GAT-based virtual asset wallet address blacklist database generation system and method

Abstract

Disclosed are a GAT-based virtual asset wallet address blacklist database generation system and method. The present invention comprises: a GAT AI engine server which trains a graph attention network-based artificial intelligence model by using respective full node indices of virtual assets, prestored in an index database server, transaction common item information prestored in a virtual asset transaction analysis database server, and a main blacklist composed of high-risk virtual asset wallet addresses, prestored in a main blacklist server, calculates GAT scores based on the trained artificial intelligence model, estimates high-risk virtual asset wallet addresses by using the calculated GAT scores, and generates a GAT blacklist composed of the estimated high-risk virtual asset wallet addresses; and a GAT blacklist database server which stores the GAT blacklist generated in the GAT AI engine server.

Description

System and method for creating a GAT-based virtual asset wallet address blacklist database

[0001] The present invention relates to a system and method for creating a virtual asset wallet address blacklist database, and more specifically, to a system and method for creating a virtual asset wallet address blacklist database based on a GAT (graph attention network).

[0002] Due to the inherent difficulty in tracking funds, blockchain-based virtual assets are often used for illegal financial transactions. If virtual asset wallets themselves are frequently used for highly risky transactions, it's important to proactively identify and prevent unauthorized use.

[0003] However, because it is very easy to create a virtual asset wallet and the security and authentication procedures are lax, anyone can create a virtual asset wallet at any time and use it for illegal purposes.

[0004] Therefore, it is necessary to proactively identify high-risk or harmful virtual asset wallets to prevent them from being used for illegal transactions or fraudulent use.

[0005] However, current virtual asset exchanges lack the means to prevent this.

[0006] In particular, there are no established standards for determining whether numerous virtual asset wallets pose a high or low risk of illegal activity. Therefore, an algorithm capable of accurately assessing the potential illegality of virtual asset wallets is urgently needed.

[0007] The purpose of the present invention is to provide a system for creating a GAT-based virtual asset wallet address blacklist database.

[0008] Another object of the present invention is to provide a method for creating a GAT-based virtual asset wallet address blacklist database.

[0009] The system for generating a GAT-based virtual asset wallet address blacklist database according to the purpose of the present invention described above may be configured to include a GAT AI engine server for training an artificial intelligence model based on a graph attention network using a main blacklist composed of an index of a full node for each virtual asset pre-stored in an index database server, transaction common item information pre-stored in a virtual asset transaction analysis database server, and high-risk virtual asset wallet addresses pre-stored in a main blacklist server, calculating a GAT score based on the trained artificial intelligence model, estimating a high-risk virtual asset wallet address using the calculated GAT score, and generating a GAT blacklist composed of the estimated high-risk virtual asset wallet addresses; and a GAT blacklist database server for storing the GAT blacklist generated by the GAT AI engine server.

[0010] Here, the GAT AI engine server may be configured to include a first data preprocessing module that preprocesses transaction common item information stored in the virtual asset transaction analysis database server, labels it by coin type, and searches for transactions corresponding to a predetermined virtual asset wallet address; an AI learning module that performs GAT learning using the transactions searched by the first data preprocessing module; a teacher module that pseudo-labels unlabeled transactions in the first data preprocessing module so that the AI ​​learning module can relearn them and feeds them back to the AI ​​learning module; and a first risk calculation module that calculates a risk corresponding to the GAT score of each virtual asset wallet address based on the GAT learning result performed by the AI ​​learning module.

[0011] And it may be configured to further include a high-risk wallet address service server that receives a virtual asset wallet address from a virtual asset exchange server, calculates a risk level for the received virtual asset wallet address, and responds to the virtual asset exchange server.

[0012] And the high-risk wallet address service server may be configured to include a second data preprocessing module that receives a virtual asset wallet address from a virtual asset exchange server and performs preprocessing; a transaction inquiry module that searches the virtual asset transaction analysis database server for transactions of the virtual asset wallet address preprocessed by the second data preprocessing module; a second risk calculation module that calculates a risk corresponding to the GAT score of the virtual asset wallet address using the transaction searched by the transaction inquiry module; and a learning update module that requests the AI ​​learning module to learn an artificial intelligence model based on the risk calculated by the second risk calculation module.

[0013] And it may be configured to further include a high-risk wallet address management server that receives a virtual asset wallet address from an administrator terminal, calculates a risk level for the received virtual asset wallet address, and responds to the virtual asset exchange server.

[0014] A method for generating a GAT-based virtual asset wallet address blacklist database according to another object of the present invention may be configured to include a step in which a GAT AI engine server learns a graph attention network-based artificial intelligence model using a main blacklist consisting of an index of each virtual asset full node pre-stored in an index database server, transaction common item information pre-stored in a virtual asset transaction analysis database server, and high-risk virtual asset wallet addresses pre-stored in a main blacklist server; a step in which the GAT AI engine server calculates a GAT score based on the learned artificial intelligence model; a step in which the GAT AI engine server estimates a high-risk virtual asset wallet address using the calculated GAT score and generates a GAT blacklist consisting of the estimated high-risk virtual asset wallet addresses; and a step in which the GAT blacklist generated by the GAT AI engine server is stored in a GAT blacklist database server.

[0015] Here, the high-risk wallet address service server may be configured to further include a step of receiving a virtual asset wallet address from a virtual asset exchange server, calculating a risk level for the received virtual asset wallet address, and responding to the virtual asset exchange server.

[0016] In addition, the high-risk wallet address management server may be configured to further include a step of receiving a virtual asset wallet address from an administrator terminal, calculating a risk level for the received virtual asset wallet address, and responding to the administrator terminal.

[0017] According to the above-described GAT-based virtual asset wallet address blacklist database creation system and method, full nodes and corresponding transaction data are collected for each virtual asset network, and learning data for detecting full nodes with a high possibility of fraudulent use, etc. in advance from the collected full nodes and transaction data is automatically extracted from the full nodes and transaction data, and a database is created by creating a blacklist and whitelist of each virtual asset wallet address, thereby performing GAT learning based on the database, thereby having the effect of more accurately estimating and monitoring the possibility of illegal use or fraudulent transactions for all virtual asset wallet addresses.

[0018] FIG. 1 is a block diagram of a GAT-based virtual asset wallet address blacklist database creation system according to one embodiment of the present invention.

[0019] FIG. 2 is a detailed configuration diagram of a GAT-based virtual asset wallet address blacklist database creation system according to one embodiment of the present invention.

[0020] FIG. 3 is a flowchart of a method for creating a GAT-based virtual asset wallet address blacklist database according to an embodiment of the present invention.

[0021] Figures 4 and 6 are detailed flowcharts of a method for creating a GAT-based virtual asset wallet address blacklist database according to an embodiment of the present invention.

[0022] Figure 5 is an example of risk calculation data according to one embodiment of the present invention.

[0023]

[0024] The present invention is susceptible to various modifications and embodiments. Specific embodiments are illustrated in the drawings and detailed descriptions of specific embodiments for carrying out the invention are provided. However, this is not intended to limit the present invention to specific embodiments, but rather to encompass all modifications, equivalents, and alternatives falling within the spirit and technical scope of the present invention. Throughout the description of each drawing, similar reference numerals have been used to designate similar components.

[0025] Terms such as first, second, A, and B may be used to describe various components, but these components should not be limited by these terms. These terms are used solely to distinguish one component from another. For example, without departing from the scope of the present invention, a first component may be referred to as a second component, and similarly, a second component may also be referred to as a first component. The term "and / or" includes a combination of multiple related items described herein or any of multiple related items described herein.

[0026] When a component is referred to as being "connected" or "connected" to another component, it should be understood that it may be directly connected or connected to that other component, but that there may be other components intervening. Conversely, when a component is referred to as being "directly connected" or "connected" to another component, it should be understood that there are no other components intervening.

[0027] The terminology used in this application is only used to describe specific embodiments and is not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, it should be understood that the terms "comprise" or "have" indicate the presence of a feature, number, step, operation, component, part, or combination thereof described in the specification, but do not preclude the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0028] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and will not be interpreted in an idealized or overly formal sense unless explicitly defined herein.

[0029] Hereinafter, a preferred embodiment according to the present invention will be described in detail with reference to the attached drawings.

[0030] FIG. 1 is a block diagram of a GAT-based virtual asset wallet address blacklist database creation system according to one embodiment of the present invention, and FIG. 2 is a detailed diagram of a GAT-based virtual asset wallet address blacklist database creation system according to one embodiment of the present invention.

[0031] Referring to FIGS. 1 and 2, a GAT-based virtual asset wallet address blacklist database creation system (300) may be configured to include a GAT AI engine server (310), a GAT blacklist database server (320), a high-risk wallet address service server (330), and a high-risk wallet address management server (340).

[0032] Below, the detailed configuration is explained.

[0033] The GAT AI engine server (310) can be configured to learn an artificial intelligence model based on a graph attention network using a main blacklist consisting of an index of a full node for each virtual asset pre-stored in an index database server (101), transaction common item information pre-stored in a virtual asset transaction analysis database server (102), and high-risk virtual asset wallet addresses pre-stored in a main blacklist server (201).

[0034] Here, the index database server (101) stores an index of each virtual asset-specific full node. The index can be pre-generated through sequential indexing of each virtual asset wallet address. Furthermore, transaction common information can include the sender's wallet address, recipient's wallet address, virtual asset transfer amount, transfer fee, transaction time, transaction hash, previous transaction output, and signature for all transactions.

[0035] And, it can be configured to calculate a GAT score based on an artificial intelligence model learned on the GAT AI engine server (310), estimate a high-risk virtual asset wallet address using the calculated GAT score, and create a GAT blacklist composed of the estimated high-risk virtual asset wallet addresses.

[0036] The GAT AI engine server (310) may be configured to include a first data preprocessing module (311), an AI learning module (312), a teacher module (313), and a first risk calculation module (314).

[0037] Below, the detailed configuration is explained.

[0038] The first data preprocessing module (311) can be configured to preprocess common transaction item information stored in the virtual asset transaction analysis database server (102), label it by coin type, and search for transactions corresponding to a given virtual asset wallet address.

[0039] The AI ​​learning module (312) can be configured to perform GAT learning using transactions retrieved from the first data preprocessing module (311).

[0040] At this time, the AI ​​learning module (312) can generate a GAT blacklist by performing a GAT learning model through the following process.

[0041] First, the AI ​​learning module (312) can extract a GAT graph structure based on a graph attention network for the entire transaction by using the index of each virtual asset full node pre-stored in the index database server (101) and the transaction common item information pre-stored in the virtual asset transaction analysis database server (102). In addition, an adjacency matrix can be generated by corresponding the connection relationships constituting the GAT graph structure to edges.

[0042] And the AI ​​learning module (312) can create a characteristic matrix by assigning preset weights to each online source site when crawling and collecting data matrixed by risk category using the main blacklist consisting of high-risk virtual asset wallet addresses pre-stored in the main blacklist server (201).

[0043] In addition, the AI ​​learning module (312) can generate a pre-training model for training data labeled with high-risk virtual asset wallet addresses by passing the characteristic matrix and tangency matrix constituting the nodes and edges of the GAT graph structure through an attention layer. In addition, the pre-training model can be utilized to predict unlabeled data for high-risk virtual asset wallet addresses, and a main learning model based on semi-supervised learning can be performed that uses the prediction result as a GAT score for estimating the high-risk virtual asset wallet address.

[0044] At this time, the AI ​​learning module (312) can be configured to automatically create and update a GAT blacklist DB in which a GAT score for estimating a high-risk virtual asset wallet address is stored by repeatedly performing the main learning model.

[0045] Meanwhile, the AI ​​learning module (312) can be configured to generate a characteristic matrix with the category classification values ​​of wallet addresses or transactions of high-risk assets by corresponding all transactions to nodes of a graph structure in the case of the BTC network and corresponding all wallet addresses to nodes of a graph structure in the case of the ETH network.

[0046] And the AI ​​learning module (312) can be configured to generate an adjacency matrix that corresponds the connection relationship between transactions to edges in the case of the BTC network, and corresponds the connection relationship between wallet addresses to edges in the case of the ETH network.

[0047] On the other hand, the process of creating a GAT blacklist is as follows:

[0048] First, a random virtual asset wallet address is entered. All transactions adjacent to the random virtual asset wallet address are then retrieved. For each transaction, the risk category classification and the estimated probability of the address being a high-risk virtual asset wallet are retrieved. The risk category classifications and the estimated probability of the high-risk virtual asset wallet address for each transaction are then combined to calculate the overall risk for the random virtual asset wallet address.

[0049] Here, the process of looking up the probability value includes the process of looking up all transactions in the blockchain network related to a random virtual asset wallet address in the GAT blacklist database, the process of predicting unlabeled data for high-risk virtual asset wallet addresses by utilizing transaction data labeled with a GAT score that estimates high-risk virtual asset wallet addresses among all transactions, and the process of using the prediction result as a risk probability value that estimates high-risk virtual asset wallet addresses, and the process of storing the risk category classification and probability value that estimates high-risk virtual asset wallet addresses for each of all transactions.

[0050] The teacher module (313) may be configured to pseudo-label unlabeled transactions in the first data preprocessing module (311) so that the AI ​​learning module can relearn them and feed them back to the AI ​​learning module.

[0051] The first risk calculation module (314) can be configured to calculate a risk corresponding to the GAT score of each virtual asset wallet address based on the GAT learning results performed in the AI ​​learning module (312).

[0052] The GAT blacklist database server (320) can be configured to store the GAT blacklist generated by the GAT AI engine server (310).

[0053] A high-risk wallet address service server (330) may be configured to receive a virtual asset wallet address from a virtual asset exchange server, calculate a risk level for the received virtual asset wallet address, and respond to the virtual asset exchange server.

[0054] The high-risk wallet address service server (330) may be configured to include a second data preprocessing module (331), a transaction inquiry module (332), a second risk calculation module (333), and a learning update module (334).

[0055] Below, the detailed configuration is explained.

[0056] The second data preprocessing module (331) may be configured to receive a virtual asset wallet address from a virtual asset exchange server and perform preprocessing.

[0057] The transaction inquiry module (332) can be configured to inquire about the transaction of a virtual asset wallet address for which preprocessing has been performed in the second data preprocessing module (331) in the virtual asset transaction analysis database server (102).

[0058] The second risk calculation module (333) can be configured to calculate a risk corresponding to the GAT score of the virtual asset wallet address using the transaction searched by the transaction search module (332).

[0059] The learning update module (334) may be configured to request the AI ​​learning module (312) to learn the artificial intelligence model based on the risk calculated by the second risk calculation module (333).

[0060] The high-risk wallet address management server (340) may be configured to receive a virtual asset wallet address from an administrator terminal (400), calculate a risk level for the received virtual asset wallet address, and respond to the administrator terminal (400).

[0061] FIG. 3 is a flowchart of a method for creating a GAT-based virtual asset wallet address blacklist database according to an embodiment of the present invention.

[0062] Referring to FIG. 3, the GAT AI engine server (310) learns an artificial intelligence model based on a graph attention network using a main blacklist consisting of an index of a full node for each virtual asset pre-stored in an index database server (101), transaction common item information pre-stored in a virtual asset transaction analysis database server (102), and high-risk virtual asset wallet addresses pre-stored in a main blacklist server (201) (S10).

[0063] Next, the GAT AI engine server (310) calculates a GAT score based on the learned artificial intelligence model (S20).

[0064] Next, the GAT AI engine server (310) estimates a high-risk virtual asset wallet address using the calculated GAT score and creates a GAT blacklist consisting of the estimated high-risk virtual asset wallet addresses (S30).

[0065] Next, the GAT blacklist generated by the GAT AI engine server (310) is stored in the GAT blacklist database server (320) (S40).

[0066] Next, the high-risk wallet address service server (330) receives a virtual asset wallet address from a virtual asset exchange server, calculates the risk level for the received virtual asset wallet address, and responds to the virtual asset exchange server (S50).

[0067] Next, the high-risk wallet address management server (330) receives a virtual asset wallet address from the administrator terminal (400), calculates the risk level for the received virtual asset wallet address, and responds to the administrator terminal (400) (S60).

[0068] FIG. 4 and FIG. 6 are detailed flowcharts of a method for creating a GAT-based virtual asset wallet address blacklist database according to an embodiment of the present invention, and FIG. 5 is an example of risk calculation data according to an embodiment of the present invention.

[0069] Here, Figure 4 illustrates the detailed process for creating a GAT-based learning model. It is configured to retrieve all transactions for a given wallet address and perform learning based on the GAT algorithm. At this time, data that was not labeled during the data preprocessing process is also configured to be retrained by the teacher module (313) to improve data accuracy. Furthermore, a risk level corresponding to the GAT score can be calculated to generate a GAT-based blacklist. Furthermore, it exemplifies a process of immediately accepting unlabeled data through a temporary teacher and performing retraining.

[0070] Figure 5 shows the results of searching for transactions of a specific BTC wallet address A, calculating the sum of A's transactions by category, and then monitoring the total amount of A's transactions.

[0071] For example, the total transmission volume of the Threat category of A is 12.4, the total transmission volume of the Tumble category is 20.1531, and the total transmission volume of all categories of A is 100.

[0072] Here, the risk of A is calculated by dividing the maximum transmission / reception volume by the total transmission / reception volume of A. In other words, the risk of A's Tumble is 20.1531 / 100, which is 20.1531%.

[0073] Figure 6 illustrates the process of calculating and responding to a request from a virtual asset exchange server with the risk level of a virtual asset wallet address. The risk level calculation results are updated to the GAT blacklist database server (320) to build a database.

[0074] As described above, the present invention collects full nodes and corresponding transaction data for each virtual asset network, automatically extracts learning data from the collected full nodes and transaction data to detect full nodes with a high possibility of fraudulent use in advance, and performs GAT learning based on a database constructed by creating a blacklist and a whitelist of each virtual asset wallet address, thereby having the effect of more accurately estimating and monitoring the possibility of illegal use or fraudulent transactions for all virtual asset wallet addresses.

[0075] Although the present invention has been described with reference to the above embodiments, it will be understood by those skilled in the art that various modifications and changes can be made to the present invention without departing from the spirit and scope of the present invention as set forth in the claims below.

Claims

1. A GAT AI engine server that trains an artificial intelligence model based on a graph attention network using a main blacklist consisting of an index of each virtual asset full node pre-stored in an index database server, common transaction item information pre-stored in a virtual asset transaction analysis database server, and high-risk virtual asset wallet addresses pre-stored in a main blacklist server, calculates a GAT score based on the trained artificial intelligence model, estimates high-risk virtual asset wallet addresses using the calculated GAT score, and creates a GAT blacklist consisting of the estimated high-risk virtual asset wallet addresses; A GAT-based virtual asset wallet address blacklist database generation system including a GAT blacklist database server in which a GAT blacklist generated from the above GAT AI engine server is stored.

2. In the first paragraph, the GAT AI engine server, A first data preprocessing module that preprocesses common transaction item information stored in the virtual asset transaction analysis database server, labels it by coin type, and searches for transactions corresponding to a given virtual asset wallet address; An AI learning module that performs GAT learning using the transactions retrieved from the first data preprocessing module; A teacher module that pseudo-labels unlabeled transactions in the first data preprocessing module and feeds the pseudo-labeled data back to the AI ​​learning module so that the AI ​​learning module can relearn the unlabeled transactions; A GAT-based virtual asset wallet address blacklist database creation system, characterized in that it comprises a first risk calculation module that calculates a risk corresponding to the GAT score of each virtual asset wallet address based on the GAT learning results performed in the above AI learning module.

3. In paragraph 1, A GAT-based virtual asset wallet address blacklist database creation system characterized in that it further includes a high-risk wallet address service server that receives a virtual asset wallet address from a virtual asset exchange server, calculates a risk level for the received virtual asset wallet address, and responds to the virtual asset exchange server.

4. In paragraph 3, the high-risk wallet address service server, A second data preprocessing module that receives a virtual asset wallet address from a virtual asset exchange server and performs preprocessing. A transaction query module that queries the virtual asset transaction analysis database server for transactions of virtual asset wallet addresses for which preprocessing has been performed in the second data preprocessing module. A second risk calculation module that calculates the risk corresponding to the GAT score of the virtual asset wallet address using the transaction searched in the above transaction search module. A GAT-based virtual asset wallet address blacklist database creation system, characterized in that the AI ​​learning module is configured to include a learning update module that requests the AI ​​model to learn based on the risk calculated by the second risk calculation module.

5. In paragraph 4, A GAT-based virtual asset wallet address blacklist database creation system characterized in that it further includes a high-risk wallet address management server that receives a virtual asset wallet address from an administrator terminal, calculates a risk level for the received virtual asset wallet address, and responds to the virtual asset exchange server.

6. A step in which the GAT AI engine server learns an artificial intelligence model based on a graph attention network using a main blacklist consisting of an index of each virtual asset full node pre-stored in an index database server, common transaction item information pre-stored in a virtual asset transaction analysis database server, and high-risk virtual asset wallet addresses pre-stored in a main blacklist server; A step in which the above GAT AI engine server calculates a GAT score based on the learned artificial intelligence model; A step in which the GAT AI engine server estimates a high-risk virtual asset wallet address using the calculated GAT score and creates a GAT blacklist consisting of the estimated high-risk virtual asset wallet addresses; A method for creating a GAT-based virtual asset wallet address blacklist database, characterized in that it comprises a step of storing a GAT blacklist generated from the GAT AI engine server in a GAT blacklist database server.

7. In paragraph 6, A method for creating a GAT-based virtual asset wallet address blacklist database, characterized in that the method further comprises a step of a high-risk wallet address service server receiving a virtual asset wallet address from a virtual asset exchange server, calculating a risk level for the received virtual asset wallet address, and responding to the virtual asset exchange server.

8. In paragraph 7, A method for creating a GAT-based virtual asset wallet address blacklist database, characterized in that the method further comprises a step of a high-risk wallet address management server receiving a virtual asset wallet address from an administrator terminal, calculating a risk level for the received virtual asset wallet address, and responding to the administrator terminal.

Images