Metacosm data intelligent sharing platform and method

Abstract

The invention belongs to the technical field of intelligence, and discloses a meta universe data intelligent sharing platform and method, and the method comprises the steps: obtaining a user behavior parameter, a virtual article attribute parameter, and a virtual article evaluation parameter; based on the user behavior parameters and the virtual article attribute parameters, missing values are processed by using an interpolation method, and abnormal value processing is performed by using adaptive # imgabs0 # to obtain a preprocessed data set; for the preprocessed data set, utilizing an improved XGBoost model to predict and obtain re-purchase rate index data of the virtual goods; based on the virtual article evaluation parameters and the re-purchase rate index data of the virtual articles, setting a threshold value, and judging whether the virtual articles are suitable for sharing or not; for virtual items suitable for sharing, sharing is carried out by using a block chain technology; the data quality and analysis accuracy are improved, and meanwhile, the security, transparency and non-tampering property of data sharing are ensured through the block chain technology, so that the efficiency and reliability of meta-universe data sharing are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent technology, and specifically to a metaverse data intelligent sharing platform and method. Background Art

[0002] The existing Metaverse data intelligent sharing platform has exposed many problems, which have affected the intelligence, accuracy and security of the platform. First, the types and quantities of data obtained are insufficient, making it impossible for the platform to fully and accurately understand the characteristics of virtual items and their performance in the market, and thus unable to effectively support the decision-making process; Secondly, the platform only relies on a single indicator to determine whether a virtual item is suitable for sharing, ignoring the comprehensive effect of multi-dimensional features, resulting in a one-sided decision model, insufficient information, and a lack of sufficient scientific basis and accuracy in the decision results; In addition, the platform still relies on traditional cloud storage, centralized databases or API interfaces to share data during the process of sharing virtual items, which faces security risks such as data leakage and tampering. It cannot effectively guarantee the transparency, immutability and user privacy of shared data, which increases the risk and distrust in the sharing process.

[0003] In view of this, the present invention proposes a metaverse data intelligent sharing platform and method to solve the above problems. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solutions: a Metaverse data intelligent sharing platform, comprising: a data acquisition module: acquiring user behavior parameters, virtual item attribute parameters and virtual item evaluation parameters; Data processing module: Based on user behavior parameters and virtual item attribute parameters, use interpolation method to process missing values ​​and use adaptive Perform outlier processing to obtain a preprocessed data set; Data prediction module: for the preprocessed data set, the improved XGBoost model is used to predict the repurchase rate index data of virtual items; Judgment module: based on the virtual item evaluation parameters and the virtual item repurchase rate index data, set the threshold to determine whether the virtual item is suitable for sharing; Data sharing module: For virtual items suitable for sharing, blockchain technology is used for sharing.

[0005] Furthermore, the user behavior parameters include: visit volume characteristic data, collection volume characteristic data, and like volume characteristic data; Virtual item attribute parameters include: price feature data and scarcity feature data; Virtual item evaluation parameters include: appearance scoring feature data, function scoring feature data, and user experience scoring feature data.

[0006] Furthermore, the method of obtaining the scarcity feature data includes: Use dynamic time windows to calculate the scarcity feature data of virtual items in the Metaverse and define the range of the time window , Indicates the latest time point. Indicates the starting time point of the time window; Obtain the inventory volume, transaction frequency and price fluctuation data at each time point in the time window and perform normalization processing; In the time window, the inventory, transaction frequency and price fluctuation data obtained at different time points have different contributions to the scarcity feature data. For the inventory, transaction frequency and price fluctuation data at different time points, the weights are set through the attenuation function: ,in represents the initial weight, represents the attenuation coefficient, Indicates the distance from the current time to the latest time; The scarcity characteristic data at time point b is calculated by weighted average method, and the formula is: ,in, , and Respectively represent the weights of inventory volume, transaction frequency and price fluctuation data, , and They respectively represent the inventory volume, transaction frequency and price fluctuation data values ​​obtained at time point b.

[0007] Furthermore, the method of using interpolation to process missing values ​​includes: The user behavior parameters and virtual item attribute parameters include five feature data: number of visits, number of favorites, number of likes, price, and scarcity. The isnull data processing method is used to traverse the five feature data respectively to find the location of the missing values ​​in each feature data. True represents a missing value, and False represents a non-missing value. For True, the forward filling method is used for processing. When the missing value is at the beginning of the feature data, the backward filling method is used for missing value processing.

[0008] Furthermore, the adaptive The specific methods for handling outliers include: For the data after missing value processing, arrange each feature data value in ascending order and use the box plot method to calculate the first quartile and the third quartile , and the interquartile range ; Dynamically adjust by calculating the skewness of each feature data value , for the Feature data set ,in, Indicates The formula for the skewness of the i-th data value in a feature group is: , where n represents the The number of feature data, Indicates The average value of feature data, s represents the The standard deviation of the feature data; When the skewness is equal to 0, Feature data set It belongs to normal distribution, and the upper and lower limits of outliers are: ; ; When the skewness is greater than 0, the characteristic data set Belongs to the right skewed distribution. When the skewness is less than 0, the characteristic data group Belongs to the left skewed distribution, adjusted according to the skewness of the feature data group , and calculate the upper and lower limits of the outliers: , ,in is the adjustment coefficient, when the characteristic data set If the data value in is not within the upper and lower limits of the outlier, it is an outlier. The outlier is identified and replaced with the previous non-outlier data value using the forward filling method.

[0009] Furthermore, the specific method of using the improved XGBoost model to predict the repurchase rate index data of virtual items includes: Step E1: The sample set includes L_o groups of samples, each group of samples includes a preprocessed data set and corresponding virtual item repurchase rate index data; wherein the data contained in the preprocessed data set is used as a feature variable, and the virtual item repurchase rate index data is used as a target variable; Step E2: Use the Gray Wolf Optimization Algorithm to search for the optimal hyperparameter combination of the XGBoost model; Step E3: define the final objective function as minimizing the loss function value; Step E4: Initialize the predicted value and use the mean of the target variable repurchase rate indicator data to define the initial predicted value: ,in, Represents the initialized predicted value of the repurchase rate indicator; Step E5: Calculate the residual. In each round of iteration, calculate the residual between the current round and the previous round: ,in, represents the residual of the tth round, Indicates The predicted value of the round; Step E6: In each round, a decision tree is constructed to fit the residual, and each feature data partition point is calculated. The partition point with the largest gain is selected as the current optimal partition point, and the data sample is divided into two sub-nodes, the left sub-tree and the right sub-tree, and each sub-tree is divided using the gain until the maximum depth of the tree is reached; Step E7: Repeat the iteration. In each round of iteration, the prediction value of the model is updated through the newly generated tree until the condition of the final objective function is met, and then the iteration is stopped.

[0010] Furthermore, the specific method of using the Gray Wolf Optimization Algorithm to search for the optimal hyperparameter combination of the XGBoost model includes: Step U1: Determine the XGBoost hyperparameter combination, including the maximum depth of the tree, the learning rate, the number of trees, the proportion of samples used in each training, the proportion of features randomly selected for each tree, the minimum loss reduction required for splitting, the L2 regularization term, and the L1 regularization term; Step U2: Define the mean square error as the objective function to evaluate the quality of each set of hyperparameter combinations; Step U3: Initialize the gray wolf individuals, randomly initialize M gray wolf individuals, each individual represents a hyperparameter combination; Step U4: Evaluate the fitness of individual gray wolves. Each gray wolf individual represents a set of hyperparameter combinations. Use this hyperparameter combination to train the XGBoost model and calculate the mean square error. , as the fitness value for evaluating this hyperparameter combination, the goal is to minimize ,therefore The smaller it is, the higher the fitness value; Step U5: Determine the current best individual, second best individual, and third best individual based on the fitness value. The position of each gray wolf individual will be adjusted based on the position of the best individual in the group. The formula is: , represents the hyperparameter combination of the mth individual in the tth generation, , and Respectively represent the hyperparameter combinations of the best individual, the second best individual, and the third best individual in the current population, , and They represent the random coefficients of the hyperparameter combinations of the best individual, the second best individual, and the third best individual in the current population, respectively, and control the update amplitude. t represents the number of iterations. Step U6: Repeat step U5 until the maximum number of iterations is reached and output the optimal hyperparameter combination.

[0011] Furthermore, the specific method of setting a threshold based on the virtual item evaluation parameter and the virtual item repurchase rate index data to determine whether the virtual item is suitable for sharing includes: The repurchase rate index data, appearance score feature data, function score feature data, and user experience score feature data are standardized and time-stamped, and the weighted average method is used to calculate the comprehensive index data. ,in, Represents comprehensive indicator data, , , and They represent repurchase rate index data, appearance score feature data, function score feature data, and user experience score feature data, respectively. , , and Respectively represent the weights of the repurchase rate index data, appearance score feature data, function score feature data, and user experience score feature data; Calculate the average value of historical comprehensive indicator data and standard deviation , and set the threshold by the mean and standard deviation , where k represents the adjustment coefficient, ; When comprehensive indicator data Less than threshold When the comprehensive index data is Greater than or equal to threshold , it means that the comprehensive index of the item is high and it is shared.

[0012] Furthermore, the specific methods of using blockchain technology to share virtual items suitable for sharing include: Use ERC-721 smart contracts to create a unique digital identity NFT for each virtual item in the metaverse. The NFT contains the unique identifier of the virtual item and the historical information of the virtual item. Before sharing, the smart contract calls off-chain data to dynamically obtain the comprehensive index data of the item. When the comprehensive index data of the item is higher than the set threshold, the smart contract will automatically allow the item to enter the sharing process; During the sharing process, NFT records the blockchain address and access time limit information of the authorized user, and automatically manages the sharing permissions through smart contracts: within the valid time range, users can access the shared content by unlocking off-chain resources; After the sharing expires, the smart contract automatically revokes the user's access rights and updates the usage record to the blockchain to form a complete sharing history.

[0013] A method for intelligently sharing metaverse data, which is implemented by a metaverse data intelligent sharing platform, comprises: Step S1, collecting user behavior parameters, virtual item attribute parameters and virtual item evaluation parameters; Step S2: Use interpolation to process missing values ​​of user behavior parameters and virtual item attribute parameters, and then adaptively Perform outlier processing and obtain preprocessed data sets; Step S3, inputting the preprocessed data set into the improved XGBoost model to predict the repurchase rate index data of the virtual items; Step S4: based on the virtual item evaluation parameters and the repurchase rate index data of the virtual item, a threshold is set to determine whether the virtual item is suitable for sharing; Step S5: Use blockchain technology to share virtual items that are suitable for sharing.

[0014] Technical effects and advantages of a metaverse data intelligent sharing platform and method of the present invention: The present invention obtains the scarcity characteristic data of virtual items through dynamic time window technology, thereby ensuring that sharing decisions are based on real-time market data and reflect the actual demand for items; interpolation method is used to handle missing values ​​and adaptive The method can handle outliers and effectively address the shortcomings of traditional methods when data is incomplete or abnormal. It can automatically adjust under complex data distribution and dynamic changes to improve data quality.

[0015] Aiming at the prediction of repurchase rate of virtual items, the present invention introduces an improved XGBoost model. By combining the evaluation parameters of multiple dimensions of appearance scoring feature data, function scoring feature data and user experience scoring feature data, the weighted average method is used to calculate the comprehensive index, thereby avoiding the limitation of single feature on decision-making, thereby making the sharing decision of virtual items more comprehensive and accurate; the transparency, security and non-tamperability of the sharing process are guaranteed by blockchain technology, ensuring the authenticity of the data and the privacy protection of users.

[0016] Through multi-dimensional data processing, precise sharing judgment mechanism and application of blockchain technology, the present invention effectively improves the intelligence level of the Metaverse virtual item sharing platform, and provides strong technical support for the safe and transparent sharing of virtual items. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a schematic diagram of a Metaverse data intelligent sharing platform of the present invention; Figure 2 This is a schematic diagram of a metaverse data intelligent sharing method of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Example 1, please refer to Figure 1 As shown, the Metaverse data intelligent sharing platform described in this embodiment includes: a data acquisition module: acquiring user behavior parameters, virtual item attribute parameters and virtual item evaluation parameters; Data processing module: Based on user behavior parameters and virtual item attribute parameters, use interpolation method to process missing values ​​and use adaptive Perform outlier processing to obtain a preprocessed data set; Data prediction module: for the preprocessed data set, the improved XGBoost model is used to predict the repurchase rate index data of virtual items; Judgment module: based on the virtual item evaluation parameters and the virtual item repurchase rate index data, set the threshold to determine whether the virtual item is suitable for sharing; Data sharing module: For virtual items suitable for sharing, blockchain technology is used for sharing.

[0020] User behavior parameters include: visit volume characteristic data, collection volume characteristic data, and like volume characteristic data; Virtual item attribute parameters include: price feature data and scarcity feature data; Virtual item evaluation parameters include: appearance scoring feature data, function scoring feature data, and user experience scoring feature data.

[0021] Whenever a user clicks on a page or details of a virtual item, the system will record this visit through the tracking code embedded in the front-end or the back-end log, generate an access record and store it in the database. By querying the access records in the database, the number of visits to each Metaverse virtual item at a certain point in time can be obtained; Each Metaverse virtual item has a collection button. When a user clicks the collection button, the system will record the action in the database. By querying the records in the user collection table, the collection quantity of each Metaverse virtual item at a certain point in time can be obtained. Each Metaverse virtual item has a like button. When a user clicks the like button, the system will record the behavior through event tracking and store the like information in the like record table in the database. By querying the records in the table, the number of likes for each Metaverse virtual item at a certain point in time can be obtained; When a user purchases a virtual item, the price at the time of the transaction will be recorded. These price data will be stored in the transaction record table. By querying the transaction table, the transaction price of each Metaverse virtual item at a certain point in time can be obtained; On the platform, after users browse or purchase virtual items, the platform will provide a rating option. Users can rate the items based on their appearance, visual effects, etc. The rating is generally presented in the form of stars. Each time a user rates the appearance, the rating data will be recorded in the platform's database, and the appearance rating of the virtual item at a certain point in time can be obtained; After using Metaverse virtual items, users will rate their functionality, practicality, and operating experience. The platform will prompt users to rate virtual items on the virtual item page or after use. The rating data will be stored in the database, reflecting the performance of the item in terms of functionality. The functional rating of the virtual item at a certain point in time can be obtained; The user experience score is a measure of the user's overall experience in using virtual items, and usually includes the user's evaluation of the item's ease of use, ease of operation, and sense of immersion. The platform usually provides feedback and rating options after the user purchases and uses the virtual item. Users evaluate their usage experience and can obtain the user experience score of the virtual item at a certain point in time.

[0022] The methods for obtaining scarcity feature data include: The scarcity feature data calculation method relies on the inventory volume, transaction frequency and price fluctuation data obtained within a fixed time, ignoring the changes in market conditions. The dynamic time window can be adjusted continuously according to the latest data to reflect market changes in real time, ensuring that the calculated scarcity feature data is more timely and accurate. By setting the starting point and the latest time point of the time window, the data weight of each time point can be dynamically adjusted. The closer time point is, the greater the contribution to the scarcity feature data, thereby responding to market fluctuations more flexibly. Use dynamic time windows to calculate the scarcity feature data of virtual items in the Metaverse and define the range of the time window , Indicates the latest time point. Indicates the starting time point of the time window; Obtain the inventory volume, transaction frequency and price fluctuation data at each time point in the time window and perform normalization processing; In the time window, the inventory, transaction frequency and price fluctuation data obtained at different time points have different contributions to the scarcity feature data. For the inventory, transaction frequency and price fluctuation data at different time points, the weights are set through the attenuation function: ,in represents the initial weight, represents the attenuation coefficient, Indicates the distance from the current time to the latest time; The scarcity characteristic data at time point b is calculated by weighted average method, and the formula is: ,in, , and Respectively represent the weights of inventory volume, transaction frequency and price fluctuation data, , and They represent the inventory volume, transaction frequency and price fluctuation data values ​​obtained at time point b respectively; The inventory of virtual items refers to the number of items available for users to purchase, use or trade at a specific point in time. The inventory data is obtained through the platform's inventory management, which records the inventory of each item at each point in time; Transaction frequency refers to the number of times a virtual item is traded at a certain point in time. This data can be obtained by analyzing the transaction records of the platform; The price fluctuation of virtual items is reflected in the changes in the transaction price of the items. The difference between the price data obtained at the current moment and the price data obtained at the previous moment represents the price fluctuation amount at the current moment and is stored in the platform's database.

[0023] Ways to use interpolation to handle missing values ​​include: The user behavior parameters and virtual item attribute parameters include five feature data: number of visits, number of favorites, number of likes, price, and scarcity. The isnull data processing method is used to traverse the five feature data respectively to find the location of the missing values ​​in each feature data. True represents a missing value, and False represents a non-missing value. For True, the forward filling method is used for processing. When the missing value is at the beginning of the feature data, the backward filling method is used for missing value processing. Both forward filling and backward filling are a type of interpolation method.

[0024] Adaptive The specific methods for handling outliers include: For the data after missing value processing, arrange each feature data value in ascending order and use the box plot method to calculate the first quartile and the third quartile , and the interquartile range ; Dynamically adjust by calculating the skewness of each feature data value , for the Feature data set ,in, Indicates The formula for the skewness of the i-th data value in a feature group is: , where n represents the The number of feature data, Indicates The average value of feature data, s represents the The standard deviation of the feature data; When the skewness is equal to 0, Feature data set It belongs to normal distribution, and the upper and lower limits of outliers are: ; ; When the skewness is greater than 0, the characteristic data set Belongs to the right skewed distribution. When the skewness is less than 0, the characteristic data group Belongs to the left skewed distribution, adjusted according to the skewness of the feature data group , and calculate the upper and lower limits of the outliers: , ,in is the adjustment coefficient, when the characteristic data set If the data value in is not within the upper and lower limits of the outlier, it is an outlier. The outlier is identified and replaced with the previous non-outlier data value using the forward filling method.

[0025] The specific methods of using the improved XGBoost model to predict the repurchase rate index data of virtual items include: When dealing with the prediction of virtual item repurchase rate indicators, the XGBoost model usually relies on fixed hyperparameter combinations and standard objective functions, such as mean square error, which cannot fully capture the complexity of the repurchase rate and the relationship between different features. The Gray Wolf Optimization Algorithm is introduced to automatically search for the optimal hyperparameter combination to better control overfitting, and dynamically adjust the model through an iterative optimization process to improve prediction accuracy. Step E1: The sample set includes L_o groups of samples, each group of samples includes a preprocessed data set and corresponding virtual item repurchase rate index data; wherein the data contained in the preprocessed data set is used as a feature variable, and the virtual item repurchase rate index data is used as a target variable; Step E2: Use the Gray Wolf Optimization Algorithm to search for the optimal hyperparameter combination of the XGBoost model; Step E3: Define the final objective function as minimizing the loss function value. The loss function value is: , among which It represents the predicted value of the repurchase rate index of the csth sample. A sample refers to a row of data corresponding to the feature variable and the target variable. represents the true value of the repurchase rate indicator of the csth sample, represents the number of samples, Indicates The weight of a leaf node, Indicates the number of leaf nodes, represents L1 regularization, represents the L1 regularization coefficient, represents L2 regularization, represents the L2 regularization coefficient; Step E4: Initialize the predicted value and use the mean of the target variable repurchase rate indicator data to define the initial predicted value: ,in, Represents the initialized predicted value of the repurchase rate indicator; Step E5: Calculate the residual. In each round of iteration, calculate the residual between the current round and the previous round: ,in, Indicates The residual of the wheel, Indicates The predicted value of the round; Step E6: In each round, a decision tree is constructed to fit the residual, and each feature data partition point is calculated. The partition point with the largest gain is selected as the current optimal partition point, and the data sample is divided into two sub-nodes, the left sub-tree and the right sub-tree, and each sub-tree is divided using the gain until the maximum depth of the tree is reached; In each round, a decision tree is constructed to fit the residual, and each feature data partition point is calculated. The partition point with the largest gain is selected as the current optimal partition point, and the data sample is divided into two sub-nodes, the left sub-tree and the right sub-tree, and each sub-tree is divided using the gain until the maximum depth of the tree is reached. The split point is the position where the decision tree splits on each feature. Assuming that the split point is the historical visit volume feature data, by calculating the gain of each data value, the data value with the largest gain is selected as the split point, and the historical visit volume feature data is divided into a left subtree and a right subtree; The gain formula is: in, and denote the sum of the first-order gradients of the left subtree and the right subtree respectively, and Represents the sum of the second-order gradients of the left subtree and the right subtree, respectively. represents the L2 regularization coefficient, represents the minimum gain threshold for controlling splitting; Among them, the first-order gradient is the first-order derivative of the final objective function relative to the predicted value. In each round of iteration, the model is updated according to the prediction error of the previous round. The direction and magnitude of this error can be measured by the first-order gradient. The first-order gradient is usually expressed as To indicate that for sample cs, it represents the derivative of the final objective function to the predicted value, the formula is: ; Among them, the second-order gradient is the second-order derivative of the final objective function relative to the predicted value, which reflects the curvature of the error or the acceleration of the error change. The second-order gradient is usually expressed as To express it, the formula is: ; Step E7: Repeat the iteration. In each round of iteration, the prediction value of the model is updated through the newly generated tree until the condition of the final objective function is met, and then the iteration is stopped; The prediction value formula for updating the model is: ,in, Indicates The predicted value of the round cs sample, represents the predicted value of the tree trained in the tth round, Represents the learning rate.

[0026] The specific methods of using the Gray Wolf Optimization Algorithm to search for the optimal hyperparameter combination of the XGBoost model include: Step U1: Determine the XGBoost hyperparameter combination, including the maximum depth of the tree, the learning rate, the number of trees, the proportion of samples used in each training, the proportion of features randomly selected for each tree, the minimum loss reduction required for splitting, the L2 regularization term, and the L1 regularization term; Step U2: Define the mean square error as the objective function to evaluate the quality of each set of hyperparameter combinations. ; Among them, good means a small MSE value: assuming that a certain set of hyperparameter combinations can make the model's prediction results close to the true value, that is, the prediction error is small, then the "goodness" of this hyperparameter combination is higher, so a hyperparameter combination with a smaller MSE is usually considered to be better; bad means a large MSE value: assuming that the model prediction error of a certain set of hyperparameter combinations is large, that is, the mean square error is large, then the "goodness" of this hyperparameter combination is low, and the model's prediction results are far from the true value, so this hyperparameter combination is usually considered to be worse; Step U3: Initialize the gray wolf individuals. Randomly initialize M gray wolf individuals, each of which represents a hyperparameter combination: , where m represents the index of the individual gray wolf, represents the maximum depth of the tree in the mth gray wolf individual, represents the learning rate of the mth gray wolf individual, represents the number of trees in the mth gray wolf individual, represents the sample ratio used in each training of the mth gray wolf individual, represents the proportion of features randomly selected from each tree in the mth gray wolf individual, represents the minimum loss reduction required for the split in the mth gray wolf individual, L2 represents the regularization term in the mth gray wolf individual, L1 represents the regularization term in the mth gray wolf individual; Step U4: Evaluate the fitness of individual gray wolves. Each gray wolf individual represents a set of hyperparameter combinations. Use this hyperparameter combination to train the XGBoost model and calculate the mean square error. , as the fitness value for evaluating this hyperparameter combination, the goal is to minimize ,therefore The smaller it is, the higher the fitness value; Step U5: Determine the current best individual, second best individual, and third best individual based on the fitness value. The position of each gray wolf individual will be adjusted based on the position of the best individual in the group. The formula is: , represents the hyperparameter combination of the mth individual in the tth generation, , and Respectively represent the hyperparameter combinations of the best individual, the second best individual, and the third best individual in the current population, , and They represent the random coefficients of the hyperparameter combinations of the best individual, the second best individual, and the third best individual in the current population, respectively, and control the update amplitude. t represents the number of iterations. Step U6: Repeat step U5 until the maximum number of iterations is reached and output the optimal hyperparameter combination.

[0027] Based on the virtual item evaluation parameters and the virtual item repurchase rate index data, the threshold is set. The specific methods for judging whether the virtual item is suitable for sharing include: Based on the repurchase rate index data, appearance score feature data, function score feature data and user experience score feature data, standardization and unified timestamp processing are performed, and the comprehensive index data is calculated using the weighted average method ,in, Represents comprehensive indicator data, , , and They represent repurchase rate index data, appearance score feature data, function score feature data, and user experience score feature data, respectively. , , and Respectively represent the weights of the repurchase rate index data, appearance score feature data, function score feature data, and user experience score feature data; Calculate the average value of historical comprehensive indicator data and standard deviation , and set the threshold by the mean and standard deviation , where k represents the adjustment coefficient, ; average value , H1 represents the total number of historical comprehensive indicator data, Indicates the number index; Standard Deviation ; When comprehensive indicator data Less than threshold When the comprehensive index data is Greater than or equal to threshold When , it means that the comprehensive index of the item is high and it is shared; The comprehensive index calculated by standardization and weighted average method is a comprehensive reflection of the multi-dimensional evaluation results of Metaverse virtual items, reflecting the overall performance of the item in terms of user repurchase, appearance, function and experience; if the comprehensive index is greater than the set threshold, it means that the virtual item performs well in multiple important dimensions and is generally considered to be more attractive and higher-value items, suitable for sharing with other users or a wider market; the setting of the threshold is determined based on the mean and standard deviation of historical data, ensuring that the sharing decision has certain historical data support and rationality, and avoiding the sharing of virtual items that perform poorly in all aspects or are not attractive enough.

[0028] For virtual items suitable for sharing, specific ways to share using blockchain technology include: Use ERC-721 smart contracts to create a unique digital identity NFT for each virtual item in the metaverse. The NFT contains the unique identifier of the virtual item and the historical information of the virtual item. Before sharing, the smart contract calls off-chain data to dynamically obtain the comprehensive index data of the item. When the comprehensive index data of the item is higher than the set threshold, the smart contract will automatically allow the item to enter the sharing process; During the sharing process, NFT records the blockchain address and access time limit information of the authorized user, and automatically manages the sharing permissions through smart contracts: within the valid time range, users can access the shared content by unlocking off-chain resources; After the sharing expires, the smart contract automatically revokes the user's access rights and updates the usage record to the blockchain to form a complete sharing history; Off-chain data refers to data that is not directly stored on the blockchain. In contrast to on-chain data, off-chain data refers to data stored outside the blockchain and managed by external systems. Although these data are not stored on the blockchain, they may be closely related to operations, transactions, and smart contracts in the blockchain.

[0029] In this embodiment, the scarcity feature data of virtual items is obtained through dynamic time window technology, so as to ensure that the sharing decision is based on real-time market data and reflects the actual demand for items; interpolation method is used to handle missing values ​​and adaptive The method can handle outliers and effectively address the shortcomings of traditional methods when data is incomplete or abnormal. It can automatically adjust in the case of complex data distribution and dynamic changes to improve data quality. In order to predict the repurchase rate of virtual items, the present invention introduces an improved XGBoost model. By combining the evaluation parameters of multiple dimensions of appearance scoring feature data, function scoring feature data and user experience scoring feature data, the weighted average method is used to calculate the comprehensive index, thereby avoiding the limitation of a single feature on decision-making, thereby making the sharing decision of virtual items more comprehensive and accurate. The transparency, security and non-tamperability of the sharing process are guaranteed by blockchain technology, ensuring the authenticity of the data and the privacy protection of users. Through multi-dimensional data processing, precise sharing judgment mechanism and application of blockchain technology, the present invention effectively improves the intelligence level of the Metaverse virtual item sharing platform, and provides strong technical support for the safe and transparent sharing of virtual items.

[0030] Example 2, please refer to Figure 2 As shown, the part not described in detail in this embodiment is described in Example 1, which provides a method for intelligent sharing of Metaverse data, including: Step S1, collecting user behavior parameters, virtual item attribute parameters and virtual item evaluation parameters; Step S2: Use interpolation to process missing values ​​of user behavior parameters and virtual item attribute parameters, and then adaptively Perform outlier processing and obtain preprocessed data sets; Step S3, inputting the preprocessed data set into the improved XGBoost model to predict the repurchase rate index data of the virtual items; Step S4: based on the virtual item evaluation parameters and the repurchase rate index data of the virtual item, a threshold is set to determine whether the virtual item is suitable for sharing; Step S5: Use blockchain technology to share virtual items that are suitable for sharing.

[0031] Embodiment 3. This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the operation mode of the above-mentioned metaverse data intelligent sharing platform and method is implemented.

[0032] Since the electronic device introduced in this embodiment is an electronic device used to implement a metaverse data intelligent sharing platform and method in the embodiment of the present application, based on the metaverse data intelligent sharing platform and method introduced in the embodiment of the present application, a person skilled in the art can understand the specific implementation of the electronic device of the present embodiment and its various variations, so how the electronic device implements the method in the embodiment of the present application is not described in detail here. As long as a person skilled in the art implements the electronic device used in a metaverse data intelligent sharing platform and method in the embodiment of the present application, it belongs to the scope of protection of this application.

[0033] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0034] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A Metaverse data intelligent sharing platform, characterized in that: include: Data acquisition module: obtains user behavior parameters, virtual item attribute parameters and virtual item evaluation parameters; Data processing module: Based on user behavior parameters and virtual item attribute parameters, use interpolation method to process missing values ​​and use adaptive Perform outlier processing to obtain a preprocessed data set; Data prediction module: for the preprocessed data set, the improved XGBoost model is used to predict the repurchase rate index data of virtual items; Judgment module: based on the virtual item evaluation parameters and the virtual item repurchase rate index data, set the threshold to determine whether the virtual item is suitable for sharing; Data sharing module: For virtual items suitable for sharing, blockchain technology is used for sharing.

2. A Metaverse data intelligent sharing platform according to claim 1, characterized in that: User behavior parameters include: visit volume characteristic data, collection volume characteristic data, and like volume characteristic data; Virtual item attribute parameters include: price feature data and scarcity feature data; Virtual item evaluation parameters include: appearance scoring feature data, function scoring feature data, and user experience scoring feature data.

3. A Metaverse data intelligent sharing platform according to claim 2, characterized in that: The method for obtaining the scarcity characteristic data includes: Use dynamic time windows to calculate the scarcity feature data of virtual items in the Metaverse and define the range of the time window , Indicates the latest time point. Indicates the starting time point of the time window; Obtain the inventory volume, transaction frequency and price fluctuation data at each time point in the time window and perform normalization processing; In the time window, for the inventory volume, transaction frequency and price fluctuation data at different time points, the time decay weight is set through the decay function: ,in represents the initial weight, represents the attenuation coefficient, Indicates the distance from the current time to the latest time; The scarcity characteristic data at time point b is calculated by weighted average method, and the formula is: ,in, , and Respectively represent the weights of inventory volume, transaction frequency and price fluctuation data, , and They respectively represent the inventory volume, transaction frequency and price fluctuation data values ​​obtained at time point b.

4. A Metaverse data intelligent sharing platform according to claim 3, characterized in that: The method of using interpolation to process missing values ​​includes: The user behavior parameters and virtual item attribute parameters include five feature data: number of visits, number of favorites, number of likes, price, and scarcity. The isnull data processing method is used to traverse the five feature data respectively to find the location of the missing values ​​in each feature data. True represents a missing value, and False represents a non-missing value. For True, the forward filling method is used for processing. When the missing value is at the beginning of the feature data, the backward filling method is used for missing value processing.

5. A Metaverse data intelligent sharing platform according to claim 4, characterized in that: The adaptive The specific methods for handling outliers include: For the data after missing value processing, arrange each feature data value in ascending order and use the box plot method to calculate the first quartile and the third quartile , and the interquartile range ; Dynamically adjust by calculating the skewness of each feature data value , for the Feature data set ,in, Indicates The formula for the skewness of the i-th data value in a feature group is: , where n represents the The number of feature data, Indicates The average value of feature data, s represents the The standard deviation of the feature data; When the skewness is equal to 0, Feature data set It belongs to normal distribution, and the upper and lower limits of outliers are: ; ; When the skewness is greater than 0, the characteristic data set Belongs to the right skewed distribution. When the skewness is less than 0, the characteristic data group Belongs to the left skewed distribution, adjusted according to the skewness of the feature data group , and calculate the upper and lower limits of the outliers: , ,in is the adjustment coefficient, when the characteristic data set If the data value in is not within the upper and lower limits of the outlier, it is an outlier. The outlier is identified and replaced with the previous non-outlier data value using the forward filling method.

6. A Metaverse data intelligent sharing platform according to claim 5, characterized in that: The specific method of using the improved XGBoost model to predict the repurchase rate index data of virtual items includes: Step E1: The sample set includes L_o groups of samples, each group of samples includes a preprocessed data set and corresponding virtual item repurchase rate index data; wherein the data contained in the preprocessed data set is used as a feature variable, and the virtual item repurchase rate index data is used as a target variable; Step E2: Use the Gray Wolf Optimization Algorithm to search for the optimal hyperparameter combination of the XGBoost model; Step E3: define the final objective function as minimizing the loss function value; Step E4: Initialize the predicted value and use the mean of the target variable repurchase rate indicator data to define the initial predicted value: ,in, Represents the initialized predicted value of the repurchase rate indicator; Step E5: Calculate the residual. In each round of iteration, calculate the residual between the current round and the previous round: ,in, represents the residual of the tth round, Indicates The predicted value of the round; Step E6: In each round, a decision tree is constructed to fit the residual, and each feature data partition point is calculated. The partition point with the largest gain is selected as the current optimal partition point, and the data sample is divided into two sub-nodes, the left sub-tree and the right sub-tree, and each sub-tree is divided using the gain until the maximum depth of the tree is reached; Step E7: Repeat the iteration. In each round of iteration, the prediction value of the model is updated through the newly generated tree until the condition of the final objective function is met, and then the iteration is stopped.

7. A Metaverse data intelligent sharing platform according to claim 6, characterized in that: The specific method of using the Gray Wolf Optimization Algorithm to search for the optimal hyperparameter combination of the XGBoost model includes: Step U1: Determine the XGBoost hyperparameter combination, including the maximum depth of the tree, the learning rate, the number of trees, the proportion of samples used in each training, the proportion of features randomly selected for each tree, the minimum loss reduction required for splitting, the L2 regularization term, and the L1 regularization term; Step U2: Define the mean square error as the objective function to evaluate the quality of each set of hyperparameter combinations; Step U3: Initialize the gray wolf individuals, randomly initialize M gray wolf individuals, each individual represents a hyperparameter combination; Step U4: Evaluate the fitness of individual gray wolves. Each gray wolf individual represents a set of hyperparameter combinations. Use this hyperparameter combination to train the XGBoost model and calculate the mean square error. , as the fitness value for evaluating this hyperparameter combination, the goal is to minimize ,therefore The smaller it is, the higher the fitness value; Step U5: Determine the current best individual, second best individual, and third best individual based on the fitness value. The position of each gray wolf individual will be adjusted based on the position of the best individual in the group. The formula is: , represents the hyperparameter combination of the mth individual in the tth generation, , and Respectively represent the hyperparameter combinations of the best individual, the second best individual, and the third best individual in the current population, , and They represent the random coefficients of the hyperparameter combinations of the best individual, the second best individual, and the third best individual in the current population, respectively, and control the update amplitude. t represents the number of iterations. Step U6: Repeat step U5 until the maximum number of iterations is reached and output the optimal hyperparameter combination.

8. The Metaverse data intelligent sharing platform according to claim 7, characterized in that: The specific method of setting a threshold based on the virtual item evaluation parameter and the virtual item repurchase rate index data to determine whether the virtual item is suitable for sharing includes: Based on the repurchase rate index data, appearance score feature data, function score feature data and user experience score feature data, standardization and unified timestamp processing are performed, and the comprehensive index data is calculated using the weighted average method ,in, Represents comprehensive indicator data, , , and They represent repurchase rate index data, appearance score feature data, function score feature data, and user experience score feature data, respectively. , , and Respectively represent the weights of the repurchase rate index data, appearance score feature data, function score feature data, and user experience score feature data; Calculate the average value of historical comprehensive indicator data and standard deviation , and set the threshold by the mean and standard deviation , where k represents the adjustment coefficient, ; When comprehensive indicator data Less than threshold When the comprehensive index data is Greater than or equal to threshold , it means that the comprehensive index of the item is high and it is shared.

9. The Metaverse data intelligent sharing platform according to claim 8, characterized in that: The specific ways of sharing virtual items suitable for sharing using blockchain technology include: Use ERC-721 smart contracts to create a unique digital identity NFT for each virtual item in the metaverse. The NFT contains the unique identifier of the virtual item and the historical information of the virtual item. Before sharing, the smart contract calls off-chain data to dynamically obtain the comprehensive index data of the item. When the comprehensive index data of the item is higher than the set threshold, the smart contract will automatically allow the item to enter the sharing process; During the sharing process, NFT records the blockchain address and access time limit information of the authorized user, and automatically manages the sharing permissions through smart contracts: within the valid time range, users can access the shared content by unlocking off-chain resources; After the sharing expires, the smart contract automatically revokes the user's access rights and updates the usage record to the blockchain to form a complete sharing history.

10. A method for intelligent sharing of Metaverse data, based on a Metaverse data intelligent sharing platform according to any one of claims 1 to 9, characterized in that: include: Step S1, collecting user behavior parameters, virtual item attribute parameters and virtual item evaluation parameters; Step S2: Use interpolation to process missing values ​​of user behavior parameters and virtual item attribute parameters, and then adaptively Perform outlier processing and obtain preprocessed data sets; Step S3, inputting the preprocessed data set into the improved XGBoost model to predict the repurchase rate index data of the virtual items; Step S4: based on the virtual item evaluation parameters and the repurchase rate index data of the virtual item, a threshold is set to determine whether the virtual item is suitable for sharing; Step S5: Use blockchain technology to share virtual items that are suitable for sharing.

Images