Dynamic data value distribution system based on data utility and production cost

Abstract

The invention relates to the field of Internet of Things data, and discloses a dynamic data value distribution system based on data utility and production cost, comprising: a production cost mapping module for mapping data production cost into a linear or exponential decay relationship based on a bimodal function; the dynamic classification module is used for dynamically dividing core participants and edge participants according to the data volume of the participants; and the data utility calculation module is used for receiving the classification result of the dynamic classification module and the output of the production cost mapping module. According to the method, high-cost and low-cost data producers are distinguished, high-investment enterprises are dynamically compensated, long-term loss is avoided, a data quality evaluation module corrects an allocation weight according to timeliness, integrity and uniqueness, dilution of low-quality data to overall effectiveness is reduced, contribution of core and edge participants is accurately calculated by a hierarchical Shapley value, and the accuracy of data quality evaluation is improved. And contribution and return matching is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of Internet of Things data, and particularly relates to a dynamic data value distribution system based on data utility and production cost. BACKGROUND

[0002] In the era of Internet of Things and big data, the reasonable distribution of data value is the core problem of guaranteeing the cooperation of multiple participants and the sustainable operation of the market. Data value distribution aims to distribute the economic benefits generated by data according to the contribution degree, data quality and production cost of each participant. The current mainstream methods include proportional distribution based on data volume and Shapley value calculation based on contribution degree.

[0003] According to the search, the publication number CN116385036A discloses a multi-aggregator demand response benefit distribution method and system based on Shapley value. The method comprises: obtaining power grid operation conditions and demand response power of each aggregator bid; determining the best demand response incentive scheme according to the power grid operation conditions and the demand response power of each aggregator bid, and publishing it to each aggregator; each aggregator adjusts the user load according to the best demand response incentive scheme, and calculates the demand response reduced power and the to-be-distributed benefits of each aggregator after the demand response ends; according to the demand response reduced power of each aggregator, the Shapley value of each aggregator is calculated respectively; and the to-be-distributed benefits are distributed to each aggregator according to the Shapley value of each aggregator.

[0004] The existing value distribution system does not dynamically combine data production cost and contribution degree, which may cause long-term loss of high-cost participants, the static distribution mechanism lacks the guiding ability of market equilibrium, and it is difficult to encourage participants to spontaneously optimize resource allocation, the Shapley value calculation has high complexity, and it is difficult to support large-scale Internet of Things scenarios. SUMMARY

[0005] To solve the above-mentioned problems that the value distribution system does not dynamically combine data production cost and contribution degree, which may cause long-term loss of high-cost participants, the static distribution mechanism lacks the guiding ability of market equilibrium, and it is difficult to encourage participants to spontaneously optimize resource allocation, the present application realizes the technical scheme as follows.

[0006] The dynamic data value distribution system based on data utility and production cost comprises:

[0007] The production cost mapping module maps the data production cost into a linear or exponential decay relationship based on a bimodal function;

[0008] The dynamic classification module dynamically divides core participants and edge participants according to the data volume of the participants;

[0009] The data utility calculation module receives the classification results from the dynamic classification module and the output from the production cost mapping module. It calculates the marginal utility of core participants using hierarchical Shapley values ​​and simplifies the calculation for marginal participants according to the proportion of data volume.

[0010] The optimization modeling module uses the Lagrange multiplier method to solve the utility maximization problem. Its inputs include the classification results of the dynamic classification module and the total input constraints.

[0011] The feedback adjustment module, based on the solution results of the optimization modeling module and the classification results of the dynamic classification module, adjusts the profit rate to achieve market equilibrium and maximize total utility.

[0012] Preferably, the calculation steps for the layered Shapley values ​​are as follows:

[0013] S1: Participant stratification:

[0014] Key participants: Those who account for ≥20% of the data volume;

[0015] Marginal participants: those accounting for less than 20% of the data volume;

[0016] S2: Core participant calculation:

[0017] For the core participant set C, Monte Carlo sampling is used to generate M groups of random contribution combinations;

[0018] Calculate the marginal utility of participant i in each group. The Shapley value is obtained by taking the average:

[0019]

[0020] S3: Merging of Marginal Participants

[0021] The total data volume V of the peripheral participant set E E =∑ j∈E vj is treated as a virtual participant, and its Shapley value Shapley(E) is calculated.

[0022] S4: Allocation and Integration

[0023] Value allocated to key participants: Shapley(i)·p;

[0024] Marginal participants are allocated proportionally:

[0025] Preferably, the dual-mode function is:

[0026] Linear mapping: v i =kx i ;

[0027] Exponential decay mapping: v i =x i (1-e - βx i );

[0028] Among them, v i For effective data volume, x i The input cost is represented by k and β, which are fitting parameters.

[0029] Preferably, the feedback adjustment module dynamically adjusts the participant's cost calculation based on the difference between the marginal utility and the average marginal utility calculated by the data utility calculation module, using the formula:

[0030]

[0031] Adjusted costs The product of the average profit rate p is used as the allocation value to achieve market equilibrium and maximize total utility.

[0032] Preferably, the adjustment coefficient α in the feedback adjustment module has a value range of 0 < α ≤ 1, and α is dynamically adjusted according to market volatility.

[0033] Preferably, the threshold of the dynamic classification module is:

[0034] Key participants: those whose data volume is ≥ the top 20% of the total data volume or whose marginal utility is ≥ the mean;

[0035] Marginal participants: those whose data volume or marginal utility is below the above threshold.

[0036] Preferably, the optimization modeling module solves the problem using the Lagrange multiplier method:

[0037]

[0038] Preferred options also include:

[0039] The data quality assessment module calculates quality weights based on timeliness, completeness, and uniqueness, and adjusts marginal utility accordingly.

[0040] Preferably, the feedback adjustment module integrates a machine learning optimizer for:

[0041] Inputs: historical allocation data, market equilibrium deviation, and participant investment growth rate;

[0042] Output: Dynamically adjust the regulation coefficient α t =α t-1 +η·ΔU;

[0043] Where η is the learning rate and ΔU is the change in utility.

[0044] Preferred options also include:

[0045] The blockchain-based evidence storage module is used for:

[0046] Record participant input cost x i Distributing value and marginal utility

[0047] The allocation is automatically executed through smart contracts, and immutable credentials are generated.

[0048] This invention provides a dynamic data value allocation system based on data utility and production cost. Compared with existing technologies, it has the following advantages: by distinguishing between high-cost and low-cost data producers, it dynamically compensates high-investment enterprises, avoiding long-term losses; the data quality assessment module adjusts the allocation weights according to timeliness, completeness, and uniqueness, reducing the dilution of overall utility by low-quality data; and the hierarchical Shapley value accurately calculates the contributions of core and peripheral participants, ensuring that contributions match returns. Attached Figure Description

[0049] Figure 1 This is a system architecture diagram proposed in this invention. Detailed Implementation

[0050] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0051] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0052] Example 1

[0053] Reference Figure 1 This is a dynamic data value allocation system based on data utility and production costs. The system achieves dynamic allocation of data value through the collaborative work of the following modules:

[0054] Production cost mapping module: Maps data production costs to effective data volume, distinguishing between linear and high fixed cost scenarios;

[0055] The dynamic classification module dynamically divides participants into core participants and peripheral participants based on the amount of data they have.

[0056] The data utility calculation module receives the classification results from the dynamic classification module and the output from the production cost mapping module. It calculates the marginal utility of core participants using hierarchical Shapley values ​​and simplifies the calculation for marginal participants according to the proportion of data volume.

[0057] Feedback adjustment module: Adjusts cost calculations based on marginal utility differences to guide market equilibrium;

[0058] Optimization Modeling Module: Maximizes utility through mathematical optimization.

[0059] Among them, the production cost mapping module maps the data production cost into a linear or exponential decay relationship based on a dual-modal function;

[0060] The dual-mode function is:

[0061] Linear mapping: v i =kx i ;

[0062] Exponential decay mapping: v i =x i (1-e - βx i );

[0063] Among them, v i For effective data volume, x i The input cost is represented by k and β, which are fitting parameters.

[0064] The data utility calculation module receives the output of the production cost mapping module and calculates the marginal utility of each participant's data using hierarchical Shapley values.

[0065] The steps for calculating the hierarchical Shapley value are as follows:

[0066] S1: Participant stratification:

[0067] Key participants: Those who account for ≥20% of the data volume;

[0068] Marginal participants: those accounting for less than 20% of the data volume;

[0069] S2: Core participant calculation:

[0070] For the core participant set C, Monte Carlo sampling is used to generate M groups of random contribution combinations;

[0071] Calculate the marginal utility of participant i in each group. The Shapley value is obtained by taking the average:

[0072]

[0073] S3: Merging of Marginal Participants

[0074] The total data volume V of the peripheral participant set E E =∑ j∈E vj is treated as a virtual participant, and its Shapley value Shapley(E) is calculated.

[0075] S4: Allocation and Integration

[0076] Value allocated to key participants: Shapley(i)·p;

[0077] Marginal participants are allocated proportionally:

[0078] The dynamic classification module dynamically divides core participants into peripheral participants based on the marginal utility provided by the data utility calculation module and the amount of data of the participants;

[0079] The threshold value of the dynamic classification module is:

[0080] Key participants: those whose data volume is ≥ the top 20% of the total data volume or whose marginal utility is ≥ the mean;

[0081] Marginal participants: those whose data volume or marginal utility is below the above threshold.

[0082] The optimization modeling module uses the Lagrange multiplier method to solve the utility maximization problem. Its inputs include the classification results of the dynamic classification module and the total input constraints.

[0083] The optimization modeling module solves the problem using the Lagrange multiplier method:

[0084]

[0085] The data quality assessment module calculates quality weights based on timeliness, completeness, and uniqueness, and adjusts marginal utility accordingly.

[0086] The data quality assessment module is used to quantitatively assess the timeliness, completeness, and uniqueness of data, preventing low-quality data from diluting the overall utility; it also improves the fairness and accuracy of allocation by correcting Shapley values ​​through quality weights.

[0087] Timeliness (Ti): The time interval between data generation and use (e.g., 1 hour is the maximum score, and 10% is deducted for each hour of delay);

[0088] Completeness (Ci): The percentage of missing data fields (e.g., a missing rate of ≤5% is the maximum score, and 20% is deducted for every additional 5%).

[0089] Uniqueness (Ui): Data duplication rate (e.g., a duplication rate of ≤2% is full marks, and 15% is deducted for every 1% increase);

[0090] Weighted calculation model: The scores of each dimension are normalized and then summed with weights. The formula is as follows:

[0091] w i =α·T i +β·C i +γ·U i .

[0092] α, β, γ are weighting coefficients;

[0093] Score range: w i ∈[0,1] (0 represents low quality, 1 represents high quality).

[0094] The feedback adjustment module, based on the solution results of the optimization modeling module and the classification results of the dynamic classification module, adjusts the profit rate to achieve market equilibrium and maximize total utility.

[0095] The feedback adjustment module dynamically adjusts the participants' cost calculations based on the difference between the marginal utility and the average marginal utility calculated by the data utility calculation module, using the formula:

[0096]

[0097] Adjusted costs The product of the average profit rate p is used as the allocation value to achieve market equilibrium and maximize total utility; the adjustment coefficient α in the feedback adjustment module has a range of 0 < α ≤ 1, and α is dynamically adjusted according to market volatility.

[0098] The feedback tuning module integrates a machine learning optimizer for:

[0099] Inputs: historical allocation data, market equilibrium deviation, and participant investment growth rate;

[0100] Output: Dynamically adjust the regulation coefficient α t =α t-1 +η·ΔU;

[0101] Where η is the learning rate and ΔU is the change in utility.

[0102] The blockchain-based evidence storage module is used for:

[0103] Record participant input cost x i Distributing value and marginal utility

[0104] The allocation is automatically executed through smart contracts, and immutable credentials are generated.

[0105] Suppose that multiple agricultural sensor manufacturers and algorithm companies share farmland environmental data (temperature, humidity, pest and disease prediction, etc.), the value of the data needs to be dynamically allocated.

[0106] Technical implementation steps:

[0107] 1. Production cost mapping

[0108] Sensor manufacturer A (linear cost): Input cost x i = 100,000 yuan, linear coefficient k = 0.8, effective data volume v i =80,000 entries.

[0109] Algorithm Company B (High Fixed Costs): Input Costs x i = 500,000 yuan, attenuation coefficient β = 0.05, effective data volume v i =50(1-e -0.05x50 )≈393,000 entries.

[0110] 2. Calculation of stratified Shapley values

[0111] Key participants: Data volume ≥ 200,000 records (e.g., algorithm company B), using Monte Carlo sampling (M = 50 times) to calculate marginal utility, results:

[0112] Edge participants: Sensor manufacturers A and others are merged into virtual participants, and value is allocated according to the proportion of data volume.

[0113] 3. Feedback Adjustment

[0114] Marginal utility of algorithm company B: With a mean of 10 and an adjustment factor α = 0.2, the adjusted cost is:

[0115] Increase in distribution value: 50.4 × 15% = 75,600 yuan (originally 75,000 yuan).

[0116] Effect verification:

[0117] Fairness: The loss rate of high-cost algorithm companies decreased from 40% to 15%;

[0118] Efficiency: Computation time reduced from 8 hours to 1.5 hours;

[0119] Dynamics: Market equilibrium is reached faster by 60%.

[0120] Example 2: Industrial Equipment Monitoring Data Platform

[0121] Assume that multiple equipment manufacturers and maintenance companies share equipment operation data to optimize maintenance strategies.

[0122] Technical implementation steps:

[0123] 1. Dynamic classification module

[0124] Threshold setting: Core participants must meet the following requirements: data volume ≥ the top 15% of the total data volume or marginal utility ≥ the mean.

[0125] Classification results: Manufacturer C (1.2 million data entries, marginal utility 12) is the core, and the rest are peripheral participants.

[0126] 2. Optimize the modeling module

[0127] Solving the utility maximization problem using the Lagrange multiplier method, the equilibrium condition is:

[0128]

[0129] Results: Total utility increased by 35%, and equipment maintenance costs decreased by 20%.

[0130] 3. Blockchain Evidence Storage Module

[0131] Record participants' input costs, allocated value, and marginal utility to generate tamper-proof credentials.

[0132] Effect verification:

[0133] Transparency: Improved auditability of allocation records reduces dispute rate by 70%;

[0134] Compliance: Meets industrial data security standards (such as ISO 27001).

[0135] Example 3: Based on the dynamic data value allocation system based on data utility and production cost described in Example 1, this invention provides a dynamic data value allocation method based on data utility and production cost. This method, through dynamic game theory and the hierarchical Shapley value model, combined with data utility, production cost, and quality assessment, achieves fair value allocation in collaborative scenarios involving multiple participants. The core steps include dynamic mapping of production costs, hierarchical utility calculation, dynamic classification and feedback adjustment, and ensure the transparency of allocation through blockchain technology. The specific implementation steps are as follows:

[0136] Step 1: Dynamic Mapping of Data Production Costs

[0137] Input: Each participant's input cost x i (such as equipment procurement and research and development expenses).

[0138] Bimodal mapping function:

[0139] Linear mode (regular data production):

[0140] v i =kx i(k is the fitting coefficient, calibrated using historical data);

[0141] Exponential decay pattern (high fixed cost scenarios, such as R&D-intensive data):

[0142] (β is the attenuation coefficient);

[0143] Output: Valid data volume v i This reflects the contribution of cost differences to the value of data;

[0144] Step 2: Calculate the utility of the stratified Shapley value

[0145] Participant stratification:

[0146] Key participants: Those who account for ≥20% of the data volume;

[0147] Marginal participants: accounting for less than 20% of the data volume.

[0148] Core participant calculations:

[0149] Monte Carlo sampling generates M sets of random contribution combinations, and the mean marginal utility is calculated:

[0150]

[0151] Marginal participant allocation:

[0152]

[0153] v E Total data volume for edge participants;

[0154] Step 3: Dynamic Feedback Adjustment and Equilibrium Optimization

[0155] Marginal utility difference adjustment:

[0156]

[0157] α∈(0,1) is the adjustment coefficient, representing the cost after dynamic optimization. Participate in the allocation.

[0158] Equilibrium modeling:

[0159] Solving utility maximization problems using the Lagrange multiplier method:

[0160]

[0161] Equilibrium condition: All participants have equal marginal utility. Step 4: Data Quality Assessment and Weight Adjustment

[0162] Quality Dimensions:

[0163] Timeliness Ti: Data delay time (e.g., ≤1 hour gets 1.0, deduct 0.1 for each hour exceeding 1 hour); Completeness Ci: Field missing rate (e.g., ≤5% gets 1.0, deduct 0.2 for each 5% exceeding 5%);

[0164] Uniqueness Ui: Repetition rate (e.g., ≤2% gets 1.0, deduct 0.15 for every 1% exceeding 2%).

[0165] Weight calculation: w i =0.4T i +0.3C i +0.3U i

[0166] Utility Modification: Step 5: Blockchain Evidence Storage and Smart Contract Execution

[0167] Data on the blockchain:

[0168] Record input cost x i Allocation value Marginal utility

[0169] Generate SHA-256 hash credentials.

[0170] In summary, compared with existing technologies, it has the following beneficial effects:

[0171] By differentiating between high-cost and low-cost data producers, the system dynamically compensates high-investment enterprises to avoid long-term losses. The data quality assessment module adjusts the weight allocation based on timeliness, completeness, and uniqueness to reduce the dilution of overall utility by low-quality data. The hierarchical Shapley value accurately calculates the contributions of core and peripheral participants, ensuring that contributions match rewards.

[0172] The core participants use Monte Carlo sampling to simplify the calculation, reducing the complexity from exponential to linear and effectively shortening the calculation time. The edge participants are merged into virtual participants to reduce redundant calculations, making it suitable for large-scale IoT scenarios.

[0173] The profit margin feedback mechanism adjusts participants' inputs in real time, incentivizes optimal resource allocation, mathematical modeling supports maximizing utility, ensuring the system reaches its optimal state through dynamic adjustments, blockchain technology records input costs, allocation values, and utility data, generating tamper-proof credentials, and smart contracts automatically execute allocation logic, reducing human intervention and improving audit efficiency.

[0174] Therefore, although the invention has been described herein with reference to specific embodiments thereof, freedom of modification, various changes and substitutions are also within the scope of the foregoing disclosure, and it should be understood that in some cases, certain features of the invention may be adopted without departing from the scope and spirit of the invention and without corresponding use of other features. Thus, many modifications can be made to adapt a particular environment or material to the essential scope and spirit of the invention. The invention is not intended to be limited to the specific terminology used in the following claims and / or the specific embodiments disclosed as the best mode for carrying out the invention, but the invention will include any and all embodiments and equivalents falling within the scope of the appended claims. Therefore, the scope of the invention will be defined only by the appended claims.

Claims

1. A dynamic data value allocation system based on data utility and production cost, characterized in that, include: The production cost mapping module maps data production costs to a linear or exponentially decaying relationship based on a bimodal function. The dynamic classification module dynamically divides participants into core participants and peripheral participants based on the amount of data they have. The data utility calculation module receives the classification results from the dynamic classification module and the output from the production cost mapping module. It calculates the marginal utility of core participants using hierarchical Shapley values ​​and simplifies the calculation for peripheral participants according to the proportion of data volume. The optimization modeling module uses the Lagrange multiplier method to solve the utility maximization problem. Its inputs include the classification results of the dynamic classification module and the total input constraints. The feedback adjustment module, based on the solution results of the optimization modeling module and the classification results of the dynamic classification module, adjusts the profit rate to achieve market equilibrium and maximize total utility.

2. The dynamic data value allocation system based on data utility and production cost according to claim 1, characterized in that, The steps for calculating the hierarchical Shapley value are as follows: S1: Participant stratification: Key participants: Those who account for ≥20% of the data volume; Marginal participants: those accounting for less than 20% of the data volume; S2: Core participant calculation: For the core participant set C, Monte Carlo sampling is used to generate M groups of random contribution combinations; Calculate the marginal utility of participant i in each group. The Shapley value is obtained by taking the average: S3: Merging of marginal participants: The total data volume V of the peripheral participant set E E =∑ j∈E vj is treated as a virtual participant, and its Shapley value Shapley(E) is calculated. S4: Allocation and Integration Value allocated to key participants: Shapley(i)·p; Marginal participants are allocated proportionally:

3. The dynamic data value allocation system based on data utility and production cost according to claim 1, characterized in that, The dual-mode function is: Linear mapping: v i =kx i ; Exponential decay mapping: v i =x i (1-e - βx i ); Among them, v i For effective data volume, x i The input cost is represented by k and β, which are fitting parameters.

4. The dynamic data value allocation system based on data utility and production cost according to claim 1, characterized in that, The feedback adjustment module dynamically adjusts the participants' cost calculations based on the difference between the marginal utility and the average marginal utility calculated by the data utility calculation module, using the formula: Adjusted costs The product of the average profit rate p is used as the allocation value to achieve market equilibrium and maximize total utility.

5. The dynamic data value allocation system based on data utility and production cost according to claim 4, characterized in that, The adjustment coefficient α in the feedback adjustment module has a range of 0 < α ≤ 1, and α is dynamically adjusted according to market volatility.

6. The dynamic data value allocation system based on data utility and production cost according to claim 1, characterized in that, The threshold value of the dynamic classification module is: Key participants: those whose data volume is ≥ the top 20% of the total data volume or whose marginal utility is ≥ the mean; Marginal participants: those whose data volume or marginal utility is below the above threshold.

7. The dynamic data value allocation system based on data utility and production cost according to claim 1, characterized in that, The optimization modeling module solves the problem using the Lagrange multiplier method:

8. The dynamic data value allocation system based on data utility and production cost according to claim 1, characterized in that, Also includes: The data quality assessment module calculates quality weights based on timeliness, completeness, and uniqueness, and adjusts marginal utility accordingly.

9. The dynamic data value allocation system based on data utility and production cost according to claim 1, characterized in that, The feedback adjustment module integrates a machine learning optimizer for: Inputs: historical allocation data, market equilibrium deviation, and participant investment growth rate; Output: Dynamically adjust the regulation coefficient α t =α t-1 +η·ΔU; Where η is the learning rate and ΔU is the change in utility.

10. The dynamic data value allocation system based on data utility and production cost according to claim 1, characterized in that, Also includes: The blockchain-based evidence storage module is used for: Record participant input cost x i Distributing value and marginal utility The allocation is automatically executed through smart contracts, and immutable credentials are generated.

Images