Group decision model based on enterprise management preference relation relative projection

Abstract

The application relates to the technical field of group decision-making, and discloses a group decision-making model based on enterprise management and preference relation relative projection, which comprises a data acquisition and initialization module, a group synthesis and consensus evaluation module, a double-drive dynamic weight calculation module, an expert mutual evaluation relation updating module and an iterative convergence and decision generation module, and can acquire an initial preference relation matrix and a mutual evaluation matrix; a group synthesis preference matrix is constructed based on current expert weights, and a global consensus level is analyzed; the objective consensus contribution degree and the subjective social trust degree of the experts are calculated, then the expert weights are updated by fusion, and the mutual evaluation matrix is updated; iteration is performed until convergence, a decision scheme ranking and an expert weight distribution report are generated based on the final group synthesis preference matrix; and the group decision-making efficiency based on enterprise management and preference relation relative projection can be improved.

Description

Technical Field

[0001] This invention relates to the field of group decision-making technology, and in particular to a group decision-making model based on the relative projection of preference relationships in enterprise management. Background Technology

[0002] Existing group decision-making models in enterprise management scenarios often rely on static expert weights or single subjective peer reviews and single objective consistency calculations, making it difficult to balance the objectivity of opinions with the trust relationship within the group. Traditional methods only complete preference integration through simple weighting or fixed thresholds, without quantitatively matching the vector structure of individual expert preferences with the overall group preferences. This results in a crude consensus assessment that is easily influenced by extreme opinions. At the same time, expert peer review networks are often based on initial fixed values ​​and cannot be dynamically adjusted during the decision-making process. This leads to a disconnect between the trust relationship and the actual consensus performance, resulting in an unreasonable weight allocation, slow convergence of group consensus, and insufficient stability and credibility of decision results. Consequently, these models cannot meet the demands of complex enterprise management decisions for efficiency, accuracy, and high consensus.

[0003] Current group decision-making technologies still have key shortcomings. First, they fail to achieve a dual-driven integration of objective consensus contribution and subjective social trust, with a single dimension for weight adjustment, making it difficult to balance data facts and team collaboration trust. Second, the peer review matrix lacks an adaptive update mechanism, failing to dynamically adjust trust scores based on real-time expert consensus performance, resulting in a rigid trust network that cannot reflect the true decision-making capabilities of experts. These shortcomings collectively lead to low efficiency in group decision-making, difficulty in rapidly improving the level of global consensus, and a lack of persuasiveness in the ranking of decision-making schemes. They cannot meet the needs of modern enterprise management for group decision-making with multiple experts, high collaboration, and strong objectivity. Therefore, how to improve the efficiency of group decision-making has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a group decision-making model based on the relative projection of preference relationships in enterprise management to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a group decision-making model based on the relative projection of preference relationships in enterprise management. The model comprises a data acquisition and initialization module, a group synthesis and consensus evaluation module, a dual-drive dynamic weight calculation module, an expert peer review relationship update module, and an iterative convergence and decision generation module, wherein: The data acquisition and initialization module is used to acquire the initial preference matrix of the expert group for the alternative solutions and the mutual evaluation matrix among the experts; The group integration and consensus assessment module is used to construct a group integration preference matrix and analyze the global consensus level based on the current expert weights and the initial preference relationship matrix. The dual-drive dynamic weight calculation module is used to calculate the objective consensus contribution of experts based on the relative projection of the group comprehensive preference matrix and the expert individual preference relationship, calculate the subjective social trust of experts based on the mutual evaluation matrix, and dynamically update the expert weight by integrating the objective consensus contribution and subjective social trust. The expert peer review relationship update module is used to update the peer review matrix based on the consensus performance of individual expert preferences and the group comprehensive preference matrix; The iterative convergence and decision generation module is used to repeatedly execute the group integration and consensus evaluation module to the expert peer review relationship update module until the global consensus level reaches a preset threshold or the expert weights converge, and generate a decision scheme ranking and expert weight allocation report based on the final group integration preference matrix.

[0006] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention is the first to deeply integrate the relative projection of preference relation row vectors with a lightweight expert peer review network, constructing a dynamic expert weight allocation model driven by both objective consensus contribution and subjective social trust. The relative projection algorithm accurately quantifies the structural consistency between individual expert preferences and overall group preferences, and the peer review network obtains subjective trust scores from experts. Through normalization and linear convex combination, the weights are dynamically updated, ensuring both the objectivity and fairness of weight allocation and taking into account the expert collaboration and trust relationships in enterprise management scenarios. This effectively solves the problems of traditional models having single weights, static allocation, and susceptibility to subjective bias, significantly improving the rationality and stability of group decision-making.

[0007] 2. This invention proposes an adaptive update method for the mutual evaluation matrix based on individual consensus performance values. Using real-time expert consensus performance as a basis, the trust score is dynamically adjusted through a consensus performance benchmark value and a learning rate parameter. A truncation process ensures numerical compliance, allowing the mutual evaluation relationship to iteratively optimize throughout the decision-making process. This continuously increases the group trust of experts with high consensus performance, weakens the influence of experts with low consensus performance, and promotes rapid convergence of the global consensus level. Simultaneously, it simplifies the group decision-making calculation process, significantly improving the efficiency of group decision-making involving multiple solutions and experts in enterprise management scenarios. The output solution ranking and weight allocation reports are more persuasive and traceable. Attached Figure Description

[0008] Figure 1 This is a module architecture diagram of a group decision-making model based on the relative projection of preference relationships in enterprise management, provided in an embodiment of the present invention. The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0009] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments belong to some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0010] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “said” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0011] Depending on the context, the word "if" or "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0012] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0013] In practice, the server-side equipment deployed in the group decision-making model based on the relative projection of enterprise management preferences may consist of one or more devices. This group decision-making model based on the relative projection of enterprise management preferences can be implemented as: a business instance, a virtual machine, or hardware devices. For example, this group decision-making model based on the relative projection of enterprise management preferences can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, this group decision-making model based on the relative projection of enterprise management preferences can be understood as software deployed on a cloud node, used to provide the group decision-making model based on the relative projection of enterprise management preferences to each user terminal. Alternatively, this group decision-making model based on the relative projection of enterprise management preferences can also be implemented as a virtual machine deployed on one or more devices in a cloud node. This virtual machine contains application software for managing each user terminal. Alternatively, this group decision-making model based on the relative projection of enterprise management preferences can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more hardware devices configured to provide the group decision-making model based on the relative projection of enterprise management preferences to each user terminal.

[0014] In terms of implementation, the group decision-making model based on the relative projection of preference relationships in enterprise management and the user terminal are mutually adapted. That is, if the group decision-making model based on the relative projection of preference relationships in enterprise management is implemented as an application installed on a cloud service platform, then the user terminal is implemented as a client that establishes a communication connection with the application; or if the group decision-making model based on the relative projection of preference relationships in enterprise management is implemented as a website, then the user terminal is implemented as a webpage; or if the group decision-making model based on the relative projection of preference relationships in enterprise management is implemented as a cloud service platform, then the user terminal is implemented as a mini-program in an instant messaging application.

[0015] like Figure 1 The diagram shown is a system architecture diagram of a group decision-making model based on the relative projection of preference relationships in enterprise management, provided by an embodiment of the present invention.

[0016] The group decision-making model 100 based on the relative projection of preference relationships in enterprise management, as described in this invention, can be set up in a cloud server. In terms of implementation, it can be implemented as one or more service devices, or as an application installed in the cloud (e.g., a mobile service operator's server, server cluster, etc.), or it can be developed as a website. Depending on the implemented functions, the group decision-making model 100 based on the relative projection of preference relationships in enterprise management may include a data acquisition and initialization module 101, a group synthesis and consensus evaluation module 102, a dual-drive dynamic weight calculation module 103, an expert peer review relationship update module 104, and an iterative convergence and decision generation module 105. The modules described in this invention can also be called units, referring to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0017] In this embodiment of the invention, in the group decision-making model based on the relative projection of preference relationships in enterprise management, each of the above modules can be implemented independently and called by other modules. This calling can be understood as a module connecting to multiple modules of another type and providing corresponding services to those connected modules. In the group decision-making model based on the relative projection of preference relationships in enterprise management provided by this embodiment of the invention, the applicability of the group decision-making model architecture based on the relative projection of preference relationships in enterprise management can be adjusted by adding modules and directly calling them without modifying the program code, achieving cluster-based horizontal expansion, so as to achieve the purpose of quickly and flexibly expanding the group decision-making model based on the relative projection of preference relationships in enterprise management. In practical applications, the above modules can be set in the same device or different devices, or they can be set in a virtual device, such as a service instance in a cloud server.

[0018] The following describes, with reference to specific embodiments, each component and specific workflow of the group decision-making model based on the relative projection of preference relationships in enterprise management: The data acquisition and initialization module is used to acquire the initial preference matrix of the expert group for the alternative solutions and the mutual evaluation matrix among the experts; In this embodiment of the invention, the step of obtaining the initial preference matrix of the expert group for the alternative solutions and the mutual evaluation matrix among the experts includes: The system receives information on the expert group and alternative solutions respectively, in order to obtain the expert set and the alternative solution set. Based on the expert set and the alternative solution set, the preference degree of each expert for each pair of alternative solutions is analyzed to construct the initial preference relationship matrix of each expert; Based on the expert set, calculate the initial trust score of each expert for other experts, and construct an initial mutual evaluation matrix.

[0019] It should be noted that receiving expert group information and alternative solution information refers to receiving input parameters through a graphical user interface or application programming interface. These parameters include the number of experts and the number of alternative solutions, and the input format is subject to basic validation to ensure the integrity of the data structure for subsequent processing.

[0020] It should be noted that the expert set is the collection of decision-making participants composed of all individual experts, and it is the basic input of the group decision-making model; the alternative set is the collection of all objects to be decided, and it is the object on which experts make preference judgments.

[0021] It should be noted that the initial preference matrix is ​​constructed based on each expert's preference judgments for all pairs of alternatives; the specific process is as follows: For each expert, obtain their preference judgments for alternative x. i Compared to scheme x j The preference degree is a value between 0 and 1; the preference degree is required to satisfy an additive reciprocal relationship, that is, the sum of the preferences of the two options is equal to 1; arrange all the preferences degrees with the option index i as the row and j as the column to form an n×n matrix, which is the initial preference relationship matrix of the expert, where n is the number of alternative options.

[0022] Furthermore, the additive reciprocal relationship ensures the logical consistency of preference judgments. Expert e h It is believed that scheme x i For x j The preference score is 0.5, indicating equal importance, while a score greater than 0.5 indicates a preference for x. i A value less than 0.5 indicates a preference for x. j .

[0023] It should be noted that the initial preference relation matrix is ​​a mathematical object representing all the preference information of individual experts for pairwise comparisons of all options, and it is the original data basis for subsequent group preference integration and consensus calculation.

[0024] It should be noted that the initial peer review matrix is ​​constructed based on each expert's assessment of the trust level of each of the other experts. The specific process is as follows: obtain the initial trust rating of each expert for the other experts, which is a value between 0 and 1; usually, no expert evaluates themselves; the self-evaluation value in the matrix is ​​1, and arrange all the ratings by the evaluator index as the row and the evaluated index as the column to form a matrix, which is the initial peer review matrix.

[0025] Furthermore, the trust score is the degree to which experts believe the judgments of other experts are reliable or their professional authority in the decision-making process, with 0 indicating no trust and 1 indicating complete trust.

[0026] It should be noted that the initial peer review matrix is ​​a mathematical object representing the initial social trust network structure within the expert group. It reflects the subjective evaluation relationship among experts and is the data source for the subjective driving part in the subsequent dynamic weight calculation.

[0027] The group integration and consensus assessment module is used to construct a group integration preference matrix and analyze the global consensus level based on the current expert weights and the initial preference relationship matrix. In this embodiment of the invention, the step of constructing a group comprehensive preference matrix and analyzing the global consensus level based on the current expert weights and the initial preference relationship matrix includes: Based on the expert weights of the current iteration step and the initial set of preference relation matrices, the individual preferences of experts are aggregated by a weighted average operator to construct the group comprehensive preference matrix for this iteration. Based on the group's comprehensive preference matrix and the initial preference relationship matrix, the global consensus level of the expert group at the current iteration step is calculated using a consensus metric algorithm.

[0028] It should be noted that the weighted average operator aggregation refers to the linear weighted summation of corresponding matrix elements in the initial preference relation matrix of all experts, and its mathematical expression is as follows: ; In the formula, Indicates the group's opinion on the proposal Compared to the plan Overall preference Expert e h The weights at the t-th iteration, and satisfying , Experts The proposed solution for preference This represents the total number of experts. This indicates the expert sequence number index. This is the current iteration step.

[0029] Furthermore, the weighted average operator aggregates the preferences of each expert for a specific pair of options by linearly combining them based on the current influence of each expert in the decision-making process, thereby forming a unified preference matrix that represents the common opinion of the entire expert group. The calculation is iterated once in the first calculation.

[0030] It should be noted that the group comprehensive preference matrix is ​​an n×n matrix, and the value of each element is also between [0,1], and it also satisfies the additive reciprocal relationship.

[0031] It should be noted that the global consensus level ranges from 0 to 1. A global consensus level of 1 indicates that the individual preferences of all experts are completely consistent with the overall group preferences, and a complete consensus has been reached. The closer the global consensus level is to 1, the higher the degree of group consensus. This indicator is used to quantify the consistency and convergence of the expert group's opinions in the current iteration round, and is the core basis for judging whether the iteration process has terminated.

[0032] In this embodiment of the invention, the mathematical expression of the consensus metric algorithm is as follows: ; In the formula, Indicates the first The level of global consensus at the next iteration The number of alternative options, For experts Preference matrix elements, Indicates the group's opinion on the proposal Compared to the plan Overall preference Indicates to All expert preference matrices and all scheme pairs Perform a traversal and summation. This represents the total number of experts. This indicates the expert sequence number index. and This is the scheme index.

[0033] The dual-drive dynamic weight calculation module is used to calculate the objective consensus contribution of experts based on the relative projection of the group comprehensive preference matrix and the expert individual preference relationship, calculate the subjective social trust of experts based on the mutual evaluation matrix, and dynamically update the expert weight by integrating the objective consensus contribution and subjective social trust. In this embodiment of the invention, the step of calculating the objective consensus contribution of experts based on the relative projection of the group comprehensive preference matrix and the relationship between individual expert preferences includes: For each expert, iterate through each alternative and extract the data of the current option row in the initial preference relation matrix of each expert to obtain the preference relation row vector; Extract the data from the row containing the current scheme in the group comprehensive preference matrix to obtain the group preference row vector; Based on the preference relationship row vector and the group preference row vector, the relative projection value of the expert on each scheme is calculated by the relative projection algorithm; The projection similarity of the experts is obtained by taking the arithmetic mean of their relative projection values ​​across all proposals. Based on the projection similarity, the objective consensus contribution of experts is calculated using an exponential mapping function.

[0034] It should be noted that the preference relation row vector representation is what experts believe the solution to. The sequence of preference levels relative to all other options; the group preference row vector represents the overall expert group's preference for the option in the current iteration. The sequence of collective preference levels relative to all other options.

[0035] It should be noted that the mathematical expression for the relative projection algorithm is as follows: ; In the formula, Experts Regarding the plan Preference vectors and group pair schemes The relative projection values ​​between preference vectors, It represents the similarity between the row vectors of preference relationships and the row vectors of group preferences. It is the square of the magnitude of the group preference row vector. For experts Preference matrix elements, Indicates the group's opinion on the proposal Compared to the plan Overall preference The number of alternative options, and For scheme indexing, This is the current iteration step.

[0036] Furthermore, the relative projection value reflects the degree of consistency between individual expert opinions and group opinions on the preference structure of a single option. The closer this value is to 1, the stronger the expert's opinion. Regarding the plan The preference structure is highly consistent with the group; if the value is less than 1, it indicates that there is a bias; the algorithm quantifies the degree of consistency into a unitless scalar through vector dot product and normalization.

[0037] It should be noted that the mathematical expression for the exponential mapping function is as follows: ; In the formula, Experts In the Objective consensus contribution at each iteration Indicates projection similarity, This is the current iteration step.

[0038] Furthermore, when the projection similarity of an expert is equal to 1, their objective consensus contribution is 1; when the projection similarity is less than 1, the objective consensus contribution decreases exponentially as the projection similarity decreases, which significantly reduces the contribution weight of experts with low consensus; when it is greater than 1, the objective consensus contribution increases slowly, avoiding excessive rewards for experts with slightly higher consensus, thus maintaining the stability of weight adjustment.

[0039] It should be noted that a higher objective consensus contribution value indicates a higher consistency between the expert's opinion and the overall group opinion, and therefore should be given higher weight in the objective driving aspect.

[0040] In this embodiment of the invention, the calculation of the expert's subjective social trust level based on the mutual evaluation matrix includes: For each expert, extract the trust ratings of all other experts (excluding the expert's self-rating) from the peer review matrix of the current iteration step to obtain the trust rating vector. Based on the trust scoring vector, the subjective social trust level obtained by the expert in the current round is calculated.

[0041] It should be noted that extracting the trust rating vector refers to selecting the corresponding person being evaluated as an expert from the peer review matrix. All ratings, excluding expert ratings. Self-evaluation; self-rating scores were set to a fixed value of 1 and removed to avoid interference from self-evaluation in the measurement of group trust.

[0042] Furthermore, the trust scoring vector is a vector that contains... A vector of n elements, representing the result at the end of the previous iteration, excluding the expert. Besides oneself, others in the group Experts to experts A set of trust assessment values ​​for the decision-making ability or reliability of experts; this vector directly reflects the level of trust in the expert's decision-making ability or reliability. The position and reputation gained within a group's social network.

[0043] It should be noted that the mathematical expression used to calculate the subjective social trust level obtained by the expert in the current round is as follows: ; in, Experts Subjective social trust level at the t-th iteration This represents the total number of experts. For the evaluator index, This indicates the experts in the peer review matrix of the previous iteration. For experts Trust rating For the current iteration step, This is the previous iteration step.

[0044] It should be noted that a subjective social trust level of 1 means that all other experts trust the expert. Complete trust; a subjective social trust level of 0 indicates that all other experts trust the expert. Complete distrust.

[0045] In this embodiment of the invention, the step of dynamically updating expert weights by fusing the objective consensus contribution and subjective social trust includes: Normalize the objective consensus contribution vector and the subjective social trust vector of all experts in the current iteration step to obtain the normalized objective contribution vector and the normalized social trust vector. By using preset objective factor weight parameters, a linear convex combination is performed on the normalized objective contribution vector and the normalized social trust vector to obtain the expert weight vector for the next iteration.

[0046] It should be noted that normalization refers to processing the elements in each vector so that their sum is 1. For the objective consensus contribution vector, the normalization formula is as follows: ;

[0047] The normalization formula for the subjective social trust vector is as follows: ;

[0048] In the formula, Experts The normalized objective contribution vector, Experts In the Objective consensus contribution at each iteration This represents the total number of experts. For the normalization process expert sequence index. This indicates the expert sequence number index. Experts The subjective social trust vector, Experts In the Subjective social trust level at the next iteration This is the current iteration step.

[0049] Furthermore, the normalized objective contribution reflects the experts'... The proportion of experts whose opinions and group consensus are objectively consistent; normalized social trust reflects the proportion of experts' opinions and group consensus. The proportion of subjective reputation gained in a group among all experts.

[0050] It should be noted that the mathematical expression for a linear convex combination is as follows: ; In the formula, As preset objective factor weights, Experts The normalized objective contribution vector, Experts The subjective social trust vector, This indicates the expert sequence number index. For experts In the New weights in the next iteration This is the current iteration step.

[0051] It should be noted that the preset objective factor weight is a key adjustment parameter of the model. It is a fixed parameter set by management based on the specific decision-making scenario, organizational culture, and the emphasis on objective facts and teamwork. Its value range is set to [0,1]. When the preset objective factor weight is equal to 1, the expert weight is entirely determined by the contribution of objective consensus, and the model degenerates into a purely objective model, focusing on the consistency between expert opinions and group consensus. When the preset objective factor weight is equal to 0, the expert weight is entirely determined by subjective social trust, and the model degenerates into a purely peer-evaluation model, focusing on the reputation of experts in the group. When the preset objective factor weight is between 0 and 1, the model achieves a dual-driven integration of objective and subjective factors. By adjusting the value of the preset objective factor weight, it can adapt to different decision-making scenarios and corporate cultures, achieving a balance between respecting objective facts and emphasizing teamwork.

[0052] The expert peer review relationship update module is used to update the peer review matrix based on the consensus performance of individual expert preferences and the group comprehensive preference matrix; In this embodiment of the invention, updating the peer review matrix based on the consensus performance of individual expert preferences and the group's overall preference matrix includes: For each expert, their individual consensus performance value in the current round is calculated based on their preference relationship matrix and the group comprehensive preference matrix at the current iteration step. For each pair of expert evaluation relationships, the trust score in the previous round of mutual evaluation matrix is ​​adjusted based on the individual consensus performance value, the preset consensus performance benchmark value, and the preset learning rate parameter to obtain the trust score to be updated. The trust score to be updated is truncated to ensure that it falls within the preset score range [0,1], so as to generate the updated mutual evaluation matrix elements for the current iteration step.

[0053] It should be noted that the mathematical expression for calculating an individual consensus performance value is as follows: ; In the formula, Experts In the Individual consensus performance values ​​in round iterations For experts Preference matrix elements, Indicates the group's opinion on the proposal Compared to the plan Overall preference The number of alternative options, and For scheme indexing, This is the current iteration step.

[0054] It should be noted that the mathematical expression for adjusting the trust score is as follows: ; In the formula, This indicates a trust score that needs updating. Indicates the experts in the previous iteration For experts Trust rating For learning rate, This serves as a benchmark for consensus performance. It is a sign function, with a value of +1 when x>0, a value of -1 when x<0, and a value of 0 when x=0. The learning rate is 0.02, and the consensus performance benchmark is 0.8.

[0055] Furthermore, adjusting trust scores essentially involves adjusting them based on expert opinions. The consensus reached in this round is used to fine-tune other experts' level of trust in it.

[0056] It should be noted that the truncation operation is defined as follows: First, determine whether the trust score to be updated is less than 0. If so, set it to 0, indicating the boundary of complete distrust. Otherwise, check if it is greater than 1. If it is, set it to 1 to indicate the boundary of complete trust. If the trust score to be updated is already in the range [0,1], then keep the value directly.

[0057] It should be noted that the updated mutual evaluation matrix elements constitute the mutual evaluation matrix of the current iteration step t. This matrix reflects the latest state of trust relationships within the expert group after this round of consensus process. Experts who continuously output highly consensus opinions will gradually accumulate higher group trust under this mechanism, thus occupying a more favorable position in the subsequent subjective social trust calculation. This simulates the social process of building reputation based on ability and performance in reality.

[0058] The iterative convergence and decision generation module is used to repeatedly execute the group integration and consensus evaluation module to the expert peer review relationship update module until the global consensus level reaches a preset threshold or the expert weights converge, and generate a decision scheme ranking and expert weight allocation report based on the final group integration preference matrix.

[0059] In this embodiment of the invention, the step of repeatedly executing the group comprehensive and consensus evaluation module to the expert peer review relationship update module until the global consensus level reaches a preset threshold or the expert weights converge, and generating a decision scheme ranking and expert weight allocation report based on the final group comprehensive preference matrix, includes: Determine whether the global consensus level of the current iteration step meets the expert weight convergence condition to obtain the convergence judgment result; When the convergence judgment result is convergence, the iteration process is terminated, and the group comprehensive preference matrix corresponding to the current iteration step is taken as the final group comprehensive preference matrix. At the same time, the expert weight vector corresponding to the current iteration step is taken as the final expert weight vector. Based on the final group comprehensive preference matrix, calculate the comprehensive score of each alternative, and sort all alternatives in descending order according to the comprehensive score to generate a decision scheme ranking; Based on the final expert weight vector, an expert weight allocation report is generated.

[0060] It should be noted that the overall score of the calculation scheme is essentially the arithmetic mean of all elements in the i-th row of the matrix.

[0061] It should be noted that after the decision options are ranked, the option with the highest ranking is the optimal choice with the strongest group consensus, and the subsequent options are ranked in order.

[0062] It should be noted that the expert weight allocation report is a formatted output generated based on the final expert weight vector, presented in the form of lists, charts, etc. It clearly lists each expert and their corresponding final weight, intuitively showing the final influence and contribution ratio of each expert in the decision-making group after the entire dynamic consensus process, providing key evidence for the transparency, interpretability and subsequent accountability of the decision-making process.

[0063] In this embodiment of the invention, determining whether the global consensus level of the current iteration step meets the expert weight convergence condition to obtain a convergence determination result includes: Determine whether the global consensus level of the current iteration step has reached or exceeded the preset consensus threshold. At the same time, calculate whether the change of the expert weight vector in the (t+1)th iteration relative to the expert weight vector in the tth iteration is less than the preset weight convergence threshold in each iteration to obtain the convergence judgment result.

[0064] It should be noted that the convergence judgment includes two independent and parallel evaluation conditions. The first condition is that the global consensus level is met, and the preset consensus threshold is 0.9. The second condition is that the expert weight changes are stable. This condition is usually determined by calculating the difference between the weight vectors between two consecutive iterations, specifically by calculating the Euclidean norm of the two weight vectors. The preset weight convergence threshold is 0.001.

[0065] It should be noted that the convergence judgment result is based on the simultaneous judgment of the above two conditions. As long as either condition is true, it is considered that the convergence state has been reached.

[0066] The details of the exemplary embodiments are provided, and the invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the invention.

[0067] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0068] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A group decision model based on the relative projection of preference relations of enterprise management, characterized in that, The model includes a data acquisition and initialization module, a group synthesis and consensus evaluation module, a dual-drive dynamic weight calculation module, an expert peer review relationship update module, and an iterative convergence and decision generation module, wherein: The data acquisition and initialization module is used to acquire the initial preference matrix of the expert group for the alternative solutions and the mutual evaluation matrix among the experts; The group integration and consensus assessment module is used to construct a group integration preference matrix based on the current expert weights and the initial preference relationship matrix, and parse out the global consensus level. The dual-drive dynamic weight calculation module is used to calculate the objective consensus contribution of experts based on the relative projection of the group comprehensive preference matrix and the expert individual preference relationship, calculate the subjective social trust of experts based on the mutual evaluation matrix, and dynamically update the expert weight by integrating the objective consensus contribution and subjective social trust. The expert peer review relationship update module is used to update the peer review matrix based on the consensus performance of individual expert preferences and the group comprehensive preference matrix; The iterative convergence and decision generation module is used to repeatedly execute the group integration and consensus evaluation module to the expert peer review relationship update module until the global consensus level reaches a preset threshold or the expert weights converge, and generate a decision scheme ranking and expert weight allocation report based on the final group integration preference matrix.

2. The group decision model based on enterprise management preference relation relative projection of claim 1, wherein, The method for obtaining the initial preference matrix of the expert group on the alternative solutions and the mutual evaluation matrix among experts includes: The system receives information on the expert group and alternative solutions respectively, in order to obtain the expert set and the alternative solution set. Based on the expert set and the alternative solution set, the preference degree of each expert for each pair of alternative solutions is analyzed to construct the initial preference relationship matrix of each expert; Based on the expert set, calculate the initial trust score of each expert for other experts, and construct an initial mutual evaluation matrix.

3. The group decision-making model based on the relative projection of preference relationships in enterprise management as described in claim 1, characterized in that, The process of constructing a group comprehensive preference matrix and analyzing the global consensus level based on the current expert weights and the initial preference relationship matrix includes: Based on the expert weights of the current iteration step and the initial set of preference relation matrices, the individual preferences of experts are aggregated by a weighted average operator to construct the group comprehensive preference matrix for this iteration. Based on the group's comprehensive preference matrix and the initial preference relationship matrix, the global consensus level of the expert group at the current iteration step is calculated using a consensus metric algorithm.

4. The group decision-making model based on relative projection of preference relationships in enterprise management as described in claim 3, characterized in that, The mathematical expression for the consensus metric algorithm is as follows: ; In the formula, Indicates the first The level of global consensus at the next iteration The number of alternative options, For experts Preference matrix elements, Indicates the group's opinion on the proposal Compared to the plan Overall preference Indicates to All expert preference matrices and all scheme pairs Perform a traversal and summation. This represents the total number of experts. This indicates the expert sequence number index. and This is the scheme index.

5. The group decision-making model based on the relative projection of preference relationships in enterprise management as described in claim 1, characterized in that, The method for calculating the objective consensus contribution of experts based on the relative projection of the group's comprehensive preference matrix and the relationship between individual expert preferences includes: For each expert, iterate through each alternative and extract the data of the current option row in the initial preference relation matrix of each expert to obtain the preference relation row vector; Extract the data from the row containing the current scheme in the group comprehensive preference matrix to obtain the group preference row vector; Based on the preference relationship row vector and the group preference row vector, the relative projection value of the expert on each scheme is calculated by the relative projection algorithm; The projection similarity of the experts is obtained by taking the arithmetic mean of their relative projection values ​​across all proposals. Based on the projection similarity, the objective consensus contribution of experts is calculated using an exponential mapping function.

6. The group decision-making model based on the relative projection of preference relationships in enterprise management as described in claim 1, characterized in that, The calculation of experts' subjective social trust based on the mutual evaluation matrix includes: For each expert, extract the trust ratings of all other experts (excluding the expert's self-rating) from the peer review matrix of the current iteration step to obtain the trust rating vector. Based on the trust scoring vector, the subjective social trust level obtained by the expert in the current round is calculated.

7. The group decision-making model based on the relative projection of preference relationships in enterprise management as described in claim 1, characterized in that, The method of dynamically updating expert weights by integrating objective consensus contribution and subjective social trust includes: Normalize the objective consensus contribution vector and the subjective social trust vector of all experts in the current iteration step to obtain the normalized objective contribution vector and the normalized social trust vector. By using preset objective factor weight parameters, a linear convex combination is performed on the normalized objective contribution vector and the normalized social trust vector to obtain the expert weight vector for the next iteration.

8. The group decision-making model based on the relative projection of preference relationships in enterprise management as described in claim 1, characterized in that, The method for updating the peer review matrix based on the consensus performance of individual expert preferences and the overall group preference matrix includes: For each expert, their individual consensus performance value in the current round is calculated based on their preference relationship matrix and the group comprehensive preference matrix at the current iteration step. For each pair of expert evaluation relationships, the trust score in the previous round of mutual evaluation matrix is ​​adjusted based on the individual consensus performance value, the preset consensus performance benchmark value, and the preset learning rate parameter to obtain the trust score to be updated. The trust score to be updated is truncated to ensure that it falls within the preset score range [0,1], so as to generate the updated mutual evaluation matrix elements for the current iteration step t.

9. The group decision-making model based on the relative projection of preference relationships in enterprise management as described in claim 1, characterized in that, The process of repeatedly executing the group comprehensive and consensus assessment module to the expert peer review relationship update module until the global consensus level reaches a preset threshold or the expert weights converge, and generating a decision scheme ranking and expert weight allocation report based on the final group comprehensive preference matrix, includes: Determine whether the global consensus level of the current iteration step meets the expert weight convergence condition to obtain the convergence judgment result; When the convergence judgment result is convergence, the iteration process is terminated, and the group comprehensive preference matrix corresponding to the current iteration step is taken as the final group comprehensive preference matrix. At the same time, the expert weight vector corresponding to the current iteration step is taken as the final expert weight vector. Based on the final group comprehensive preference matrix, calculate the comprehensive score of each alternative, and sort all alternatives in descending order according to the comprehensive score to generate a decision scheme ranking; Based on the final expert weight vector, an expert weight allocation report is generated.

10. The group decision-making model based on the relative projection of preference relationships in enterprise management as described in claim 9, characterized in that, The determination of whether the global consensus level of the current iteration step meets the expert weight convergence condition, in order to obtain the convergence determination result, includes: Determine whether the global consensus level of the current iteration step has reached or exceeded the preset consensus threshold. At the same time, calculate whether the change of the expert weight vector in the (t+1)th iteration relative to the expert weight vector in the tth iteration is less than the preset weight convergence threshold in each iteration to obtain the convergence judgment result.

Images