Systems and methods for entity labeling based on behavior
Methods and systems use a second dataset comprising independently validated data based on labeled blockchain operations previously processed through the blockchain network. The use of the second dataset in conjunction with the dataset comprising labeled blockchain characteristics data received from a plurality of sources provides a comprehensive input for a machine learning model to identify and label suspicious blockchain operations. By doing so, the system mitigates fraudulent, criminal, or suspicious transactions. For example, the system may send an alert to a user if a behavior type is deemed a high risk by the machine learning model; the system may furthermore hold the funds associated with the blockchain operation in a separate account while the transaction is manually reviewed. Thus, the system may generate alerts to notify a user of a suspicious behavior type or suspicious blockchain operation, divert funds associated with the transaction, and conduct a manual review of suspicious blockchain operations.
1 . A system for generating real-time security alerts for blockchain operations, the system comprising:
a first cryptography-based, storage application,
wherein the first cryptography-based, storage application corresponds to a first private key, and
wherein the first private key is stored on a first user device;
a second cryptography-based, storage application,
wherein the second cryptography-based, storage application corresponds to a first partial private key and a second partial private key,
wherein the first partial private key is stored on a remote device network, and
wherein the second partial private key is stored on the first user device; and
one or more computer processors configured to cause operations comprising:
identifying a first blockchain operation within first blockchain data representing a first state of a blockchain network;
generating a first feature input based on a receipt address for the first blockchain operation;
processing the first feature input using an artificial intelligence model that is trained based on a first training dataset and a second training dataset comprising independently validated data that is assigned a weight to reconcile differences between predictions of the artificial intelligence model and reference feedback information,
wherein the first training dataset comprises labeled blockchain characteristics data received from a plurality of sources, and
wherein the independently validated data is based on labeled blockchain operations, previously processed through the blockchain network, directed to an entity to confirm a labeling of the entity;
generating, based on processing the first feature input using the artificial intelligence model, a first output that indicates a security risk level, for the receipt address, corresponding to suspicious blockchain activity;
modifying, based on the first output and prior to executing the first blockchain operation, the receipt address for the first blockchain operation to obtain a modified receipt address; and
performing, without user input, the first blockchain operation with the modified receipt address to mitigate the suspicious blockchain activity from being validated on a blockchain.
2 . A method for generating real-time security alerts for blockchain operations, the method comprising:
receiving first blockchain data representing a first state of a blockchain network;
identifying a first blockchain operation within the first blockchain data;
generating a first feature input based on a first blockchain characteristic of the first blockchain operation, wherein the first blockchain characteristic is based on a receipt address for the first blockchain operation;
processing the first feature input using an artificial intelligence model,
wherein the artificial intelligence model is trained based on a first training dataset and a second training dataset,
wherein the first training dataset comprises labeled blockchain characteristics data received from a plurality of sources,
wherein the second training dataset comprises independently validated data that is based on labeled blockchain operations, previously processed through the blockchain network, directed to an entity to confirm a labeling of the entity, and
wherein the independently validated data is assigned a weight to reconcile differences between predictions of the artificial intelligence model and reference feedback information;
generating, based on processing the first feature input using the artificial intelligence model, a first output that indicates a security risk level, for the receipt address, corresponding to suspicious blockchain activity; and
modifying, based on the first output and prior to executing the first blockchain operation, the receipt address for the first blockchain operation to obtain a modified receipt address;
performing, without user input, the first blockchain operation with the modified receipt address to mitigate the suspicious blockchain activity from being validated on a blockchain.
3 . The method of claim 2 , further comprising:
determining, based on the first output, that the first blockchain operation has the security risk level of a plurality of security risk levels.
4 . The method of claim 2 , further comprising:
generating a first alert indicating that the first blockchain operation has the security risk level; and
requesting a user confirmation of the first alert.
5 . The method of claim 2 , further comprising:
generating a first alert indicating that the first blockchain operation has the security risk level.
6 . The method of claim 2 , further comprising:
requesting a user confirmation of a multifactor token; and
in response to receiving the user confirmation, authorizing the first blockchain operation to proceed.
7 . The method of claim 2 , further comprising:
receiving second blockchain data representing a second state of the blockchain network;
identifying a second blockchain operation within the second blockchain data;
generating a second feature input based on a second blockchain characteristic of the second blockchain operation;
processing the second feature input, using the artificial intelligence model, to generate a second output; and
determining, based on the second output, to automatically proceed with the first blockchain operation.
8 . The method of claim 2 , wherein the labeled blockchain characteristics data comprises a signature type used for a previous blockchain operation.
9 . The method of claim 2 , wherein the labeled blockchain characteristics data comprises a smart contract type used for a previous blockchain operation.
10 . The method of claim 2 , wherein the labeled blockchain characteristics data comprises a behavior type used for a previous blockchain operation.
11 . The method of claim 2 , wherein the labeled blockchain characteristics data comprises an entity label corresponding to a known receipt address.
12 . The method of claim 2 , wherein generating the first feature input based on the first blockchain characteristic of the first blockchain operation comprises:
determining the receipt address for the first blockchain operation; and
generating the first feature input based on the receipt address.
13 . The method of claim 2 , wherein processing the first feature input, using the artificial intelligence model, comprises determining a likelihood of a security risk for the first blockchain operation based on a pattern of activity corresponding to the receipt address.
14 . One or more non-transitory, computer readable media having instructions recorded thereon, that, when executed by one or more processors, causes operations comprising:
receiving first blockchain data representing a first state of a blockchain network;
identifying a first blockchain operation within the first blockchain data;
generating a first feature input based on a first blockchain characteristic of the first blockchain operation, wherein the first blockchain characteristic is based on a receipt address for the first blockchain operation;
processing the first feature input using an artificial intelligence model
wherein the artificial intelligence model is trained based on a first training dataset and a second training dataset,
wherein the first training dataset comprises labeled blockchain characteristics data received from a plurality of sources, and
wherein the second training dataset comprises independently validated data;
generating, based on processing the first feature input using the artificial intelligence model, a first output that indicates a security risk level, for the receipt address, corresponding to suspicious blockchain activity;
modifying, based on the first output and prior to executing the first blockchain operation, the receipt address for the first blockchain operation to obtain a modified receipt address;
performing, without user input, the first blockchain operation with the modified receipt address to mitigate the suspicious blockchain activity from being validated on a blockchain.
15 . The one or more non-transitory, computer readable media of claim 14 , wherein the operations further comprise:
determining, that the first blockchain operation has the security risk level of a plurality of security risk levels.
16 . The one or more non-transitory, computer readable media of claim 14 , wherein the operations further comprise:
providing a first alert indicating that the first blockchain operation has the security risk level of a plurality of security risk levels.
17 . The one or more non-transitory, computer readable media of claim 14 , wherein the operations further comprise:
generating a first alert indicating that the first blockchain operation has the first security risk level of a plurality of security risk levels.
18 . The one or more non-transitory, computer readable media of claim 14 , wherein the operations further comprise:
requesting a user confirmation of a multifactor token.
19 . The one or more non-transitory, computer readable media of claim 14 , wherein the operations further comprise:
receiving second blockchain data representing a second state of the blockchain network;
identifying a second blockchain operation within the second blockchain data;
generating a second feature input based on a second blockchain characteristic of the second blockchain operation;
processing the second feature input, using the artificial intelligence model, to generate a second output; and
determining, based on the second output, whether to automatically proceed with the first blockchain operation.
20 . The method of claim 2 , further comprising:
using a digital signature to execute the first blockchain operation.