisions corresponding to the mth decision; Import the historical decision accuracy and historical decision precision into the specific analysis formula of the decision accuracy coefficient: , where ω m Expressed as the decision accuracy coefficient corresponding to the mth decision, ω 1m and ω 2m They represent the historical decision accuracy and historical decision precision, ω1´ and ω2´ represent the standard decision accuracy and standard decision precision, β1 and β2 represent the influencing factors of historical decision accuracy and precision, respectively; The expert qualification preprocessing unit is used to establish an expert qualification preprocessing model, import the expert qualification parameters transmitted by the decision information parameter acquisition module into the expert qualification preprocessing model, and calculate the expert qualification coefficient. The expert qualification coefficient includes a probability index and an expert decision recall rate. The calculation formula of the probability index is: , where P m represents the probability that the system finds relevant knowledge in the expert knowledge base in the mth decision, Cw m represents the capacity of the expert knowledge base in the mth decision, PI represents the knowledge capacity required for decision making, and P M represents the standard probability of finding relevant knowledge, Z m represents the probability index of the success of the mth decision, k represents the total number of various decisions, and the calculation formula of the expert decision recall rate is: , where Rw represents the number of expert recalls, and the probability index and expert decision recall rate are introduced into the expert qualification coefficient. The specific analysis formula is: , where ζ´ represents the standard expert recall rate and ζ represents the expert seniority coefficient; The decision information analysis module is used to analyze the data of the decision information preprocessing module, establish a decision comprehensive evaluation model, import relevant decision accuracy coefficients and decision accuracy coefficients into the decision comprehensive evaluation model, and calculate the decision comprehensive evaluation index; The specific analysis formula of the decision comprehensive evaluation index is: , where ω m It is represented as the decision accuracy coefficient corresponding to the m-th decision, ζ represents the expert qualification coefficient, Represents the relevant decision accuracy coefficient. The higher the decision accuracy, the higher the probability that the system will find relevant knowledge in the knowledge base, and the greater the decision comprehensive evaluation index; The decision judgment module is used to establish a preset value of the decision comprehensive evaluation index, make a judgment based on the preset value of the decision comprehensive evaluation index and the decision comprehensive evaluation index, and issue a warning signal according to the judgment result; The human-computer interaction module is used to send signals for unexecutable decisions. The intelligent decision-making system transmits the judgment results and warning signals in the decision judgment module to the management control panel through IDSS, and transmits unexecutable decision information to the management through the control panel.

2. The intelligent decision-making system based on a large model according to claim 1, characterized in that: The decision keyword parameter acquisition unit acquires each key word in the information to be decided, forms a key word set of the information to be decided, extracts the key words and their corresponding historical key word sets, analyzes the matching degree between the key words to be decided and their corresponding historical key words, and the key word set is recorded as p u i ={p u 1,p u 2,...,p u i ,...,p u n }, where p u i It is represented as the i-th keyword in the u-th decision-making information, where u=1,2,…,z, i=1,2,…,n, and the historical decision key word set corresponding to the decision-making information stored in the decision storage database is extracted, and the corresponding historical decision key word set is recorded as q f v ={q f 1,q f 2,...,q f v ,...,q f h }, where q f v It is represented as the vth key word in the fth corresponding historical decision information, where f = 1, 2, ..., t, v = 1, 2, ..., h.

3. The intelligent decision-making system based on a large model according to claim 2, characterized in that: The decision accuracy parameter acquisition unit acquires the decision accuracy parameters corresponding to the historical decisions, the decision accuracy parameters including the number of successful decisions, the expected number of decisions and the total number of decisions, the number of successful decisions and the total number of decisions are denoted as Vz and Cz respectively; the decision is judged whether it meets expectations according to the pre-set goals or standards, and the expected number of decisions is counted and denoted as Qz.

4. The intelligent decision-making system based on a large model according to claim 1, characterized in that: The expert qualification parameter acquisition unit acquires expert qualification parameters, which include the capacity of the expert knowledge base and the number of expert recalls, which are denoted as Cw and Rw respectively.