Methods, devices, and storage media for providing financial services based on the metaverse
By constructing virtual financial institutions in the metaverse and utilizing target models trained through federated learning, combined with data from real-world branches, the problem of low accuracy in financial services within the metaverse is solved, thereby improving personalized services and data security.
Patent Information
- Application Number
- CN202310792753.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-06-29
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-06-29
AI Technical Summary
Financial services provided in the metaverse are less accurate and cannot provide personalized services to users based on user data.
By constructing virtual branches of financial institutions in the metaverse, risk control models and product recommendation models are trained using federated learning. Combined with data from real-world branches, customer identity and transaction information are obtained and input into the target model to provide personalized financial services.
It improves the accuracy and personalization of financial services in the metaverse, enhances the customer experience, and protects customer data privacy and reduces the risk of data breaches through federated learning.
Smart Images

Figure CN116579852B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence, and more specifically, to a method, apparatus, and storage medium for providing financial services based on a metaverse. Background Technology
[0002] With the rapid development of internet technology, developing businesses within the metaverse has gradually become a trend. The fusion of the virtual and real worlds of the metaverse has brought new business models and development opportunities to various industries. The metaverse is a digital ecosystem that tightly connects the real and virtual worlds, providing immersive interactive experiences that allow users to engage in various activities such as socializing, entertainment, and shopping within a virtual environment. However, the application of the metaverse in the financial industry is still in its early stages and requires further exploration and development.
[0003] However, among related technologies, the financial services provided in the metaverse are less flexible and cannot offer personalized services based on user data, resulting in lower accuracy of the financial services provided to users. With the development of technologies such as big data and artificial intelligence, traditional business models and service methods are no longer sufficient to meet customer needs.
[0004] There is currently no effective solution to the problem of low accuracy in financial services provided in the metaverse by related technologies. Summary of the Invention
[0005] The main objective of this application is to provide a method, apparatus, and storage medium for providing financial services based on the metaverse, in order to solve the problem of low accuracy of financial services provided in the metaverse in related technologies.
[0006] To achieve the above objectives, according to one aspect of this application, a method for providing financial services based on a metaverse is provided. The method includes: acquiring financial services to be provided by a financial institution; determining a target model based on the financial services to be provided, wherein the target model is of at least one of the following types: a risk control model and a product recommendation model; the target model is jointly trained using federated learning based on sample data from multiple branches of the financial institution; acquiring the financial institution's customer's first identity information, first transaction behavior information, second identity information, and second transaction behavior information, wherein the first identity information and first transaction behavior information are information stored in the financial institution's database, and the second identity information and second transaction behavior information are information stored in a virtual financial institution for a virtual character in the metaverse, where the virtual character is a character created by the customer in the metaverse, and the virtual financial institution is a branch created by the financial institution in the metaverse; inputting the first identity information, first transaction behavior information, second identity information, and second transaction behavior information into the target model to obtain the model output; and providing the customer with a preset financial service based on the model output.
[0007] Optionally, the target model is obtained by: identifying multiple branches that need to provide financial services from all branches of the financial institution; training each branch based on its local dataset to obtain a local model for the branch, and determining the model parameters of the local model to obtain a set of model parameters, wherein each branch's local model corresponds to one model parameter; processing the set of model parameters using a federated averaging algorithm to obtain the processed target model parameters; and determining the target model based on a preset model framework and the target model parameters.
[0008] Optionally, the target model is a risk control model. Training is performed on each branch based on its local dataset to obtain a local model for that branch. Determining the model parameters of the local model includes: identifying the local dataset for each branch and extracting multiple sets of training samples from it. Each set of training samples includes the customer's first identity information, second identity information, first historical transaction behavior information, second historical transaction behavior information, and historical risk assessment value from the local dataset. The neural network model is iteratively trained using these multiple sets of training samples to obtain a set of undetermined models. Each iteration trains the neural network model by randomly extracting multiple sets of training samples from the local dataset. The minimum loss function is determined from the loss functions of the undetermined models. The minimum loss function is processed based on preset parameters and regularization coefficients to obtain the processed target loss function, which is then used as the model parameters of the local model.
[0009] Optionally, the target model is a product recommendation model. Training is performed on each branch based on its local dataset to obtain a local model for that branch. Determining the model parameters of the local model includes: identifying the local dataset for each branch and extracting multiple sets of training samples from it. Each set of training samples includes the customer's first identity information, second identity information, first historical transaction behavior information, second historical transaction behavior information, and historical recommended products from the local dataset. The neural network model is iteratively trained using these multiple sets of training samples to obtain a set of undetermined models. Each iteration trains the neural network model by randomly extracting multiple sets of training samples from the local dataset. The minimum loss function is determined from the loss functions of the undetermined models. The minimum loss function is processed based on preset parameters and regularization coefficients to obtain the processed target loss function, which is then used as the model parameters of the local model.
[0010] Optionally, the target model parameters are processed using a federated averaging algorithm, including: calculating the total amount of data in all local datasets of multiple branches, and calculating the ratio of the amount of data in the local dataset of each branch to the total amount of data to obtain the target weights of the branches; weighted summing of the target loss functions and target weights of the local models of multiple branches to obtain the loss function of the target model; and determining the loss function of the target model as the target model parameters.
[0011] Optionally, obtaining the first identity information and first transaction behavior information of a financial institution's customers includes: obtaining all types of customer identity information and transaction behavior information from a database, wherein the types of identity information include at least one of the following: age, job title, assets, and interests, and the types of transaction behavior information include at least one of the following: transaction frequency, transaction amount, transaction cycle, and transaction method; calculating the first information gain for each type of identity information and calculating the second information gain for each type of transaction behavior information; identifying identity information whose first information gain is greater than or equal to an information gain threshold as first identity information, and identifying transaction behavior information whose second information gain is greater than or equal to an information gain threshold as first transaction behavior information.
[0012] Optionally, calculating the first information gain for each type of identity information and the second information gain for each type of transaction behavior information includes: determining the first information entropy of all customers in the database transacting with financial institutions within a preset period; determining the second information entropy of customers in the database with target type identity information transacting with financial institutions within a preset period; determining the third information entropy of customers in the database with target type transaction behavior information transacting with financial institutions within a preset period; calculating the difference between the first and second information entropies to obtain the first information gain of the target type identity information; and calculating the difference between the first and third information entropies to obtain the second information gain of the target type transaction behavior information.
[0013] To achieve the above objectives, according to another aspect of this application, a financial service provision device based on a metaverse is provided. The device includes: a first acquisition unit, configured to acquire financial services to be provided by a financial institution, and determine a target model based on the financial services to be provided, wherein the target model is of at least one of the following types: a risk control model and a product recommendation model, and the target model is jointly trained using federated learning based on sample data from multiple branches of the financial institution; a second acquisition unit, configured to acquire first identity information, first transaction behavior information, second identity information, and second transaction behavior information of a customer of the financial institution, wherein the first identity information and first transaction behavior information are information stored in the database of the financial institution, and the second identity information and second transaction behavior information are information stored in a virtual financial institution for a virtual character in the metaverse, where the virtual character is a character created by the customer in the metaverse, and the virtual financial institution is a branch created by the financial institution in the metaverse; an input unit, configured to input the first identity information, first transaction behavior information, second identity information, and second transaction behavior information into the target model to obtain the model output result; and a service unit, configured to provide the customer with a preset financial service based on the model output result.
[0014] This application employs the following steps: First, it obtains the financial services to be provided by a financial institution; second, it determines a target model based on these financial services. The target model includes at least one of the following types: a risk control model and a product recommendation model. The target model is trained using federated learning on sample data from multiple branches of the financial institution. Third, it obtains the first identity information, first transaction behavior information, second identity information, and second transaction behavior information of the financial institution's customers. The first identity information and first transaction behavior information are stored in the financial institution's database, while the second identity information and second transaction behavior information are stored in a virtual financial institution within a virtual universe. The virtual character is a role created by the customer in the metaverse, and the virtual financial institution is a branch created by the financial institution within the metaverse. Fourth, it inputs the first identity information, first transaction behavior information, second identity information, and second transaction behavior information into the target model to obtain the model output. Finally, it provides the customer with pre-defined financial services based on the model output, thus solving the problem of low accuracy in financial services provided in the metaverse in related technologies. By building a federated learning platform connecting virtual financial institutions in the metaverse and various branches in the real world, and training the target model for providing financial services using federated learning, the accuracy of financial services provided in the metaverse is improved. Attached Figure Description
[0015] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:
[0016] Figure 1 This is a flowchart of a method for providing financial services based on the metaverse, according to an embodiment of this application.
[0017] Figure 2 This is a flowchart of an optional metaverse-based financial service provision method provided according to embodiments of this application;
[0018] Figure 3 This is a schematic diagram of a metaverse-based financial service provisioning device provided according to an embodiment of this application;
[0019] Figure 4 This is a schematic diagram of an electronic device provided according to an embodiment of this application. Detailed Implementation
[0020] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.
[0021] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.
[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0023] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties.
[0024] For ease of description, the following explains some of the nouns or terms used in the embodiments of this application:
[0025] The Metaverse is a virtual world comprised of various technologies, including virtual reality, blockchain, and artificial intelligence. It is a digital and virtualized environment where people can engage in virtual interactions, digital asset transactions, virtual property management, and digital identity verification.
[0026] Federated Learning is a data privacy-preserving distributed machine learning method that trains models locally on various nodes and shares model parameters, enabling learning from the entire dataset without revealing the original data. It allows for the collaborative training of a global model without sharing the original data. In federated learning, each participant holds its own local data, and the federated algorithm updates the local model before aggregating the model parameters into the global model. In the financial industry, especially banking, data security and privacy protection are paramount; therefore, federated learning plays a crucial role in risk control and fraud prevention.
[0027] Secure Multi-Party Computation (SMPC): Secure multi-party computation is a computing technique that protects data privacy. In secure multi-party computation, multiple participants use technologies such as encryption and distributed computing to perform computations and make decisions without exposing their data.
[0028] Information Gain: Information gain is a metric used for feature selection. It calculates the information gain of a feature by comparing the information entropy of the original dataset with that of the dataset after a certain feature partition, thereby determining which features are most important for the classification task.
[0029] Matrix Factorization (MF): Matrix factorization is a technique used to process matrix data. It constructs relationships in matrix data by decomposing a matrix into a product of multiple low-dimensional matrices.
[0030] Isolation Forest (iForest): Isolation Forest is an algorithm used for anomaly detection. It determines whether a data point is an anomaly by constructing multiple isolated trees and evaluating the average path length of a data point within each isolated tree.
[0031] The present invention will now be described in conjunction with preferred implementation steps. Figure 1 This is a flowchart of a method for providing financial services based on the metaverse, according to an embodiment of this application. Figure 1 As shown, the method includes the following steps:
[0032] Step S101: Obtain the financial services to be provided by the financial institutions, and determine the target model based on the financial services to be provided. The target model is of at least one of the following types: risk control model and product recommendation model. The target model is jointly trained by federated learning based on sample data from multiple branches of the financial institutions.
[0033] Specifically, financial services can include loan services, risk assessment services, and product recommendation services. The financial services to be provided can be those offered by virtual financial institutions within the metaverse. The target model includes, but is not limited to, risk control models and product recommendation models—models used to provide financial services to customers. The target model is determined based on the financial service to be provided; for example, if a product recommendation service is required, then the target model is determined to be a product recommendation model. The target model can be trained using federated learning based on sample datasets from various branches of the financial institution. By constructing virtual branches of financial institutions within the metaverse and utilizing the target model trained through federated learning to provide personalized services to customers, a better customer experience and data security can be achieved.
[0034] For example, by building a cross-bank federated learning platform, branches can share parameters of risk control and anti-fraud models while protecting customer data privacy. Each branch selects an appropriate machine learning model based on its own business needs and trains and optimizes the target model through federated learning. Utilizing virtual identity information from the metaverse can provide richer data features to the target model, improving its predictive accuracy. Regular updates to the target model can then address the ever-changing financial environment.
[0035] Step S102: Obtain the first identity information, first transaction behavior information, second identity information, and second transaction behavior information of the financial institution's customer. The first identity information and first transaction behavior information are information stored in the financial institution's database. The second identity information and second transaction behavior information are information stored in the virtual financial institution by the virtual character in the metaverse. The virtual character is a role created by the customer in the metaverse, and the virtual financial institution is a branch created by the financial institution in the metaverse.
[0036] Specifically, for customers who have transaction records with financial institutions and have created virtual avatars in the metaverse, virtual branches of financial institutions can be built in the metaverse to provide financial services such as account opening, fund transfers, and loans in virtual reality scenarios. These virtual branches utilize blockchain technology to ensure data security and transaction transparency, enhancing the customer experience. The primary identity information includes the customer's name, age, occupation, and interests; the secondary identity information includes the appearance and virtual occupation information of the virtual avatar created in the metaverse; the primary transaction behavior information includes the customer's transaction records in real-world financial institutions; and the secondary transaction behavior information includes the customer's transaction records in virtual financial institutions within the metaverse.
[0037] It should be noted that the virtual branches established in the metaverse include outlets, counters, and self-service machines, simulating the environment of real-world financial institutions. Intuitive and user-friendly interfaces are designed for these virtual branches to facilitate financial transactions for customers. The financial service modules of these virtual branches offer customers functions such as account opening, fund transfers, loans, and wealth management. Blockchain technology is used to ensure the security, transparency, and traceability of financial transactions.
[0038] Step S103: Input the first identity information, the first transaction behavior information, the second identity information, and the second transaction behavior information into the target model to obtain the model output results.
[0039] Specifically, by inputting the first identity information, the first transaction behavior information, the second identity information, and the second transaction behavior information into the target model, a model output result corresponding to the customer can be obtained. Based on the model output result, personalized services can be provided to the customer. For example, if the target model is a product recommendation model, the product to be recommended can be obtained as the model output result based on the first identity information, the first transaction behavior information, the second identity information, and the second transaction behavior information input by the customer. If the target model is a risk control model, the risk assessment value can be obtained as the model output result based on the first identity information, the first transaction behavior information, the second identity information, and the second transaction behavior information input by the customer.
[0040] Step S104: Provide customers with preset financial services based on the model output results.
[0041] Specifically, if the model output is a product to be recommended, then the virtual financial institution will push the product to be recommended to the customer's virtual avatar to provide financial services. If the model output is a risk assessment value, then the virtual financial institution will provide the customer's virtual avatar with a loan amount corresponding to the risk assessment value to provide financial services.
[0042] For example, by designing and building virtual financial institutions within a metaverse, and establishing intelligent customer service systems within these institutions, personalized services such as loan recommendations and investment advice can be provided to customers using target models trained through federated learning. Simultaneously, the intelligent customer service system can predict transaction risks based on the target model, providing real-time monitoring and warnings for potentially risky transactions. The intelligent customer service system also designs and develops intelligent customer service robots, offering multiple communication methods such as text, voice, and video. Personalized services, such as loan recommendations and investment advice, are provided to customers based on target models trained through federated learning. The intelligent customer service system monitors and issues real-time warnings for potentially risky transactions based on model predictions.
[0043] The method for providing financial services based on the metaverse provided in this application involves acquiring the financial services to be provided by financial institutions, determining a target model based on these services, wherein the target model includes at least one of the following types: a risk control model and a product recommendation model. The target model is jointly trained using federated learning based on sample data from multiple branches of the financial institution. The method acquires the first identity information, first transaction behavior information, second identity information, and second transaction behavior information of the financial institution's customers. The first identity information and first transaction behavior information are information stored in the financial institution's database, while the second identity information and second transaction behavior information are information stored in a virtual financial institution within the metaverse for virtual characters. The virtual character is a role created by the customer in the metaverse, and the virtual financial institution is a branch created by the financial institution within the metaverse. The first identity information, first transaction behavior information, second identity information, and second transaction behavior information are input into the target model to obtain the model output. Based on the model output, a preset financial service is provided to the customer, solving the problem of low accuracy in financial services provided in the metaverse in related technologies. By building a federated learning platform between virtual financial institutions in the metaverse and various branches in the real world, and training the target model for providing financial services using federated learning, the accuracy of financial services provided in the metaverse is improved.
[0044] The target model is trained using a federated averaging algorithm and the model parameters of the local models of each branch. Optionally, in the metaverse-based financial service provision method provided in this application embodiment, the target model is obtained in the following manner: multiple branches that need to provide financial services are identified from all branches of the financial institution; each branch is trained based on its local dataset to obtain a local model of the branch, and the model parameters of the local model are determined to obtain a set of model parameters, wherein each branch's local model corresponds to one model parameter; the set of model parameters is processed using a federated averaging algorithm to obtain the processed target model parameters; and the target model is determined based on a preset model framework and the target model parameters.
[0045] Specifically, some branches of financial institutions are unable to provide local datasets for training local models without permission. Therefore, multiple branches that can provide local datasets for training local models and provide financial services are identified from all branches. Each branch trains a local model based on its locally stored customer information dataset. The local model can be a neural network model, and the loss function of the local model is used as the model parameters. A set of model parameters from multiple branches is processed using a federated averaging algorithm and used as the target model parameters. The target model is then constructed based on the target model parameters and a pre-defined model framework. By combining the metaverse with the field of federated learning, the constructed target model provides financial institutions with a new financial service model, enhancing customer experience, protecting customer data privacy through federated learning, reducing the risk of data leakage, and improving the security performance of financial institutions.
[0046] Optionally, in the method for providing financial services based on the metaverse provided in this application embodiment, the target model is a risk control model. Training is performed on each branch based on its local dataset to obtain a local model for the branch. Determining the model parameters of the local model includes: determining the local dataset of the branch and extracting multiple sets of training samples from the local dataset, wherein each set of training samples includes the customer's first identity information, second identity information, first historical transaction behavior information, second historical transaction behavior information, and historical risk assessment value from the local dataset; iteratively training the neural network model using multiple sets of training samples to obtain a set of undetermined models, wherein each iteration trains the neural network model by randomly extracting multiple sets of training samples from the local dataset; determining the minimum loss function from the loss functions of the set of undetermined models; processing the minimum loss function based on preset parameters and regularization coefficients to obtain the processed target loss function, and determining the target loss function as the model parameters of the local model.
[0047] Specifically, when the target model is a risk control model, training samples for training the local model are extracted from the local dataset of the branch offices. For example, the primary identity information of customers and their first historical transaction behavior information at the branch offices stored in the local dataset, as well as the secondary identity information of customers and their second historical transaction behavior information at the virtual financial institution stored in the virtual financial institution, and the customers' historical risk assessment values are used as training samples. In each iteration, a batch of training samples is randomly selected from the local dataset to train the neural network model, resulting in a candidate model. Multiple iterations yield a set of candidate models. From this set of candidate models, the candidate model corresponding to the minimum loss function is determined. The minimum loss function is the loss function with the smallest possible value among the loss functions of each candidate model. The target loss function is determined by the following formula:
[0048]
[0049] Where t represents the iteration number, and for each iteration t = 1, 2, ..., k represents the k-th branch, D k For the local dataset of the kth branch, w k t Let L(w; x) represent the target loss function of the k-th branch after the t-th iteration. i y i The loss function is represented by argmin, where λ represents the regularization coefficient. ‖w‖ represents the initialization of global model parameters. The parameters of the target model in federated learning are provided by determining the parameters of the local model.
[0050] Optionally, in the metaverse-based financial service provision method provided in this application embodiment, the target model is a product recommendation model. Training is performed on each branch based on its local dataset to obtain a local model for that branch. Determining the model parameters of the local model includes: determining the local dataset of the branch and extracting multiple sets of training samples from the local dataset, wherein each set of training samples includes the customer's first identity information, second identity information, first historical transaction behavior information, second historical transaction behavior information, and historical recommended products from the local dataset; iteratively training the neural network model using multiple sets of training samples to obtain a set of undetermined models, wherein each iteration trains the neural network model by randomly extracting multiple sets of training samples from the local dataset; determining the minimum loss function from the loss functions of the set of undetermined models; processing the minimum loss function based on preset parameters and regularization coefficients to obtain the processed target loss function, and determining the target loss function as the model parameters of the local model.
[0051] Specifically, when the target model is a product recommendation model, training samples for training the local model are extracted from the local dataset of the branch offices. For example, the primary identity information of customers and their first historical transaction behavior information at the branch offices stored in the local dataset, as well as the secondary identity information of customers and their second historical transaction behavior information at the virtual financial institution stored in the virtual financial institution, and the historical recommended products to customers are used as training samples. In each iteration of training, a batch of training samples is randomly selected from the local dataset to train the neural network model, resulting in a candidate model. Multiple iterations yield a set of candidate models. From this set of candidate models, the candidate model corresponding to the minimum loss function is determined. The minimum loss function is the loss function with the smallest possible value among the loss functions of each candidate model.
[0052] The loss function of the target model is determined by the model parameters of the local models and the federated averaging algorithm. Optionally, in the metaverse-based financial service provision method provided in this application embodiment, the federated averaging algorithm is used to process a set of model parameters to obtain the processed target model parameters, including: calculating the total data volume of all local datasets of multiple branches, and calculating the ratio of the data volume of the local dataset of each branch to the total data volume to obtain the target weight of the branch; weighted summing of the target loss function and target weight of the local models of multiple branches to obtain the loss function of the target model; and determining the loss function of the target model as the target model parameters.
[0053] Specifically, assume there are K branch offices, each branch office k having its own local dataset D. k n k Let represent the data volume of the k-th branch, and let n represent the total data volume n, where n is the sum of the data volumes of all branches. The goal of federated learning is to train a global model w, or the target model, without sharing the original data. The parameters of the target model are determined by the following formula:
[0054]
[0055] Among them, w t+1 Indicates the target model parameters. This represents a weighted summation of the target loss function and target weights for the local models of multiple branches.
[0056] It should be noted that after determining the target model, it is also necessary to optimize it. For example, if the target model is a model that recommends products to customers based on product ratings, after determining the rating algorithm for the target model, the rating algorithm of the target model is optimized based on the recommendation system algorithm. Taking matrix factorization (MF) as an example, assume the rating algorithm is a rating matrix P∈R. m×n Where m represents the number of users, n represents the number of products, and R ij Let R represent the rating of user i for product j. The goal of matrix factorization is to decompose the rating matrix R into two low-rank matrices P∈R. m×k and Q∈R n×k This makes R≈PQ T The optimization objective of matrix factorization can be expressed as:
[0057]
[0058] Here, Ω represents the set of indices of known ratings in the rating matrix R, and λ is the regularization coefficient. The rating algorithm of the target model is optimized using methods such as gradient descent.
[0059] Optionally, in the metaverse-based financial service provision method provided in this application embodiment, obtaining the first identity information and first transaction behavior information of a financial institution's customers includes: obtaining all types of customer identity information and transaction behavior information from a database, wherein the types of identity information include at least one of the following: age, position, assets, and interests, and the types of transaction behavior information include at least one of the following: transaction frequency, transaction amount, transaction cycle, and transaction method; calculating the first information gain for each type of identity information, and calculating the second information gain for each type of transaction behavior information; determining the identity information whose first information gain is greater than or equal to the information gain threshold as the first identity information, and determining the transaction behavior information whose second information gain is greater than or equal to the information gain threshold as the first transaction information.
[0060] Specifically, the data features of the training target model are determined by calculating information gain. These data features are, in this embodiment, the first identity information and the first transaction behavior information. When determining the data features of the training target model, the influence of each type of data feature on the output of the target model is used to determine the data features, and information gain is used to determine which data features are most important to the target model. For example, the first identity information and the first transaction behavior information are used as data features of the training target model. The types of identity information include age, position, assets, and interests, while the types of transaction behavior information include transaction frequency, transaction amount, transaction cycle, and transaction method. By calculating the information gain of each type of identity information and transaction behavior information on the transaction between the customer and the financial institution, the influence of each type of identity information and transaction behavior information on the output of the target model is determined. If the target model is a product recommendation model, the information gain of age is greater than or equal to the information gain threshold, the information gain of transaction frequency is greater than or equal to the information gain threshold, the information gain of interest is less than the information gain threshold, and the information gain of transaction amount is less than the information gain threshold. This indicates that the customer's age and transaction frequency have an impact on the product recommendation results, while interests and transaction amount have no impact. Therefore, age is used as the first identity information, and transaction frequency is used as the first transaction behavior information. By using information gain to filter the first identity information and the first transaction behavior information, the data features of the training target model are strongly correlated, thereby improving the accuracy of the target model.
[0061] It should be noted that after selecting the primary identity information and primary transaction behavior information as data features, anomaly detection algorithms can be used to remove anomalous samples from the training dataset corresponding to these features. Anomaly detection algorithms such as Isolation Forest (IForest) are tree-based anomaly detection algorithms that construct a forest of multiple isolation trees. Whether a data point is an anomaly is determined by evaluating the average path length of each isolation tree. For an isolation tree, the path length h(x) represents the number of edges from the root node to a leaf node. The anomaly score s(x) of an isolation forest is defined as follows:
[0062]
[0063] Where E(h(x)) is the average path length of data point x in all isolated trees, c(n) is the average path length from the root node to the leaf node in a balanced binary tree, and n represents the number of training samples. The anomaly score s(x) ranges from [0, 1], and the closer the value is to 1, the more likely the data point is to be an anomaly.
[0064] Information gain is obtained by calculating the difference in information entropy. Optionally, in the metaverse-based financial service provision method provided in this application embodiment, calculating the first information gain for each type of identity information and the second information gain for each type of transaction behavior information includes: determining the first information entropy of all customers in the database transacting with financial institutions within a preset period; determining the second information entropy of customers in the database with target type identity information transacting with financial institutions within a preset period; determining the third information entropy of customers in the database with target type transaction behavior information transacting with financial institutions within a preset period; calculating the difference between the first and second information entropies to obtain the first information gain of the target type identity information; and calculating the difference between the first and third information entropies to obtain the second information gain of the target type transaction behavior information.
[0065] Specifically, suppose we have a feature set F = f1, f2, ..., f m And a classification task objective variable Y. Information gain represents the information gain given feature f. i Under the given conditions, the decrease in information entropy of the target variable Y. The formula for calculating information gain is as follows:
[0066] IG(Y, f) i )=H(Y)-H(Y|f i )
[0067] Where H(Y) is the information entropy of the target variable Y, also known as the first information entropy, H(Y|f i ) indicates that given feature fi Under the given conditions, the conditional entropy of the target variable Y, also known as the second or third information entropy, is IG(Y, f). i ) represents information gain, also known as first information gain or second information gain. First identity information is filtered from multiple types of identity information, and second transaction behavior information is filtered from multiple types of transaction behavior information, by calculating the first information gain for each type of identity information and the second information gain for each type of transaction behavior information.
[0068] According to another embodiment of this application, an optional metaverse-based financial services provision method is also provided. Figure 2 This is a flowchart of an optional metaverse-based financial service provision method provided according to embodiments of this application. For example... Figure 2 As shown, the method includes:
[0069] Step S201: Design and build a virtual bank branch in the metaverse, and design an interactive interface for the virtual bank branch.
[0070] Specifically, the virtual bank branch has financial service modules with functions such as account opening, fund transfer, loan, and wealth management. Customers can log in to the virtual bank branch in the metaverse through virtual reality devices to experience immersive financial services.
[0071] Step S202: Establish a cross-bank federated learning platform, whereby branches of various banks share parameters of risk control and anti-fraud models while protecting customer data privacy.
[0072] Specifically, each bank branch selects a suitable machine learning model based on its own business needs and trains it through federated learning. Simultaneously, virtual identity information from the metaverse is used to provide the model with richer data features, improving prediction accuracy.
[0073] Step S203: Regularly update the risk control and anti-fraud models to address the ever-changing financial environment.
[0074] Step S204: Set up an intelligent customer service robot in the metaverse to provide multiple communication methods such as text, voice, and video.
[0075] Specifically, intelligent customer service robots utilize federated learning models to provide personalized services to customers, such as loan recommendations and investment advice. Simultaneously, intelligent customer service can also monitor and issue real-time warnings for potentially risky transactions based on model predictions.
[0076] Step S205: Use blockchain technology to ensure the security and transparency of financial transactions in the metaverse and enhance customer trust.
[0077] The optional metaverse-based financial service delivery method provided in this application combines the metaverse with the field of federated learning, offering a novel financial service model for the banking industry and enhancing customer experience. Federated learning protects customer data privacy, reduces the risk of data breaches, and improves the security performance of the banking industry. It fully utilizes virtual identity information within the metaverse to provide richer data features for risk control and anti-fraud models, improving the predictive accuracy of these models. The intelligent customer service system, combined with models trained through federated learning, provides personalized financial services and real-time risk monitoring, enhancing service quality and security. Blockchain technology ensures the security and transparency of financial transactions within the metaverse, enhancing customer trust.
[0078] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.
[0079] This application also provides a metaverse-based financial service provisioning device. It should be noted that this metaverse-based financial service provisioning device can be used to execute the metaverse-based financial service provisioning method provided in this application. The following describes the metaverse-based financial service provisioning device provided in this application.
[0080] Figure 3 This is a schematic diagram of a metaverse-based financial service provision device provided according to an embodiment of this application. Figure 3 As shown, the device includes:
[0081] The first acquisition unit 10 is used to acquire the financial services to be provided by financial institutions and determine the target model based on the financial services to be provided. The target model is of at least one of the following types: risk control model and product recommendation model. The target model is jointly trained by federated learning based on sample data from multiple branches of the financial institution.
[0082] The second acquisition unit 20 is used to acquire the first identity information, first transaction behavior information, second identity information, and second transaction behavior information of a customer of a financial institution. The first identity information and first transaction behavior information are information stored in the database of the financial institution, while the second identity information and second transaction behavior information are information stored in the virtual financial institution by a virtual character in the metaverse. The virtual character is a character created by the customer in the metaverse, and the virtual financial institution is a branch created by the financial institution in the metaverse.
[0083] Input unit 30 is used to input the first identity information, the first transaction behavior information, the second identity information, and the second transaction behavior information into the target model to obtain the model output results;
[0084] Service unit 40 is used to provide customers with pre-defined financial services based on the model output results.
[0085] The financial service provision device based on the metaverse provided in this application embodiment acquires the financial services to be provided by financial institutions through a first acquisition unit 10, and determines a target model based on the financial services to be provided. The target model includes at least one of the following types: a risk control model and a product recommendation model. The target model is jointly trained using federated learning based on sample data from multiple branches of the financial institution. The second acquisition unit 20 acquires the first identity information, first transaction behavior information, second identity information, and second transaction behavior information of the financial institution's customers. The first identity information and first transaction behavior information are information stored in the financial institution's database, while the second identity information and second transaction behavior information are virtual characters in the metaverse. The information stored in the virtual financial institution is as follows: the virtual role is the role created by the customer in the metaverse, and the virtual financial institution is the branch created by the financial institution in the metaverse; the input unit 30 inputs the first identity information, the first transaction behavior information, the second identity information, and the second transaction behavior information into the target model to obtain the model output result; the service unit 40 provides the customer with preset financial services based on the model output result, which solves the problem of low accuracy of financial services provided in the metaverse in related technologies. By building a federated learning platform between the virtual financial institution in the metaverse and various branches in the real world, and training the target model for providing financial services based on the federated learning method, the accuracy of financial services provided in the metaverse is improved.
[0086] Optionally, in the metaverse-based financial service provision device provided in this application embodiment, the target model is obtained in the following manner: a first determining module, used to determine multiple branches that need to provide financial services from all branches of a financial institution; a training module, used to train each branch based on the local dataset of the branch to obtain a local model of the branch, and determine the model parameters of the local model to obtain a set of model parameters, wherein each branch's local model corresponds to one model parameter; a processing module, used to process the set of model parameters through a federated averaging algorithm to obtain the processed target model parameters; and a second determining module, used to determine the target model based on a preset model framework and the target model parameters.
[0087] Optionally, in the metaverse-based financial service provision device provided in this application embodiment, the target model is a risk control model, and the training module includes: a first determining submodule, used to determine the local dataset of the branch and extract multiple sets of training samples from the local dataset, wherein each set of training samples includes the customer's first identity information, second identity information, first historical transaction behavior information, second historical transaction behavior information, and historical risk assessment value from the local dataset; a first training submodule, used to iteratively train the neural network model through multiple sets of training samples to obtain a set of undetermined models, wherein each iteration of training trains the neural network model by randomly extracting multiple sets of training samples from the local dataset; a second determining submodule, used to determine the minimum loss function from the loss functions of a set of undetermined models; and a first processing submodule, used to process the minimum loss function based on preset parameters and regularization coefficients to obtain the processed target loss function, and determine the target loss function as the model parameters of the local model.
[0088] Optionally, in the metaverse-based financial service provision device provided in this application embodiment, the target model is a product recommendation model, and the training module includes: a third determining submodule, used to determine the local dataset of the branch and extract multiple sets of training samples from the local dataset, wherein each set of training samples includes the customer's first identity information, second identity information, first historical transaction behavior information, second historical transaction behavior information, and historical recommended products in the local dataset; a second training submodule, used to iteratively train the neural network model through multiple sets of training samples to obtain a set of undetermined models, wherein each iteration of training trains the neural network model by randomly extracting multiple sets of training samples from the local dataset; a fourth determining submodule, used to determine the minimum loss function from the loss functions of a set of undetermined models; and a second processing submodule, used to process the minimum loss function based on preset parameters and regularization coefficients to obtain the processed target loss function, and determine the target loss function as the model parameters of the local model.
[0089] Optionally, in the metaverse-based financial service provisioning device provided in this application embodiment, the processing module includes: a first calculation submodule, used to calculate the total data volume of all local datasets of multiple branches, and calculate the ratio of the data volume of the local dataset of each branch to the total data volume, to obtain the target weight of the branch; a summation submodule, used to perform weighted summation of the target loss function and target weight of the local models of multiple branches, to obtain the loss function of the target model; and a fifth determination submodule, used to determine the loss function of the target model as the target model parameters.
[0090] Optionally, in the metaverse-based financial service provision device provided in this application embodiment, the second acquisition unit 20 includes: an acquisition module, used to acquire all types of customer identity information and transaction behavior information from a database, wherein the types of identity information include at least one of the following: age, position, assets, and interests, and the types of transaction behavior information include at least one of the following: transaction frequency, transaction amount, transaction cycle, and transaction method; a calculation module, used to calculate a first information gain for each type of identity information and a second information gain for each type of transaction behavior information; and a third determination module, used to determine identity information with a first information gain greater than or equal to an information gain threshold as first identity information, and to determine transaction behavior information with a second information gain greater than or equal to an information gain threshold as first transaction behavior information.
[0091] Optionally, in the metaverse-based financial service provision device provided in this application embodiment, the calculation module includes: a sixth determining submodule, used to determine the first information entropy of all customers in the database transacting with financial institutions within a preset period; a seventh determining submodule, used to determine the second information entropy of customers in the database with target type identity information transacting with financial institutions within a preset period; an eighth determining submodule, used to determine the third information entropy of customers in the database with target type transaction behavior information transacting with financial institutions within a preset period; a second calculation submodule, used to calculate the difference between the first information entropy and the second information entropy to obtain the first information gain of the target type identity information; and a third calculation submodule, used to calculate the difference between the first information entropy and the third information entropy to obtain the second information gain of the target type transaction behavior information.
[0092] The metaverse-based financial service provision device includes a processor and a memory. The first acquisition unit 10, the second acquisition unit 20, the input unit 30, and the service unit 40 mentioned above are all stored in the memory as program units. The processor executes the program units stored in the memory to realize the corresponding functions.
[0093] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and adjusting kernel parameters can improve the accuracy of financial services provided in the metaverse.
[0094] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.
[0095] This invention provides a computer-readable storage medium storing a program that, when executed by a processor, implements a method for providing financial services based on the metaverse.
[0096] This invention provides a processor for running a program, wherein the program executes a method for providing financial services based on the metaverse.
[0097] Figure 4 This is a schematic diagram of an electronic device provided according to an embodiment of this application. For example... Figure 4 As shown, electronic device 401 includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs the following steps: obtaining financial services to be provided by a financial institution; determining a target model based on the financial services to be provided, wherein the target model type includes at least one of the following: a risk control model and a product recommendation model; the target model is jointly trained using federated learning based on sample data from multiple branches of the financial institution; obtaining the financial institution's customer's first identity information, first transaction behavior information, second identity information, and second transaction behavior information, wherein the first identity information and first transaction behavior information are information stored in the financial institution's database, and the second identity information and second transaction behavior information are information stored in a virtual financial institution for a virtual character in a metaverse; the virtual character is a character created by the customer in the metaverse, and the virtual financial institution is a branch created by the financial institution in the metaverse; inputting the first identity information, first transaction behavior information, second identity information, and second transaction behavior information into the target model to obtain the model output; and providing the customer with a preset financial service based on the model output. The device in this paper can be a server, PC, PAD, mobile phone, etc.
[0098] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program with the following method steps: obtaining financial services to be provided by a financial institution; determining a target model based on the financial services to be provided, wherein the type of the target model includes at least one of the following: a risk control model and a product recommendation model, and the target model is jointly trained by federated learning based on sample data from multiple branches of the financial institution; obtaining the first identity information, first transaction behavior information, second identity information, and second transaction behavior information of the financial institution's customers, wherein the first identity information and first transaction behavior information are information stored in the database of the financial institution, and the second identity information and second transaction behavior information are information stored in a virtual financial institution by a virtual character in a metaverse, where the virtual character is a character created by the customer in the metaverse, and the virtual financial institution is a branch created by the financial institution in the metaverse; inputting the first identity information, first transaction behavior information, second identity information, and second transaction behavior information into the target model to obtain the model output result; and providing the customer with a preset financial service based on the model output result.
[0099] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0100] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0101] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0102] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0103] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0104] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0105] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0106] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.
[0107] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0108] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.
Claims
1. A method for providing a financial service based on a meta universe, the method comprising: receiving a request for a financial service from a user; and providing the financial service based on the request. include: Obtain the financial services to be provided by financial institutions, and determine a target model based on the financial services to be provided. The type of the target model includes at least one of the following: risk control model and product recommendation model. The target model is jointly trained by federated learning based on sample data from multiple branches of the financial institution. The system obtains the first identity information, first transaction behavior information, second identity information, and second transaction behavior information of the financial institution's customer. The first identity information and the first transaction behavior information are information stored in the database of the financial institution, and the second identity information and the second transaction behavior information are information stored in the virtual financial institution by a virtual character in the metaverse. The virtual character is a character created by the customer in the metaverse, and the virtual financial institution is a branch created by the financial institution in the metaverse. Input the first identity information, the first transaction behavior information, the second identity information, and the second transaction behavior information into the target model to obtain the model output result; Based on the output of the model, the customer is provided with a pre-set financial service.
2. The method of claim 1, wherein, The target model is obtained in the following way: Identify the plurality of branches of the financial institution that require the provision of the financial services; The local dataset of the branch is used to train each branch to obtain the local model of the branch, and the model parameters of the local model are determined to obtain a set of model parameters, wherein each branch's local model corresponds to one model parameter. The set of model parameters is processed using a federated averaging algorithm to obtain the processed target model parameters; The target model is determined based on the preset model framework and the target model parameters.
3. The method according to claim 2, characterized in that, The target model is a risk control model. It is trained on each branch based on its local dataset to obtain a local model for that branch. The model parameters of the local model are determined as follows: Determine the local dataset of the branch office and extract multiple sets of training samples from the local dataset, wherein each set of training samples includes the customer's first identity information, second identity information, first historical transaction behavior information, second historical transaction behavior information and historical risk assessment value from the local dataset; The neural network model is iteratively trained using the multiple sets of training samples to obtain a set of undetermined models. Each iteration of training involves randomly extracting multiple sets of training samples from the local dataset to train the neural network model. Determine the minimum loss function from the loss functions of the set of undetermined models; The minimum loss function is processed based on preset parameters and regularization coefficients to obtain the processed target loss function, which is then determined as the model parameter of the local model.
4. The method of claim 2, wherein, The target model is a product recommendation model. It is trained on each branch office using its local dataset to obtain a local model for that branch office. The model parameters of the local model are determined as follows: Determine the local dataset of the branch office and extract multiple sets of training samples from the local dataset, wherein each set of training samples includes the customer's first identity information, second identity information, first historical transaction behavior information, second historical transaction behavior information, and historical recommended products in the local dataset; The neural network model is iteratively trained using the multiple sets of training samples to obtain a set of undetermined models. Each iteration of training involves randomly extracting multiple sets of training samples from the local dataset to train the neural network model. Determine the minimum loss function from the loss functions of the set of undetermined models; The minimum loss function is processed based on preset parameters and regularization coefficients to obtain the processed target loss function, which is then determined as the model parameter of the local model.
5. The method of claim 2, wherein, The target model parameters are obtained by processing the set of model parameters using a federated averaging algorithm, including: Calculate the total data volume of all local datasets of the multiple branches, and calculate the ratio of the data volume of the local dataset of each branch to the total data volume to obtain the target weight of the branch. The target loss function of the target model is obtained by weighted summing of the target loss function and target weights of the local models of the multiple branches. The loss function of the target model is determined as the parameters of the target model.
6. The method of claim 1, wherein, Obtaining the first identity information and first transaction behavior information of the financial institution's customers includes: Retrieve all types of identity information and transaction behavior information of the customer from the database, wherein the types of identity information include at least one of the following: age, position, assets and interests, and the types of transaction behavior information include at least one of the following: transaction frequency, transaction amount, transaction cycle and transaction method; Calculate the first information gain for each type of identity information, and calculate the second information gain for each type of transaction behavior information; The identity information whose first information gain is greater than or equal to the information gain threshold is determined as the first identity information, and the transaction behavior information whose second information gain is greater than or equal to the information gain threshold is determined as the first transaction behavior information.
7. The method of claim 6, wherein, Calculate the first information gain for each type of identity information, and calculate the second information gain for each type of transaction behavior information, including: Determine the first information entropy of all transactions between the customers in the database and the financial institution within a preset period; The second information entropy is determined by identifying customers with target type identity information in the database who transact with the financial institution within the preset period; The third information entropy is determined by identifying customers with target-type transaction behavior information in the database who transact with the financial institution within the preset period; Calculate the difference between the first information entropy and the second information entropy to obtain the first information gain of the identity information of the target type; The difference between the first information entropy and the third information entropy is calculated to obtain the second information gain of the transaction behavior information of the target type. 8.A meta-universe based financial service providing apparatus, characterized by comprising: include: The first acquisition unit is used to acquire the financial services to be provided by the financial institution, and determine the target model based on the financial services to be provided. The target model is of at least one of the following types: risk control model and product recommendation model. The target model is jointly trained by federated learning based on sample data from multiple branches of the financial institution. The second acquisition unit is used to acquire the first identity information, first transaction behavior information, second identity information, and second transaction behavior information of the customer of the financial institution. The first identity information and the first transaction behavior information are information stored in the database of the financial institution, and the second identity information and the second transaction behavior information are information stored in the virtual financial institution by a virtual character in the metaverse. The virtual character is a character created by the customer in the metaverse, and the virtual financial institution is a branch created by the financial institution in the metaverse. The input unit is used to input the first identity information, the first transaction behavior information, the second identity information, and the second transaction behavior information into the target model to obtain the model output result; A service unit is used to provide preset financial services to the customer based on the output results of the model.
9. A non-volatile storage medium, comprising: The non-volatile storage medium includes a stored program, wherein the program, when running, controls the device where the non-volatile storage medium is located to execute the metaverse-based financial service provision method according to any one of claims 1 to 7.
10. An electronic device, comprising: It includes one or more processors and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the metaverse-based financial services delivery method according to any one of claims 1 to 7.
Citation Information
Patent Citations
Credit business recommendation method based on federal learning
CN114782176A
Metacosmic application scene system for popularizing recreation and sports industry
CN115554692A
