Data allocation method and system based on affine VCG and original dual iteration

Abstract

This invention provides a data allocation method and system based on affine VCG and primal-dual iteration, relating to the technical field of resource allocation optimization and algorithm mechanism design. The method includes: S1, inputting data transaction parameters; S2, generating virtual participants to form an extended bidding set; S3, setting corresponding value adjustment coefficients for each participant in the extended bidding set; S4, executing multiple rounds of allocation iteration based on the extended bidding set and value adjustment coefficients until a clearing state is reached; S5, after reaching the clearing state, determining the final optimal allocation scheme and the final personalized payment chips for each winning bidder; S6, calculating and outputting the real payment chips for each winning bidder based on the final personalized payment chips and value adjustment coefficients. This invention's method, by introducing virtual participants to change the competitive structure and directly amplifying the payment amount through adjustment coefficients, ensures a significant increase in the seller's expected return from an algorithmic perspective, enhancing the participation willingness of resource suppliers and the stability of the system ecosystem.

Description

Technical Field

[0001] This invention relates to the field of resource allocation optimization and algorithm mechanism design technology, specifically to a data allocation method and system based on affine VCG and primal dual iteration. Background Technology

[0002] As the reform of the market-oriented allocation system for data elements continues to deepen, the paradigm for realizing the value of data elements has undergone a fundamental shift: from the traditional resale of original data ownership to intelligent value creation driven by large-scale artificial intelligence models.

[0003] Against this industry backdrop, two core contradictions have given rise to new transaction demands: On the one hand, the massive and rigid demand for high-quality labeled datasets in the training and iteration of large-scale artificial intelligence models is increasingly at odds with the persistently high costs of acquiring, cleaning, and labeling high-quality data; on the other hand, the same dataset can be applied by different market players to differentiated strategic scenarios such as model training, industry analysis, and business decision-making, and its value assessment is highly scenario-dependent and subject-specific. These factors have jointly driven the data element market, generating a rigid market demand for the same dataset to be configured with multiple buyers and multiple transaction modes.

[0004] Traditional data transaction pricing models, including fixed pricing and bilateral negotiation, are difficult to adapt to the complex valuation and resource allocation needs that are dynamic, multi-party, and highly differentiated: fixed pricing cannot cover the differentiated value assessments of the same data by different parties, which can easily lead to serious value mismatch; bilateral negotiation has inherent defects such as long transaction cycles, low efficiency, and poor standardization, and cannot support large-scale, multi-party synchronous transactions.

[0005] In the field of allocation theory, the Vickrey-Clark-Groves (VCG) mechanism, as a classic multi-item allocation mechanism, is considered a core theoretical benchmark for multi-agent resource allocation due to its excellent theoretical properties such as incentive compatibility (which can induce participating buyers to truthfully disclose their valuations of the target) and Pareto optimality. However, in the actual implementation scenarios of data element market transactions, the VCG mechanism faces insurmountable technical application bottlenecks and cannot adapt to the aforementioned multi-agent dataset transaction needs: First, the pricing logic and revenue guarantee capability of the VCG mechanism are highly dependent on a fully competitive market environment. However, in data transaction scenarios, the number of target buyers for specific vertical domain datasets is usually limited, and fully competitive conditions are extreme exceptions rather than the market norm. Directly applying the VCG mechanism will result in the inability to effectively guarantee the reasonable revenue of the trading platform, and may even lead to revenues significantly lower than the reasonable market level. Second, the implementation of existing VCG mechanisms mostly adopts a single-round sealed bidding model, and the bidding and pricing calculation process is opaque. This not only easily leads to widespread questioning of the fairness of the transaction by trading entities and significantly increases the costs of handling transaction disputes and system maintenance, but also seriously damages the industry credibility and user trust of the trading platform.

[0006] In summary, the current data element trading field urgently needs a new allocation method. This method should retain the excellent theoretical properties such as incentive compatibility and efficient resource allocation, while overcoming the two core technical bottlenecks of the existing VCG mechanism in data trading scenarios: difficulty in guaranteeing reasonable platform returns and insufficient transparency in the trading process. This is necessary to adapt to the multi-entity, multi-mode, and large-scale dataset trading allocation needs in the process of data element marketization. Summary of the Invention

[0007] To address the shortcomings of the existing technology, the present invention aims to provide a data allocation method and system based on affine VCG and primal dual iteration, thereby solving the technical problems of difficulty in guaranteeing seller revenue and lack of transparency in the allocation process in traditional VCG-like mechanisms in scenarios such as data transactions.

[0008] Specifically, on the one hand, the present invention provides a data allocation method based on affine VCG and primal dual iteration, which includes the following steps: S1. Input data transaction parameters: Receive the set of data sharing modes to choose from, the set of real participants, the bid valuation of each real participant for different sharing modes, additional secondary development intention options, and the upper limit of the number of successful bidders allowed by the data distributor to conduct secondary data development; S2. Generate Virtual Participants: Construct a set of virtual participants and set their bid valuations for each data sharing mode, forming a set that includes real participants. With virtual participants Extended bid set and a collection of data sharing models , where any participant Any data sharing mode ; S3. Set a value adjustment factor: to expand the bid set. Participants Set the corresponding value adjustment coefficient ; S4. Perform multi-round iterative allocation: based on an expanded bid set. With value adjustment coefficient The process involves multiple rounds of allocation iterations until a clearing state is reached. S5. Determine the final allocation: After the clearing state is reached, determine the final optimal allocation plan and the final personalized payment chips for each successful bidder; S6. Calculate the final payment: Based on the final personalized payment chips and value adjustment factor. Calculate and output the actual payment chips of each winning bidder.

[0009] Preferably, step S4 specifically includes the following sub-steps: S41. Initialize the iteration rounds The initial temporary stakes for each of the current data-sharing models will be announced to all participants. All are 0; S42. Calculate the adjusted net utility of the current active participants. It also collects a combination of data items that maximize net utility, forming the current demand set for each active participant. ; S43. Set the objective function and determine whether there is a conflict in the needs of active participants by finding a feasible allocation scheme that can simultaneously satisfy all constraints. If there is a conflict, proceed to step S44; if there is no conflict, proceed to step S46. S44, Based on the current demand set Calculate and identify a minimum supply shortage set of participants. ; S45. Personalized payment tokens for participants within the minimum supply shortage set, targeting the demand-sharing model. Increase by a preset positive increment ,renew The other chips remain unchanged, and then the iteration rounds are set. Then return to step S41 for the next iteration; S46. The iteration terminates, the current allocation state and chip matrix are locked, and the clearing state is output.

[0010] Preferably, in step S2, the bidding strategy of the virtual participants is set independently in advance by the data allocator; the strategy is configured to inject one or more preset bidding parameters into the allocation system, which serve as implicit dynamic reserved chips to ensure the minimum chips of the seller while maintaining the incentive compatibility of real participants.

[0011] Preferably, the specific value of the value adjustment coefficient in step S2 is preset by the data allocator according to the expected revenue target or allocation strategy.

[0012] Preferably, the objective function in step S43 is: ; The constraints are: Wherein, constraint (1) represents the data sharing mode for any chosen method. There must be exactly one Each participant shares the data; constraint (2) indicates that among all available data sharing modes, only one mode can be selected for clearing; constraint (3) indicates that each participant can be assigned to at most one data sharing mode; constraint (4) indicates that in the selected data sharing mode... Under this condition, the total number of successful bidders who obtain the right to secondary development of data cannot exceed the upper limit preset by the data allocation party. , Let be a binary indicator variable, where i represents a real participant, m represents a virtual participant, j represents a participant, N is the set of real participants, M is the set of virtual participants, h represents an arbitrarily chosen data sharing pattern h, and H is the set of data sharing patterns, H={1,2,...,h}. and It is a variable belonging to 0-1, representing whether the actual participant i selects the data sharing mode h. This indicates whether virtual participant m selects data sharing mode h. This indicates whether participant k selects data sharing mode h. This represents the valuation of data sharing pattern h by the actual participant i. This represents the valuation of virtual participant m for data sharing pattern h. It is the weight adjustment coefficient for real participant i. It is the weight adjustment coefficient for virtual participant m. The value h indicates whether the entire system selects the data sharing mode. If selected, the value is 1; otherwise, it is 0.

[0013] Preferably, when participants In data sharing mode When secondary development rights are required ,otherwise .

[0014] Preferably, the participant valuation and allocation need to meet the following conditions: ; Where K' is K is a subset of K', and V(K) represents the social welfare of all participants in set K performing the above allocation process. The left side of the inequality represents the marginal utility of participant j in set K, and the right side of the inequality represents the marginal utility of participant j in set K'.

[0015] Preferably, in step S44, the set of participants with the minimum supply shortage is identified. The specific steps are as follows: S441. Define the supply shortage set: any subset consisting of all currently active participants. If no allocation scheme satisfying all the aforementioned constraints can be found, such that the subset... Each participant can be assigned to their current demand set. If a data sharing pattern is selected, then the subset is determined. This is a set of supply shortages where demand conflicts exist; S442. Determine the set of participants with the minimum supply shortage: If there exists a specific set among the identified supply shortage sets... such that from this set If removing any single participant results in any proper subset of the set no longer being a supply shortage set, then the set is considered to be in a state of supply shortage. It is the set of participants with the minimum supply shortage.

[0016] Secondly, the present invention provides an allocation system for implementing a data allocation method based on affine VCG and primal dual iteration, comprising: The participant interface module is used to receive valuations from real participants. Feedback on requirements during iteration ; The virtual participant strategy module is used to store and execute preset demand reporting rules for virtual participants; The iterative allocation engine, the core processing module, is used to manage personalized payment chip matrices. Calculate the allocation state value and identify the minimum supply shortage set. And control the chip update and iteration cycle; The allocation and payment calculation module is used to determine the final allocation when the iteration terminates. And adjust based on the final personalized payment chips and profit adjustment coefficient. Calculate the final payment list.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The data allocation method based on affine VCG and original dual iteration of the present invention changes the competitive structure by introducing virtual participants and directly amplifies the amount of payment chips by adjusting the coefficients. From the algorithm level, it ensures that the seller's expected return is significantly improved, and enhances the participation willingness of resource suppliers and the stability of the system ecosystem.

[0018] (2) The method of the present invention decomposes the originally one-time sealed calculation process into a series of logically clear and open interactive steps, namely a complete set of quotation, demand feedback and chip price adjustment. The intermediate state of each step, such as chip price and temporary allocation, is visible to all participants, which greatly improves the transparency, fairness and credibility of the allocation process and reduces subsequent disputes.

[0019] (3) The data allocation method based on affine VCG and primal dual iteration of the present invention has been theoretically proven to maintain the core properties such as incentive compatibility (IC) and individual rationality (IR) after the above improvements are introduced, thus ensuring the theoretical rigor and practical reliability of the algorithm.

[0020] (4) The entire iterative process of this invention has a defined logic and clear steps, making it easy to implement through programming. It is particularly suitable for distributed technology environments that require high transparency and automated execution, such as blockchain smart contracts. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the architecture of the data allocation system based on affine VCG and primal dual iteration of the present invention; Figure 2 This is the overall flowchart of the data allocation method based on affine VCG and primal dual iteration of the present invention; Figure 3 This is a detailed flowchart of the chip adjustment iteration process in this invention; Figure 4 This is a schematic diagram illustrating the relationship between the number of virtual participants and platform revenue in an embodiment of the present invention. Detailed Implementation

[0022] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

[0023] like Figure 1As shown in the figure, this invention provides a data allocation system 100 based on affine VCG (Vickery-Clark-Grovsian mechanism) and primal-dual iteration. Affine VCG is one of the most fundamental and important mechanisms in the field of mechanism design, with dominant strategy incentive compatibility and social welfare maximization as its core characteristics. This system can be deployed on a cloud server or in a blockchain smart contract environment and mainly includes the following modules: Participant Interface Module 101: Used to receive input parameters such as the valuation of real participants and their willingness to further develop the data.

[0024] Virtual Participant Strategy Module 102: Based on the underlying logic set by the data allocator, it injects the preset requirements and bidding parameters of virtual participants into the system as an implicit dynamic chip retention mechanism.

[0025] Iterative Allocation Engine 103: This is the core computing component of the system, receiving data from modules 101 and 102. Internally, it further includes a state and utility calculation unit (for managing the personalized payment chip matrix and calculating net utility) and a supply shortage set identification unit (for determining market conflicts and locking the set to be adjusted).

[0026] Allocation and Payment Calculation Module 104: After the iterative allocation engine 103 determines that the market has reached an equilibrium clearing state, it receives the output results, calculates the final allocation scheme, and generates the final payment list for the winning bidders according to the correction rules. The correction rules are as follows: the final payment amount for the real participants who won the bid is their final personalized payment token divided by their value adjustment coefficient, and the final payment amount for the virtual participants who won the bid is set to 0.

[0027] For ease of subsequent description, let's first define relevant symbols, and denote the extended participant set as... It contains a set of real participants. With virtual participants Let the set of all possible data sharing modes be denoted as . For any participant any round and any data sharing mode ,use Indicates the user's current personal chip count; sets the iteration round. The initial temporary stakes for the current data-sharing model will be announced to all participants. .

[0028] Based on current participants Exposed bid value And value adjustment coefficient and current personalized bargaining chips Calculate its value for each shared data combination Adjusted net utility And calculate what would make it The maximum combination of one or more data items constitutes its current demand set. .

[0029] Based on all active participants (i.e., maximum adjusted net utility) The current demand set of the participants The system attempts to find a compatible allocation scheme. If the needs of active participants conflict and cannot be satisfied collectively (i.e., there is a resource shortage), the system will allocate resources based on the demand set of all participants. Calculate and identify a minimum supply shortage set of participants. Subsequently, this set Participants The data-sharing model with demand adds a pre-set positive increment to its personalized leverage. (Typically, (i.e., update) Other chips remain unchanged. After this round of updates is complete, set... Then return to the previous step for the next iteration.

[0030] If all the needs of active participants can be satisfied by a conflict-free allocation scheme, or if the maximum adjusted net utility of all participants is achieved... If all values ​​are 0, it indicates that the market has reached a state of equilibrium clearing, and the iteration ends.

[0031] Select a final allocation scheme from all currently feasible allocation schemes. And record the winning participants at this time. The combination of the data items obtained The ultimate personalized bargaining chip .

[0032] According to the final allocation And the corresponding final personalized chips, determining the payment for each winning bidder, for each real participant who won the bid. Its final payment stake is its final personalized stake divided by its value adjustment factor, i.e. For each winning virtual participant, the final payout chip is 0.

[0033] like Figure 2 As shown, the data allocation method provided in this embodiment of the invention, based on the above-mentioned system 100, includes the following steps in its overall macro-process: Step S1: Input data transaction parameters. The system receives the set of selectable data sharing modes, the set of real participants, the bid valuations of each real participant for different sharing modes, additional secondary development willingness options, and the upper limit of the number of successful bidders allowed to conduct secondary data development by the data allocator through the participant interface module 101.

[0034] Step S2: Generate virtual participants. The virtual participant strategy module 102 constructs a set of virtual participants and sets the bidding valuation of virtual participants for each data sharing mode, forming a set that includes real participants. With virtual participants Extended bid set and a collection of data sharing models , where any participant Any data sharing mode The bidding strategy of virtual participants is set independently in advance by the data allocator; the strategy involves injecting one or more preset parameters into the allocation system. These parameters serve as implicit, dynamically reserved chips to ensure the seller's minimum chips while maintaining incentive compatibility with real participants.

[0035] Step S3: Set the value adjustment coefficient. The system expands the bid set. Participants Set the corresponding value adjustment coefficient The specific value of the value adjustment coefficient is preset by the data allocator based on the expected return target or allocation strategy. Step S4: Execute a multi-round primitive-dual iterative allocation process. The iterative allocation engine 103 is based on an extended bid set. With value adjustment coefficient The allocation process involves multiple rounds of iterative allocation cycles until a clearing state is reached. Specifically, the allocation iterative cycle refers to the cycle of price quotation, feedback, and price adjustment.

[0036] Combination Figure 3 As shown, the multi-round primitive-dual iterative allocation process in step S4 above is specifically executed by the iterative allocation engine 103 in a loop according to the following sub-steps: S41. Initialize the iteration rounds The initial temporary stakes for each of the current data-sharing models will be announced to all participants. All are 0.

[0037] S42. Calculate the adjusted net utility of the current active participants. It also collects a combination of data items that maximize net utility, forming the current demand set for each active participant. .

[0038] S43. Set the objective function and determine whether there is a conflict in the needs of active participants by finding a feasible allocation scheme that can simultaneously satisfy all constraints. If there is a conflict, proceed to step S44; if there is no conflict, proceed to step S46.

[0039] The objective function in step S43 is: ; The constraints are: Wherein, constraint (1) represents the data sharing mode for any chosen method. There must be exactly one Each participant shares the data; constraint (2) indicates that among all available data sharing modes, only one mode can be selected for clearing; constraint (3) indicates that each participant can be assigned to at most one data sharing mode; constraint (4) indicates that in the selected data sharing mode... Under this condition, the total number of successful bidders who obtain the right to secondary development of data cannot exceed the upper limit preset by the data allocation party. , As a binary indicator variable, when participants In data sharing mode When secondary development rights are required ,otherwise Let i represent a real participant, m represent a virtual participant, j represent a participant, N be the set of real participants, M be the set of virtual participants, h represent an arbitrarily chosen data sharing mode h, and H be the set of data sharing modes, where H = {1, 2, ..., h}. and It is a variable belonging to 0-1, representing whether the actual participant i selects the data sharing mode h. This indicates whether virtual participant m selects data sharing mode h. This indicates whether participant k selects data sharing mode h. This represents the valuation of data sharing pattern h by the actual participant i. This represents the valuation of virtual participant m for data sharing pattern h. It is the weight adjustment coefficient for real participant i. It is the weight adjustment coefficient for virtual participant m. The value h indicates whether the entire system selects the data sharing mode. If selected, the value is 1; otherwise, it is 0.

[0040] The valuation and allocation of participants must meet the following conditions: .

[0041] This formula implies diminishing marginal returns, meaning that as the pool of participants expands, the return on investment decreases with each new bidder. The resulting increase in social welfare is diminishing. This is a sufficient condition to ensure that the iterative auction mechanism terminates at VCG payment and maintains the true bids of the bidders (incentive compatibility). Where K' is... Let K be a subset of K', and let V(K) represent the social welfare of all participants in set K for the above allocation process. The left side of the inequality represents the social welfare of a smaller subset K plus the social welfare of participant j minus the social welfare of only set K (that is, the marginal utility of participant j in set K). The right side of the inequality represents the marginal utility of participant j in the larger set K'.

[0042] S44, Based on the current demand set Calculate and identify a minimum supply shortage set of participants. The set satisfies the condition that "removing any one member allows the remaining members to be assigned compatiblely".

[0043] Identify the set of participants with minimum supply shortage The specific steps are as follows: S441. Define the supply shortage set: any subset consisting of all currently active participants. If no allocation scheme satisfying all the aforementioned constraints can be found, such that the subset... Each participant can be assigned to their current demand set. If a data sharing pattern is selected, then the subset is determined. This is the set of supply shortages where there is a demand conflict.

[0044] S442. Determine the set of participants with the minimum supply shortage: If there exists a specific set among the identified supply shortage sets... such that from this set If removing any single participant results in any proper subset of the set not being a supply shortage set, then the set is considered to be in a state of supply shortage. It is the set of participants with the minimum supply shortage.

[0045] S45. Personalized payment tokens for participants within the minimum supply shortage set, targeting the demand-sharing model. Increase by a preset positive increment ,renew The other chips remain unchanged, and then the iteration rounds are set. and return to step S41 for the next iteration; positive increment It is usually set to 1.

[0046] S46. The iteration terminates, the current allocation state and chip matrix are locked, and the clearing state is output.

[0047] To more intuitively demonstrate the operational effect of the above iterative system, a specific application scenario numerical example is provided below. In a data transaction scenario, three real participants (denoted as a, c, and d) and one virtual participant (denoted as b) are involved in bidding. Available data sharing modes include: dedicated mode (…). ), two-person sharing mode ( ) and three-person sharing mode ( At the same time, the maximum number of successful bidders allowed to obtain the right to further develop the data is set at 1.

[0048] First, the participant interface module receives the actual bid valuations and secondary development intentions (with) from these three real participants. (The number indicates that secondary development rights are required). Simultaneously, the virtual participant strategy module is triggered, automatically injecting one virtual participant (denoted as b) and its preset bidding parameters into the system. The system sets a value adjustment coefficient of 2 for virtual participant b, and a value adjustment coefficient of 1 for the remaining real participants (a, c, d). The initial data collected by the system at this time is shown in Table 1: Table 1 After data collection is complete, the iterative allocation engine is started. Its internal state and utility calculation unit first multiplies the original valuation by an adjustment coefficient as the iteration baseline value and initializes the temporary chips to 0. Subsequently, the iterative allocation engine and its internal supply shortage set identification unit execute multiple rounds of iterations according to the aforementioned logic. The specific iteration process is shown in Table 2: Table 2 Once the iterative allocation engine determines a clear, the allocation and payment calculation module takes over the output. This module records that the final personalized payment chip obtained by real participant A in the "two-person sharing mode" is 4. Subsequently, the allocation and payment calculation module calls the correction rule to calculate that the final payment chip of real participant A is its personalized payment chip divided by the adjustment coefficient, i.e., 4 / 1=4; at the same time, the actual payment amount of virtual participant B, who is just a participant in the mechanism, is automatically counted as 0.

[0049] Compared to traditional VCG allocations that do not introduce virtual participants (where participant A exclusively enjoys the data and the seller only benefits by 3), this embodiment increases the seller's total revenue to 4 through the system's iterative engine and virtual bid injection. This implementation not only maintains the incentive compatibility of participants' real bids but also effectively overcomes the technical bottleneck of low platform revenue caused by traditional mechanisms, achieving high transparency throughout the allocation process.

[0050] In this embodiment, as Figure 4 As shown, a repeated experiment was conducted in this embodiment: with 20 real participants, 1 to 10 virtual participants were added sequentially, and the distribution of bidding value between real and virtual participants was made similar. Figure 4 This diagram illustrates the relationship between the proportion of virtual participants in the winning set and platform revenue. The horizontal axis represents the number of virtual participants added; the left vertical axis represents the change in platform revenue (blue line) and the change in social welfare (red line), both based on the scenario without virtual participants; the right vertical axis represents the composition of the winning set, with the green area representing the proportion of real participants and the orange area representing the proportion of virtual participants.

[0051] Depend on Figure 4 It is evident that the proportion of virtual participants in the winning set increases with the number of virtual participants. However, when the proportion of virtual winners is too high, platform revenue does not increase accordingly, and may even decrease. This result contradicts the general intuition that "an excessively high proportion of virtual winners will severely damage platform revenue." Therefore, introducing an appropriate number of virtual participants can significantly improve platform revenue.

[0052] Step S5: Determine the final allocation. After market clearing, the allocation and payment calculation module 104 determines the final optimal allocation scheme and the final personalized payment chips for each successful bidder.

[0053] Step S6: Calculate the final payment. Based on the final personalized payment tokens and the value adjustment coefficient, calculate and output the actual payment tokens for each winning participant. Specifically, for each winning real participant, divide their final personalized payment tokens by their value adjustment coefficient to obtain the final payment tokens; for each winning virtual participant, record their final payment tokens as 0.

[0054] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A data allocation method based on affine VCG and primal-dual iteration, characterized in that: It includes the following steps: S1. Input data transaction parameters: Receive the set of data sharing modes to choose from, the set of real participants, the bid valuation of each real participant for different sharing modes, additional secondary development intention options, and the upper limit of the number of successful bidders allowed by the data distributor to conduct secondary data development; S2, generating virtual participants: constructing a set of virtual participants, and setting the bidding evaluation of the virtual participants to each data sharing mode, forming a set of real participants and the set of virtual participants extended bidding set and the set of data sharing modes wherein any participant , any data sharing mode ; S3, set value adjustment coefficient: for extending the bid set of each participant ;​​ S4, performing multiple rounds of iterative allocation: based on the extended bid set with the value adjustment factor , performing multiple rounds of allocation iteration loops until reaching a clearing state; S5. Determine the final allocation: After the clearing state is reached, determine the final optimal allocation plan and the final personalized payment chips for each successful bidder; S6. Calculate the final payment: Based on the final personalized payment chips and value adjustment factor. Calculate and output the actual payment chips of each winning bidder.

2. The data allocation method based on affine VCG and primal-dual iteration according to claim 1, characterized in that: Step S4 specifically includes the following sub-steps: S41. Initialize the iteration rounds The initial temporary stakes for each of the current data-sharing models will be announced to all participants. All are 0; S42. Calculate the adjusted net utility of the current active participants. It also collects a combination of data items that maximize net utility, forming the current demand set for each active participant. ; S43. Set the objective function and determine whether there is a conflict in the needs of active participants by finding a feasible allocation scheme that can simultaneously satisfy all constraints. If there is a conflict, proceed to step S44; if there is no conflict, proceed to step S46. S44, Based on the current demand set Calculate and identify a minimum supply shortage set of participants. ; S45. Personalized payment tokens for participants within the minimum supply shortage set, targeting the demand-sharing model. Increase by a preset positive increment ,renew The other chips remain unchanged, and then the iteration rounds are set. Then return to step S41 for the next iteration; S46. The iteration terminates, the current allocation state and chip matrix are locked, and the clearing state is output.

3. The data allocation method based on affine VCG and primal-dual iteration according to claim 1, characterized in that: In step S2, the bidding strategy of the virtual participants is set independently in advance by the data allocator. The strategy is to inject one or more preset parameters into the allocation system. These parameters serve as implicit dynamic reserve chips to ensure the minimum chips for the seller while maintaining the incentive compatibility of real participants.

4. The data allocation method based on affine VCG and primal-dual iteration according to claim 1, characterized in that: The specific value of the value adjustment coefficient in step S2 is preset by the data allocator based on the expected return target or allocation strategy.

5. The data allocation method based on affine VCG and primal-dual iteration according to claim 1, characterized in that: The objective function in step S43 is: ； The constraints are: Wherein, constraint (1) represents the data sharing mode for any chosen method. There must be exactly one Each participant shares the data; constraint (2) indicates that among all available data sharing modes, only one mode can be selected for clearing; constraint (3) indicates that each participant can be assigned to at most one data sharing mode; constraint (4) indicates that in the selected data sharing mode... Under this condition, the total number of successful bidders who obtain the right to secondary development of data cannot exceed the upper limit preset by the data allocation party. , Let be a binary indicator variable, where i represents a real participant, m represents a virtual participant, j represents a participant, N is the set of real participants, M is the set of virtual participants, h represents an arbitrarily chosen data sharing pattern h, and H is the set of data sharing patterns, H={1,2,...,h}. and It is a variable belonging to 0-1, representing whether the actual participant i selects the data sharing mode h. This indicates whether virtual participant m selects data sharing mode h. This indicates whether participant k selects data sharing mode h. This represents the valuation of data sharing pattern h by the actual participant i. This represents the valuation of virtual participant m for data sharing pattern h. It is the weight adjustment coefficient for real participant i. It is the weight adjustment coefficient for virtual participant m. The value h indicates whether the entire system selects the data sharing mode. If selected, the value is 1; otherwise, it is 0.

6. The data allocation method based on affine VCG and primal-dual iteration according to claim 5, characterized in that: When participants In data sharing mode When secondary development rights are required ,otherwise .

7. The data allocation method based on affine VCG and primal-dual iteration according to claim 1, characterized in that: The valuation and allocation of participants must meet the following conditions: ； Where K' is K is a subset of K', and V(K) represents the social welfare of all participants in set K performing the above allocation process. The left side of the inequality represents the marginal utility of participant j in set K, and the right side of the inequality represents the marginal utility of participant j in set K'.

8. The data allocation method based on affine VCG and primal-dual iteration according to claim 1, characterized in that: In step S44, the set of participants with the minimum supply shortage is identified. The specific steps are as follows: S441. Define the supply shortage set: any subset consisting of all currently active participants. If no allocation scheme satisfying all the aforementioned constraints can be found, such that the subset... Each participant can be assigned to their current demand set. If a data sharing pattern is selected, then the subset is determined. This is a set of supply shortages where demand conflicts exist; S442. Determine the set of participants with the minimum supply shortage: If there exists a specific set among the identified supply shortage sets... such that from this set If removing any single participant results in any proper subset of the set not being a supply shortage set, then the set is considered to be in a state of supply shortage. It is the set of participants with the minimum supply shortage.

9. An allocation system for the data allocation method based on affine VCG and primal dual iteration as described in claim 1, characterized in that: The distribution system includes: The participant interface module is used to receive valuations from real participants. Feedback on requirements during iteration ; The virtual participant strategy module is used to store and execute preset demand reporting rules for virtual participants; The iterative allocation engine, the core processing module, is used to manage personalized payment chip matrices. Calculate the allocation state value and identify the minimum supply shortage set. And control the chip update and iteration cycle; The allocation and payment calculation module is used to determine the final allocation when the iteration terminates. And adjust based on the final personalized payment chips and profit adjustment coefficient. Calculate the final payment list.

Images