Systems, methods, and computer program products for learning continuous embedding spaces for real-time payment transactions

Abstract

Methods, systems, and computer program products for learning a continuous embedding space for real-time payment (RTP) transactions are provided. A method can include receiving RTP data including a plurality of attributes, the plurality of attributes including a sender and a receiver. An attribute is selected as a target attribute. The remaining attributes are input into a first machine learning model (e.g., an NLP model) including at least one embedding layer and a hidden layer, the first machine learning model being trained to predict the target attribute. After training the model, each of the remaining attributes is converted into a first vector using the at least one embedding layer of the machine learning model to form a first set of vectors. The first set of vectors is stored and subsequently input into a second machine learning model to perform at least one second task different from the first task.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to U.S. patent application No. 17 / 490,325 filed on September 30, 2021, the disclosure of which is hereby incorporated by reference in its entirety. Background Art 1. Technical Field

[0004] The present disclosure generally relates to learning a continuous embedding space, and in some non-limiting embodiments or aspects, to systems, methods, and computer program products for learning a continuous embedding space for real-time payment (RTP) transactions.

[0005] 2. Technical considerations

[0006] Real-time payment (RTP) transactions are an emerging type of financial transactions. RTP transactions always involve at least a sender and a receiver (e.g., a debtor and a creditor, respectively). For certain tasks, such as risk management or behavioral analysis, it may be necessary to understand RTP transaction data.

[0007] However, some techniques require domain knowledge to better understand the data and select the most relevant features for machine learning and artificial intelligence modeling. In addition, some types of data, such as categorical data (e.g., transaction type, certain account information, and / or certain customer information) are not suitable for machine learning models. In addition, some techniques focus on one type of embedding (e.g., consumer, issuer, or account-level embedding) or a portion of transaction entities. Using only customer, client, or account-level embeddings results in embeddings that do not include transaction-level information and result in a cold start for new customers' data. Using only a portion of transaction entities results in embeddings that do not include issuer or account information.

[0008] Therefore, there is a need for improved techniques for learning RTP transaction embedding. Summary of the invention

[0009] According to a non-limiting embodiment or aspect, a method is provided, comprising: using at least one processor to receive RTP data associated with a plurality of RTP transactions, wherein the RTP data of each of the plurality of RTP transactions comprises a plurality of attributes, the plurality of attributes comprising a sender and a receiver; using the at least one processor to select one of the plurality of attributes as a target attribute, wherein the plurality of attributes other than the target attribute comprises a plurality of remaining attributes; using the at least one processor to train a first machine learning model to perform a first task, the first task comprising predicting the target attribute based on inputting the plurality of remaining attributes of each of the plurality of RTP transactions into the first machine learning model, wherein the first machine learning model comprises at least one embedding layer and at least one hidden layer; after training the model, using the at least one processor, converting each of the plurality of remaining attributes into a first vector using the at least one embedding layer of the first machine learning model to form a first group of vectors; using the at least one processor to store the first group of vectors; and using the at least one processor to input the first group of vectors into a second machine learning model to perform at least one second task different from the first task.

[0010] In a non-limiting embodiment or aspect, the RTP data of each RTP transaction in the plurality of RTP transactions includes a plurality of fields, each corresponding field in the plurality of fields corresponding to a corresponding attribute in the plurality of attributes. In some non-limiting embodiments or aspects, the first machine learning model includes a natural language processing (NLP) model, and wherein the training further includes inputting each of the plurality of attributes of each RTP transaction into the NLP model as a word of a sentence using the at least one processor. In some non-limiting embodiments or aspects, the target attribute includes a continuous value field. In some non-limiting embodiments or aspects, the method further includes generating a plurality of categories using the at least one processor; and using the at least one processor, based on the value of the target attribute, classifying each target attribute into one of the plurality of categories, wherein the plurality of categories each represent a range of values ​​for the continuous value field. In some non-limiting embodiments or aspects, training the first machine learning model to perform the first task also includes: using the at least one processor, creating a negative sample of the target attribute based on the equal probability of selecting any single target attribute classified into one of the multiple categories, wherein creating the negative sample includes: using the at least one processor to determine the frequency of occurrence of the target attribute in each of the multiple categories; and using the at least one processor, based on the frequency of occurrence of the target attribute in each of the multiple categories, performing negative sampling of the target attribute. In some non-limiting embodiments or aspects, the second machine learning model is a fraud detection model. In some non-limiting embodiments or aspects, the method also includes: using at least one processor to generate a graph including a plurality of nodes based on the first set of vectors, wherein each of the plurality of nodes represents one of the plurality of RTP transactions.

[0011] According to a non-limiting embodiment or aspect, a system comprising at least one processor is provided, wherein the at least one processor is programmed or configured to: receive RTP data associated with a plurality of RTP transactions, wherein the RTP data of each of the plurality of RTP transactions comprises a plurality of attributes, wherein the plurality of attributes comprises a sender and a receiver; select one of the plurality of attributes as a target attribute, wherein the plurality of attributes other than the target attribute comprises a plurality of remaining attributes; train a first machine learning model to perform a first task, wherein the first task comprises predicting the target attribute based on inputting the plurality of remaining attributes of each of the plurality of RTP transactions into the first machine learning model, wherein the first machine learning model comprises at least one embedding layer and at least one hidden layer; after training the model, convert each of the plurality of remaining attributes into a first vector using the at least one embedding layer of the first machine learning model to form a first set of vectors; store the first set of vectors; and input the first set of vectors into a second machine learning model to perform at least one second task different from the first task.

[0012] In some non-limiting embodiments or aspects, the RTP data of each RTP transaction in the plurality of RTP transactions includes a plurality of fields, and each corresponding field in the plurality of fields corresponds to a corresponding attribute in the plurality of attributes. In some non-limiting embodiments or aspects, the first machine learning model includes an NLP model, and wherein when training the first machine learning model to perform the first task, the at least one processor is also programmed or configured to: input each of the plurality of attributes of each RTP transaction into the NLP model as a word of a sentence. In some non-limiting embodiments or aspects, the target attribute includes a continuous value field. In some non-limiting embodiments or aspects, the processor is also programmed or configured to: generate a plurality of categories; and based on the value of the target attribute, store each target attribute in one of the plurality of categories, wherein the plurality of categories each represent a value range of the continuous value field. In some non-limiting embodiments or aspects, when training the first machine learning model to perform the first task, the at least one processor is also programmed or configured to: create a negative sample of the target attribute based on the equal probability of selecting any single target attribute classified into one of the multiple categories, wherein creating the negative sample includes: using at least one processor to determine the frequency of occurrence of the target attribute in each of the multiple categories; and using the at least one processor, based on the frequency of occurrence of the target attribute in each of the multiple categories, performing negative sampling of the target attribute. In some non-limiting embodiments or aspects, the second machine learning model includes a fraud detection model. In some non-limiting embodiments or aspects, the at least one processor is also programmed or configured to: generate a graph including a plurality of nodes based on the first set of vectors, wherein each of the plurality of nodes represents one of the plurality of RTP transactions.

[0013] According to a non-limiting embodiment or aspect, there is provided a computer program product comprising at least one non-transitory computer-readable medium, the at least one non-transitory computer-readable medium comprising one or more instructions, which, when executed by at least one processor, cause the at least one processor to: receive RTP data associated with a plurality of RTP transactions, wherein the RTP data of each of the plurality of RTP transactions comprises a plurality of attributes, the plurality of attributes comprising a sender and a receiver; select one of the plurality of attributes as a target attribute, wherein the plurality of attributes other than the target attribute comprises a plurality of remaining attributes; train a first machine learning model to perform a first task, the first task comprising predicting the target attribute based on inputting the plurality of remaining attributes of each of the plurality of RTP transactions into the first machine learning model, wherein the first machine learning model comprises at least one embedding layer and at least one hidden layer; after training the model, convert each of the plurality of remaining attributes into a first vector using the at least one embedding layer of the first machine learning model to form a first set of vectors; store the first set of vectors; and input the first set of vectors into a second machine learning model to perform at least one second task different from the first task.

[0014] In a non-limiting embodiment or aspect, the RTP data of each RTP transaction in the plurality of RTP transactions includes a plurality of fields, each respective field in the plurality of fields corresponding to a respective attribute in the plurality of attributes. In some non-limiting embodiments or aspects, the first machine learning model includes an NLP model, and wherein the one or more instructions that cause the at least one processor to train the first machine learning model to perform the first task further cause the at least one processor to: input each of the plurality of attributes of each RTP transaction into the NLP model as a word of a sentence, wherein the target attribute includes a continuous value field. In some non-limiting embodiments or aspects, the method further generates a plurality of categories; based on the value of the target attribute, stores each target attribute into one of the plurality of categories, wherein the plurality of categories each represent a range of values ​​of the continuous value field; based on the equal probability of selecting any single target attribute classified into one of the plurality of categories, creates a negative sample of the target attribute, wherein creating the negative sample comprises: using at least one processor to determine the frequency with which the target attribute appears in each of the plurality of categories; and using the at least one processor, based on the frequency with which the target attribute appears in each of the plurality of categories, performing negative sampling of the target attribute; and generating a graph comprising a plurality of nodes based on the first set of vectors, wherein each of the plurality of nodes represents an RTP transaction in the plurality of RTP transactions. In some non-limiting embodiments or aspects, the second machine learning model comprises a fraud detection model.

[0015] Additional non-limiting embodiments or aspects are set forth in the following numbered clauses:

[0016] Clause 1: A computer-implemented method, comprising: using at least one processor to receive RTP data associated with a plurality of RTP transactions, wherein the RTP data of each of the plurality of RTP transactions includes a plurality of attributes, the plurality of attributes including a sender and a receiver; using the at least one processor to select one of the plurality of attributes as a target attribute, wherein the plurality of attributes other than the target attribute include a plurality of remaining attributes; using the at least one processor to train a first machine learning model to perform a first task, the first task comprising predicting the target attribute based on inputting the plurality of remaining attributes of each of the plurality of RTP transactions into the first machine learning model, wherein the first machine learning model comprises at least one embedding layer and at least one hidden layer; after training the model, using the at least one processor, converting each of the plurality of remaining attributes into a first vector using the at least one embedding layer of the first machine learning model to form a first set of vectors; using the at least one processor to store the first set of vectors; and using the at least one processor to input the first set of vectors into a second machine learning model to perform at least one second task different from the first task.

[0017] Clause 2: The method of clause 1, wherein the RTP data of each RTP transaction in the plurality of RTP transactions comprises a plurality of fields, each respective field in the plurality of fields corresponding to a respective attribute in the plurality of attributes.

[0018] Clause 3: A method as described in clause 1 or 2, wherein the first machine learning model includes a NLP model, and wherein the training also includes using the at least one processor to input each of the multiple attributes of each RTP transaction as a word of a sentence into the NLP model.

[0019] Clause 4: The method of any one of clauses 1 to 3, wherein the target attribute comprises a continuous-valued field.

[0020] Clause 5: The method as described in any one of Clauses 1 to 4 further includes: generating a plurality of categories using the at least one processor; and classifying each target attribute into one of the plurality of categories based on the value of the target attribute using the at least one processor, wherein the plurality of categories each represent a value range of the continuous value field.

[0021] Clause 6: A method as described in any one of clauses 1 to 5, wherein training the first machine learning model to perform the first task further includes: using the at least one processor to create a negative sample of the target attribute based on an equal probability of selecting any single target attribute classified into one of the multiple categories, wherein creating the negative sample includes: using the at least one processor to determine the frequency with which the target attribute appears in each of the multiple categories; and using the at least one processor to perform negative sampling of the target attribute based on the frequency with which the target attribute appears in each of the multiple categories.

[0022] Clause 7: A method as described in any of clauses 1 to 6, wherein the second machine learning model comprises a fraud detection model.

[0023] Clause 8: The method of any one of clauses 1 to 7, further comprising: using at least one processor to generate a graph comprising a plurality of nodes based on the first set of vectors, wherein each node of the plurality of nodes represents an RTP transaction of the plurality of RTP transactions.

[0024] Item 9: A system comprising at least one processor, wherein the at least one processor is programmed or configured to: receive RTP data associated with a plurality of RTP transactions, wherein the RTP data of each of the plurality of RTP transactions comprises a plurality of attributes, wherein the plurality of attributes comprises a sender and a receiver; select one of the plurality of attributes as a target attribute, wherein the plurality of attributes other than the target attribute comprises a plurality of remaining attributes; train a first machine learning model to perform a first task, wherein the first task comprises predicting the target attribute based on inputting the plurality of remaining attributes of each of the plurality of RTP transactions into the first machine learning model, wherein the first machine learning model comprises at least one embedding layer and at least one hidden layer; after training the model, convert each of the plurality of remaining attributes into a first vector using the at least one embedding layer of the first machine learning model to form a first set of vectors; store the first set of vectors; and input the first set of vectors into a second machine learning model to perform at least one second task different from the first task.

[0025] Clause 10: The system of clause 9, wherein the RTP data of each RTP transaction in the plurality of RTP transactions comprises a plurality of fields, each respective field in the plurality of fields corresponding to a respective attribute in the plurality of attributes.

[0026] Clause 11: A system as described in clause 9 or 10, wherein the first machine learning model includes a NLP model, and wherein when training the first machine learning model to perform the first task, the at least one processor is also programmed or configured to: input each of the multiple attributes of each RTP transaction into the NLP model as a word of a sentence.

[0027] Clause 12: The system of any of clauses 9 to 11, wherein the target attribute comprises a continuous-valued field.

[0028] Clause 13: A system as described in any of clauses 9 to 12, wherein the processor is further programmed or configured to: generate a plurality of categories; and based on the value of the target attribute, store each target attribute into one of the plurality of categories, wherein the plurality of categories each represent a range of values ​​for the continuous value field.

[0029] Clause 14: A system as described in any of clauses 9 to 13, wherein when training the first machine learning model to perform the first task, the at least one processor is also programmed or configured to: create a negative sample of the target attribute based on an equal probability of selecting any single target attribute classified in one of the multiple categories, wherein creating the negative sample includes: using at least one processor to determine the frequency of occurrence of the target attribute in each of the multiple categories; and using the at least one processor, based on the frequency of occurrence of the target attribute in each of the multiple categories, performing negative sampling of the target attribute.

[0030] Clause 15: A system as described in any of clauses 9 to 14, wherein the second machine learning model comprises a fraud detection model.

[0031] Clause 16: The system of any one of clauses 9 to 15, wherein the at least one processor is further programmed or configured to: generate a graph comprising a plurality of nodes based on the first set of vectors, wherein each of the plurality of nodes represents an RTP transaction of the plurality of RTP transactions.

[0032] Clause 17: A computer program product comprising at least one non-transitory computer-readable medium, the at least one non-transitory computer-readable medium comprising one or more instructions, which, when executed by at least one processor, cause the at least one processor to: receive RTP data associated with a plurality of RTP transactions, wherein the RTP data of each of the plurality of RTP transactions comprises a plurality of attributes, the plurality of attributes comprising a sender and a receiver; select one of the plurality of attributes as a target attribute, wherein the plurality of attributes other than the target attribute comprises a plurality of remaining attributes; train a first machine learning model to perform a first task, the first task comprising predicting the target attribute based on inputting the plurality of remaining attributes of each of the plurality of RTP transactions into the first machine learning model, wherein the first machine learning model comprises at least one embedding layer and at least one hidden layer; after training the model, convert each of the plurality of remaining attributes into a first vector using the at least one embedding layer of the first machine learning model to form a first set of vectors; store the first set of vectors; and input the first set of vectors into a second machine learning model to perform at least one second task different from the first task.

[0033] Clause 18: The computer program product of Clause 17, wherein the RTP data of each RTP transaction in the plurality of RTP transactions comprises a plurality of fields, each respective field in the plurality of fields corresponding to a respective attribute in the plurality of attributes.

[0034] Clause 19: A computer program product as described in clause 17 or 18, wherein the first machine learning model comprises a NLP model, and wherein the one or more instructions that cause the at least one processor to train the first machine learning model to perform the first task also cause the at least one processor to: input each of the multiple attributes of each RTP transaction into the NLP model as a word of a sentence.

[0035] Clause 20: The computer program product of any of clauses 17 to 19, wherein the target attribute comprises a continuous-valued field.

[0036] Clause 21: A computer program product as described in any of clauses 17 to 20, wherein the one or more instructions that cause the at least one processor to train the first machine learning model to perform the first task also cause the at least one processor to: generate a plurality of categories; and based on the value of the target attribute, store each target attribute into one of the plurality of categories, wherein the plurality of categories each represent a range of values ​​for the continuous value field.

[0037] Clause 22: A computer program product as described in any of clauses 17 to 21, wherein the one or more instructions that cause the at least one processor to train the first machine learning model to perform the first task also cause the at least one processor to: create a negative sample of the target attribute based on an equal probability of selecting any single target attribute classified in one of the multiple categories, wherein creating the negative sample comprises: determining, using at least one processor, a frequency with which the target attribute appears in each of the multiple categories; and performing negative sampling of the target attribute based on the frequency with which the target attribute appears in each of the multiple categories, using the at least one processor.

[0038] Clause 23: A computer program product as described in any of clauses 17 to 22, wherein the second machine learning model comprises a fraud detection model.

[0039] Clause 24: A computer program product as described in any of clauses 17 to 23, wherein the one or more instructions that cause the at least one processor to train the first machine learning model to perform the first task also cause the at least one processor to: generate a graph comprising a plurality of nodes based on the first set of vectors, wherein each of the plurality of nodes represents an RTP transaction among the plurality of RTP transactions.

[0040] These and other features and characteristics of the present disclosure, as well as methods of operation and functions of related structural elements and combinations of parts and economies of manufacture, will become more apparent when considering the following description and appended claims with reference to the accompanying drawings, all of which form a part of this specification, wherein like reference numerals designate corresponding parts in the various figures. It is to be expressly understood, however, that the drawings are for illustration and description purposes only and are not intended as a definition of the limits of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Additional advantages and details are explained in more detail below with reference to a non-limiting exemplary embodiment shown in the schematic drawings, in which:

[0042] Figure 1 is a schematic diagram of a system for learning a continuous embedding space for real-time payment (RTP) transactions according to a non-limiting embodiment or aspect;

[0043] Figure 2 is a flow chart of a process for learning a continuous embedding space for RTP transactions according to a non-limiting embodiment or aspect;

[0044] Figure 3 yes Figure 2 Flowcharts of non-limiting embodiments or exemplary implementations of aspects of the processes in;

[0045] Figure 4 yes Figure 2 illustrative diagrams of non-limiting embodiments or aspects of the processes in;

[0046] Figure 5 is an exemplary input data table including data from a plurality of RTP transactions;

[0047] Fig. 6A and Figure 6B They are the distribution of target attributes and the use Figure 5 An exemplary field distribution bar chart of an example of frequent negative sampling of fields of an input data table in;

[0048] Fig. 7A and Figure 7B are an exemplary first machine learning model and an exemplary embedding vector generated for an exemplary attribute, respectively;

[0049] Figure 8 is a diagram of a non-limiting embodiment or aspect of an environment in which the methods, systems, and / or computer program products described herein may be implemented according to the principles of the presently disclosed subject matter; and

[0050] Fig. 9 Example components of an apparatus for use in conjunction with non-limiting embodiments or aspects are shown. DETAILED DESCRIPTION

[0051] For the purposes of the following description, the terms "end," "upper," "lower," "right," "left," "vertical," "horizontal," "top," "bottom," "lateral," "longitudinal," and their derivatives shall relate to the orientation of the embodiments in the accompanying drawings. However, it shall be understood that the embodiments may employ various alternative variations and step orders, except where expressly specified to the contrary. It shall also be understood that the specific devices and processes illustrated in the accompanying drawings and described in the following description are merely exemplary embodiments or aspects of the present disclosure. Accordingly, specific dimensions and other physical characteristics relating to the embodiments or aspects disclosed herein shall not be considered limiting.

[0052] The aspects, components, elements, structures, actions, steps, functions, instructions, etc. used herein should not be understood as critical or necessary unless explicitly described as such. Also, as used herein, the article "one" is intended to include one or more items, and can be used interchangeably with "one or more" and "at least one". In addition, as used herein, the term "set" is intended to include one or more items (such as related items, unrelated items, combinations of related items and unrelated items, etc.), and can be used interchangeably with "one or more" or "at least one". In the case of wishing only one item, the term "one" or similar language is used. Also, as used herein, the term "having" and the like are intended to be open terms. In addition, unless otherwise explicitly stated, the phrase "based on" is intended to mean "based at least in part on".

[0053] As used herein, the term "acquirer" may refer to an entity that is licensed and / or approved by a transaction service provider to initiate transactions (e.g., payment transactions) using a payment device associated with the transaction service provider. Transactions that may be initiated by an acquirer may include payment transactions (e.g., purchases, original credit transactions (OCTs), account funding transactions (AFTs), etc.). In some non-limiting embodiments or aspects, an acquirer may be a financial institution, such as a bank. As used herein, the term "acquirer system" may refer to one or more computing devices operated by or on behalf of an acquirer, such as a server computer executing one or more software applications.

[0054] As used herein, the term "account identifier" may include one or more primary account numbers (PANs), tokens, or other identifiers associated with a customer account. The term "token" may refer to an identifier used as a replacement or substitute identifier for an original account identifier such as a PAN. An account identifier may be alphanumeric or any combination of characters and / or symbols. A token may be associated with a PAN or other original account identifier in one or more data structures (e.g., one or more databases, etc.) such that the token can be used to conduct transactions without directly using the original account identifier. In some examples, an original account identifier such as a PAN may be associated with multiple tokens for different individuals or purposes.

[0055] As used herein, the term "communication" may refer to the reception, acceptance, transmission, transfer, provision, etc. of data (e.g., information, signals, messages, instructions, commands, etc.). A unit (e.g., a device, a system, a component of a device or system, a combination thereof, etc.) communicating with another unit means that the unit is able to directly or indirectly receive information from the other unit and / or send information to the other unit. This may refer to a direct or indirect connection (e.g., a direct communication connection, an indirect communication connection, etc.) that is wired and / or wireless in nature. In addition, although the information sent can be modified, processed, relayed, and / or routed between the first unit and the second unit, the two units may also communicate with each other. For example, even if the first unit passively receives information and does not actively send information to the second unit, the first unit may communicate with the second unit. As another example, if at least one intermediate unit processes the information received from the first unit and transmits the processed information to the second unit, the first unit may communicate with the second unit.

[0056] As used herein, the term "computing device" may refer to one or more electronic devices configured to process data. In some examples, a computing device may include the necessary components to receive, process, and output data, such as a processor, a display, a memory, an input device, a network interface, etc. The computing device may be a mobile device. As an example, a mobile device may include a cellular phone (e.g., a smartphone or a standard cellular phone), a portable computer, a wearable device (e.g., a watch, glasses, lenses, clothing, etc.), a personal digital assistant (PDA), and / or other similar devices. A computing device may also be a desktop computer or other form of non-mobile computer.

[0057] As used herein, the terms "electronic wallet" and "electronic wallet application" refer to one or more electronic devices and / or software applications configured to initiate and / or conduct payment transactions. For example, an electronic wallet may include a mobile device executing an electronic wallet application, and may also include server-side software and / or databases for maintaining and providing transaction data to the mobile device. An "electronic wallet provider" may include an entity that provides and / or maintains electronic wallets for customers, such as Google. Android Apple Samsung And / or other similar electronic payment systems. In some non-limiting examples, the issuing bank may be an electronic wallet provider.

[0058] As used herein, the term "issuer institution" may refer to one or more entities, such as a bank, that provide accounts to customers for conducting transactions (e.g., payment transactions), such as initiating credit and / or debit payments. For example, an issuer institution may provide customers with account identifiers, such as PANs, that uniquely identify one or more accounts associated with the customer. Account identifiers may be implemented on portable financial devices, such as physical financial instruments (e.g., payment cards), and / or may be electronic and used for electronic payments. The term "issuer system" refers to one or more computer devices operated by or on behalf of an issuer institution, such as a server computer executing one or more software applications. For example, an issuer system may include one or more authorization servers for authorizing transactions.

[0059] As used herein, the term "merchant" may refer to a person or entity that provides goods and / or services, or access to goods and / or services, to a customer based on a transaction, such as a payment transaction. The term "merchant" or "merchant system" may also refer to one or more computer systems operated by or on behalf of a merchant, such as a server computer that executes one or more software applications. As used herein, a "point of sale (POS) system" may refer to one or more computers and / or peripheral devices used by a merchant to conduct payment transactions with customers, including one or more card readers, near field communication (NFC) receivers, RFID receivers and / or other contactless transceivers or receivers, contact-based receivers, payment terminals, computers, servers, input devices, and / or other similar devices that can be used to initiate payment transactions.

[0060] As used herein, the term "payment device" may refer to an electronic payment device, a portable financial device, a payment card (e.g., a credit or debit card), a gift card, a smart card, smart media, a payroll card, a healthcare card, a wristband, a machine-readable medium containing account information, a key chain device or pendant, an RFID transponder, a retailer discount or membership card, a cellular phone, an electronic wallet mobile application, a PDA, a pager, a security card, a computing device, an access card, a wireless terminal, a transponder, etc. In some non-limiting embodiments or aspects, a payment device may include volatile or non-volatile memory to store information (e.g., an account identifier, the name of an account holder, etc.).

[0061] As used herein, the term "payment gateway" may refer to an entity and / or a payment processing system operated by or on behalf of such an entity, which provides payment services (e.g., transaction service provider payment services, payment processing services, etc.) to one or more merchants (e.g., transaction service provider payment services, payment processing services, etc.). The payment service may be associated with the use of a portable financial device managed by a transaction service provider. As used herein, the term "payment gateway system" may refer to one or more computer systems, computer devices, servers, server groups, etc. operated by or on behalf of a payment gateway.

[0062] As used herein, the term "server" may refer to or include one or more computing devices operated by or facilitating communications and processing by multiple parties in a network environment such as the Internet, but it should be understood that communications may be facilitated through one or more public or private network environments, and that various other arrangements may be possible. In addition, multiple computing devices (e.g., servers, point-of-sale (POS) devices, mobile devices, etc.) that communicate directly or indirectly in a network environment may constitute a "system." As used herein, references to a "server" or "processor" may refer to a previously described server and / or processor stated to perform a previous step or function, a different server and / or processor, and / or a combination of servers and / or processors. For example, as used in the specification and claims, a first server and / or first processor stated to implement a first step or function may refer to the same or different server and / or processor stated to implement a second step or function.

[0063] As used herein, the term "transaction service provider" may refer to an entity that receives transaction authorization requests from merchants or other entities and, in some cases, provides payment assurance through an agreement between the transaction service provider and the issuer organization. For example, a transaction service provider may include, for example The term "transaction processing system" may refer to one or more computer systems operated by or on behalf of a transaction service provider, such as a transaction processing server executing one or more software applications. A transaction processing server may include one or more processors and, in some non-limiting embodiments or aspects, may be operated by or on behalf of a transaction service provider.

[0064] The present disclosure provides a new technique for learning RTP transaction embedding by simultaneously learning consumer, issuer and / or account-level information and transaction-level information. For example, the disclosed technique describes receiving RTP data including multiple attributes, including a sender and a receiver. Select one attribute from the multiple attributes as a target attribute. The remaining attributes are input into a first machine learning model (e.g., an NLP model) including at least one embedding layer and at least one hidden layer, and the first machine learning model is trained to perform a first task (e.g., predict the target attribute). After training the model, each of the remaining attributes is converted into a first vector using the at least one embedding layer of the first machine learning model to form a first set of vectors. The first set of vectors is stored and subsequently input into a second machine learning model, which is trained to perform at least one second task different from the first task. One advantage of using RTP transaction data is that there are paired relationships in RTP transactions. In other words, RTP transactions always involve at least two entities, namely a sender and a receiver (e.g., a debtor and a creditor, respectively). An additional advantage of the disclosed subject matter is that the machine learning model can learn continuous vector embeddings of RTP transactions without domain knowledge. Once a machine learning model is trained, these embeddings can be used as input to other machine learning models, such as fraud detection models, product recommendation models, non-compliance detection models, anomaly detection models, etc. Thus, the disclosed technology solves problems associated with fraud detection, risk management, and behavioral analysis.

[0065] Reference now Figure 1 , Figure 1 A system 100 for learning a continuous embedding space of RTP transactions according to a non-limiting embodiment or aspect is depicted. The system 100 includes a transaction database 102, a first machine learning model system 104, an embedding vector database 106, and / or a second machine learning model system 108.

[0066] The transaction database 102 may include one or more devices capable of receiving information from the first machine learning model system 104 and / or the second machine learning model system 108 and / or transmitting information to the first machine learning model system and / or the second machine learning model system. For example, the transaction database 102 may include a computing device, such as a computer, a server, a server group, and / or other similar devices. In some non-limiting embodiments or aspects, the transaction database 102 may communicate with a data storage device, which may be local or remote to the transaction database 102. In some non-limiting embodiments or aspects, the transaction database 102 may be capable of receiving information from the data storage device, storing information in the data storage device, transmitting information to the data storage device, or searching for information stored in the data storage device.

[0067] The first machine learning model system 104 may include one or more devices capable of receiving information from the transaction database 102, the embedding vector database 106, and / or the second machine learning system 108 and / or transmitting information to the transaction database, the embedding vector database, and / or the second machine learning system. For example, the first machine learning model system 104 may include a computing device, such as a computer, a server, a group of servers, and / or other similar devices. In some non-limiting embodiments or aspects, as described herein, the first machine learning model system 104 may be associated with a financial institution, a transaction service provider, and / or an issuer. For example, the first machine learning model system 104 may be operated by a financial institution, a transaction service provider, and / or an issuer.

[0068] The embedding vector database 106 may include one or more devices capable of receiving information from the first machine learning model system 104 and / or the second machine learning model system 108 and / or transmitting information to the first machine learning model system and / or the second machine learning model system. For example, the embedding vector database 106 may include a computing device, such as a computer, a server, a group of servers, and / or other similar devices. In some non-limiting embodiments or aspects, the embedding vector database 106 may communicate with a data storage device, which may be local or remote to the embedding vector database 106. In some non-limiting embodiments or aspects, the embedding vector database 106 may be capable of receiving information from the data storage device, storing information in the data storage device, transmitting information to the data storage device, or searching for information stored in the data storage device.

[0069] The second machine learning model system 108 may include one or more devices capable of receiving information from the transaction database 102, the first machine learning model system 104, and / or the embedding vector database 106 and / or transmitting information to the transaction database, the first machine learning model system, and / or the embedding vector database. For example, the second machine learning model system may include a computing device, such as a computer, a server, a group of servers, and / or other similar devices. In some non-limiting embodiments or aspects, the second machine learning model system 108 may be associated with a financial institution, a transaction service provider, and / or an issuer, as described herein. For example, the second machine learning model system 108 may be operated by a financial institution, a transaction service provider, and / or an issuer.

[0070] In some non-limiting embodiments or aspects, as described herein, the first machine learning model system 104 may receive (e.g., from the transaction database 102) RTP data associated with a plurality of RTP transactions. For example, as described herein, the RTP data for each RTP transaction may include a plurality of attributes. As described herein, the plurality of attributes may include a sender and a receiver. Additionally or alternatively, the first machine learning model system 104 may select one of the plurality of attributes as a target attribute, as described herein. The plurality of attributes in addition to the target attribute may include a plurality of remaining attributes, as described herein. Additionally or alternatively, the first machine learning model system 104 may train a first machine learning model to perform a first task, the first task comprising predicting a target attribute based on inputting a plurality of remaining attributes of each RTP transaction in a plurality of RTP transactions into a first machine learning model, as described herein. As described herein, the first machine learning model may include at least one embedding layer and at least one hidden layer. Additionally or alternatively, the first machine learning model system 104 may, after training the model, convert each of the plurality of attributes into a first vector using at least one embedding layer of the first machine learning model to form a first set of vectors, as described herein. Additionally or alternatively, the first machine learning model system 104 can store the first set of vectors (e.g., in an embedding vector database 106), as described herein. Additionally or alternatively, the first machine learning model system 104 can input the first set of vectors into a second machine learning model to perform at least one second task different from the first task, as described herein.

[0071] supply Figure 1 The number and arrangement of devices and networks shown in are examples. Figure 1 Additional devices and / or networks, fewer devices and / or networks, different devices and / or networks, or differently arranged devices and / or networks than those shown in . In addition, Figure 1 Two or more of the devices shown in the figure may be implemented in a single device, or Figure 1 The single device shown in the embodiment may be implemented as multiple distributed devices. Additionally or alternatively, one or more devices of the environment 100 may perform one or more functions described as being performed by another set of devices of the environment 100.

[0072] Reference now Figure 2 , illustrates a process 200 for learning a continuous embedding space for RTP transactions according to a non-limiting embodiment or aspect. Figure 2The steps shown in are for example purposes only. It will be appreciated that in non-limiting embodiments or aspects, additional, fewer, different, and / or different orders of steps may be used. In some non-limiting embodiments or aspects, one or more of the steps of process 200 may be performed (e.g., completely, partially, etc.) by the first machine learning model system 104 (e.g., one or more devices of the first machine learning model system 104). In some non-limiting embodiments or aspects, one or more of the steps of process 200 (e.g., completely, partially, etc.) may be performed by another device or another group of devices that is separate from or includes the first machine learning model system 104, such as the transaction database 102, the embedding vector database 106, and / or the second machine learning model system 108 (e.g., one or more devices of the second machine learning model system 108).

[0073] like Figure 2 As shown in, at step 202, process 200 may include receiving a data set associated with a plurality of RTP transactions. For example, the first machine learning model system 104 may receive RTP transaction data associated with a plurality of RTP transactions from the transaction database 102. In some non-limiting embodiments or aspects, the RTP transaction data of each RTP transaction in a plurality of RTP transactions may include a plurality of attributes, including a sender and a receiver (e.g., a debtor and a creditor, respectively). Additionally or alternatively, a plurality of attributes may include client, customer and / or account information and / or transaction information. For example, a plurality of attributes may include one or more of the following: creditor bank, creditor routing number, creditor account number, debtor bank, debtor routing number, debtor account number, RTP transaction amount, etc. In some non-limiting embodiments or aspects, the RTP data of each RTP transaction in a plurality of RTP transactions may include a plurality of fields. For example, each corresponding field in a plurality of fields may correspond to a corresponding attribute in a plurality of attributes (e.g., creditor bank, creditor routing number, creditor account number, debtor bank, debtor routing number, debtor account number, transaction amount, etc.).

[0074] like Figure 2As shown in , at step 204, process 200 may include selecting an attribute from a plurality of attributes as a target attribute. For example, the first machine learning model system 104 may select one of the following as the target attribute: a creditor bank, a creditor routing number, a creditor account number, a debtor bank, a debtor routing number, a debtor account number, or a transaction amount. In some non-limiting embodiments or aspects, the plurality of attributes may include a plurality of remaining attributes (i.e., any attributes that are not selected as target attributes). For example, if the first machine learning model system 104 selects the transaction amount as the target attribute, all other attributes may be considered as a plurality of remaining attributes. In some non-limiting embodiments or aspects, the target attribute may be a continuous value field. For example, the target attribute may be the transaction amount.

[0075] like Figure 2 As shown in , at step 206, process 200 may include training a first machine learning model to perform a first task. For example, the first machine learning model system 104 may train the first machine learning model to predict a target attribute of each of the multiple RTP transactions based on inputting a plurality of remaining attributes of each of the multiple RTP transactions into the first machine learning model. In some non-limiting embodiments or aspects, the first machine learning model may include two or more layers. For example, the first machine learning model may include at least one embedding layer and at least one hidden layer. In some non-limiting embodiments or aspects, the first machine learning model may be a natural language processing (NLP) model. In some non-limiting embodiments or aspects, a continuous bag of attributes model (e.g., a continuous bag of words model) may be used to train the first machine learning model. For example, training the first machine learning model to perform the first task may include inputting each of the remaining attributes of each RTP transaction into the first machine learning model as a word of a sentence, and the target attribute may be a target word to be predicted based on the remaining attributes (e.g., words) of the RTP transaction (e.g., a sentence). In some non-limiting embodiments or aspects, training the first machine learning model to perform the first task may include creating negative samples of RTP data. For example, the first machine learning model system 104 may create negative samples of multiple attributes.

[0076] like Figure 2 As shown in , at step 208, process 200 may include converting a plurality of remaining attributes of each RTP transaction in the plurality of RTP transactions into a first vector to form a first set of vectors. For example, the first machine learning model system 104 may use an embedding layer to convert each of the plurality of remaining attributes into a first vector to form a first set of vectors.

[0077] like Figure 2As shown in , at step 210, process 200 may include storing the first set of vectors. For example, the first machine learning model system 104 may store the first set of vectors in the embedding vector database 106. For example, the first machine learning model system 104 may transmit the first set of vectors to the embedding vector database 106, which may receive and store the first set of vectors.

[0078] like Figure 2 As shown in , at step 212, process 200 may include inputting the first set of vectors into a second machine learning model to perform a second task different from the first task. For example, the embedding vector database 106 may transmit the first set of vectors to the second machine learning model system 108. Additionally or alternatively, the second machine learning model system 108 may input the first set of vectors into its second machine learning model. In some non-limiting embodiments or aspects, the second machine learning model may be a fraud detection model. For example, the second machine learning model system 108 may input the first set of vectors into a fraud detection model. In some non-limiting embodiments or aspects, the second machine learning model may be a product recommendation model, a non-compliance detection model, an anomaly detection model, or the like.

[0079] supply Figure 2 The number and arrangement of the steps shown in are provided as examples. Figure 2 There may be additional steps, fewer steps, different steps, or differently arranged steps compared to the steps shown in . Figure 2 Two or more steps shown in may be implemented in a single step, or Figure 2 A single step shown in can be implemented as multiple steps.

[0080] Reference now Figure 3 , showing Figure 2 An exemplary implementation 300 of a non-limiting embodiment or aspect of the process in. Figure 3The steps shown in are for example purposes only. It will be appreciated that in non-limiting embodiments or aspects, additional, fewer, different and / or different orders of steps may be used. In some non-limiting embodiments or aspects, one or more of the steps of the specific implementation 300 may be performed (e.g., completely, partially, etc.) by the first machine learning model system 104 (e.g., one or more devices of the first machine learning model system 104). In some non-limiting embodiments or aspects, one or more of the steps of the specific implementation 300 may be performed (e.g., completely, partially, etc.) by another device or another group of devices that is separate from or includes the first machine learning model system 104, such as the transaction database 102, the embedding vector database 106, and / or the second machine learning model system 108 (e.g., one or more devices of the second machine learning model system 108).

[0081] like Figure 3 As shown in, at step 302, process 300 can include receiving a data set associated with a plurality of RTP transactions. For example, the first machine learning model system 104 can receive RTP transaction data associated with a plurality of RTP transactions from the transaction database 102.

[0082] like Figure 3 As shown in , at step 304, process 300 may include selecting a continuous value field attribute from multiple attributes as a target attribute. For example, the first machine learning model system 104 may select an attribute that is a continuous value field attribute (e.g., transaction amount) from multiple attributes as the target attribute.

[0083] like Figure 3 As shown in , at step 306, process 300 may include generating a plurality of categories. For example, the first machine learning model system 104 may generate two or more categories. In some non-limiting embodiments or aspects, each of the plurality of categories may represent a range of values ​​for a continuous field value. For example, the first machine learning model system 104 may generate two or more categories representing a range of values ​​for a transaction amount (e.g., a first category associated with a range of $0 to $500, a second category associated with a range of $501 to $1,000, a third category associated with a range of $1,001 to $1,500, etc.).

[0084] like Figure 3As shown in , at step 308, process 300 may include classifying each target attribute into one of a plurality of categories based on the value of the target attribute. For example, the first machine learning model system 104 may classify each transaction amount into one of a plurality of categories representing a certain value range based on the value of the transaction amount (e.g., a transaction amount of $108.11 will be classified into a first category associated with a range of $0 to $500).

[0085] like Figure 3 As shown in , at step 310, process 300 may include inputting each of the multiple attributes into the NLP model. For example, the first machine learning model system 104 may be trained to predict the target attribute based on inputting the multiple remaining attributes of each RTP transaction in the multiple RTP transactions into the NLP model. In some non-limiting embodiments or aspects, the NLP model may include two or more layers. For example, the NLP model may include at least one embedding layer and at least one hidden layer. In some non-limiting embodiments or aspects, the NLP model may include a continuous bag of attributes (CBOA) model (e.g., a continuous bag of words (CBOW) model, in which the RTP transactions are regarded as sentences, and the attributes of the corresponding RTP transactions are input as words of the sentence). For example, training the NLP model to perform the first task may include inputting each of the multiple attributes of each RTP transaction into the NLP model as a word of the sentence to perform the task of predicting the target attribute.

[0086] like Figure 3As shown in , at step 312, process 300 may include performing frequent negative sampling of fields to make training NLP models more efficient. For example, the first machine learning model system 104 may create negative samples by generating a vector representing a false RTP transaction, for which the transaction amount (category) is incorrect for the remaining attributes (e.g., starting with a vector representing a true RTP transaction, the true RTP transaction includes a true target value in a first category and the associated remaining attributes, using those same remaining attributes to generate the vector, but making the category of the target value different, such as a third category). In some non-limiting embodiments or aspects, step 312 may include performing negative sampling by creating negative samples of multiple attributes based on the equal probability of selecting any single attribute classified into one of the multiple categories. In some non-limiting embodiments or aspects, step 312 may include performing frequent negative sampling of fields by determining the frequency with which the target attribute appears in each of the multiple categories, and selecting a false category for the target attribute based on the relative frequency with which the target attribute appears in each corresponding category. For example, the first machine learning model system 104 may determine the frequency at which each transaction amount appears in each category of a plurality of categories (e.g., a value range). In some non-limiting embodiments or aspects, step 312 may include generating negative samples based on the frequency at which the target attribute appears in the category. For example, the first machine learning model system 104 may generate negative samples with false target attributes in a particular category of the plurality of categories based on the probability of the frequency at which the transaction amount appears in the particular category.

[0087] like Figure 3 As shown in , at step 314, process 300 may include converting a plurality of remaining attributes of each RTP transaction in the plurality of RTP transactions into a first vector to form a first set of vectors. For example, the first machine learning model system 104 may use at least one embedding layer to convert each of the plurality of remaining attributes into a first vector to form a first set of vectors.

[0088] like Figure 3 As shown in FIG. 3 , at step 316 , process 300 can include storing the first set of vectors. For example, the first set of vectors can be stored in the embedding vector database 106 .

[0089] like Figure 3 As shown in , at step 318, process 300 may include inputting the first set of vectors into a second machine learning model to perform a second task different from the first task. For example, embedding vector database 106 may input the first set of vectors into second machine learning model system 108. In some non-limiting embodiments or aspects, the second machine learning model may be a fraud detection model.

[0090] supply Figure 3 The number and arrangement of the steps shown in are provided as examples. Figure 3 There may be additional steps, fewer steps, different steps, or differently arranged steps compared to the steps shown in . Figure 3 Two or more steps shown in may be implemented in a single step, or Figure 3 A single step shown in can be implemented as multiple steps.

[0091] Reference now Figure 4 , showing Figure 2 An example diagram 400 of a non-limiting implementation or aspect of the process in. Figure 4 The elements shown in the drawings are for illustrative purposes only. It will be appreciated that additional, fewer and / or different elements may be used in non-limiting embodiments or aspects. Figure 4 As shown in , the graph 400 includes a plurality of nodes 410, a first dashed ellipse 402, a second dashed ellipse 404, a solid ellipse 406, and a block 408. In some non-limiting embodiments or aspects, the process 200 may include generating a vector graph. For example, the first machine learning model system 104 may generate a vector graph representing a plurality of attributes. In some non-limiting embodiments or aspects, a plurality of nodes 410 may represent a plurality of vector embeddings for each of a plurality of attributes of the RTP data for each RTP transaction in a plurality of RTP transactions. In some non-limiting embodiments or aspects, a plurality of nodes 410 may form a cluster. In some non-limiting embodiments or aspects, a cluster of nodes 410 may represent vectors with similar attributes. For example, a plurality of nodes 410 representing a vector of attributes of a day of the week may be clustered together, as indicated by the first dashed ellipse 402. A plurality of nodes 410 representing a vector of attributes of a bank name may be clustered together, as indicated by the second dashed ellipse 404. Multiple nodes 410 representing vectors of attributes of person-to-person (e.g., PTP1), person-to-business (e.g., PTB1), business-to-person (e.g., BTP1), and / or business-to-business (e.g., BTB1) may be clustered together, as indicated by solid ovals 406. Figure 4As shown in , the node representing the vector PTP1 is roughly near the cluster indicated by the solid ellipse 406, however, the reason why the node may overlap with another cluster is because the graph 400 represents the projection of the high-dimensional embedding space into the two-dimensional graph. A plurality of nodes 410 representing multiple categories (e.g., categories 1 to 8, which may be categories of the target attribute transaction amount, as described herein) can be clustered together, as indicated by the block 408. In some non-limiting embodiments or aspects, the cluster indicated by the first dotted ellipse 402, the second dotted ellipse 404, the solid ellipse 406, and the block 408 can be used to prove that the first machine learning model (e.g., an NLP model) has learned the meaning of each attribute without any prior knowledge. In other words, the vectors represent relatively good embeddings of the attributes they represent, because similar attributes have vectors with similar values ​​(and therefore form clusters, because the nodes representing the endpoints of such vectors are close to each other).

[0092] Reference now Figure 5 , shows a screenshot of an exemplary input data table 500 including data from multiple RTP transactions. Figure 5 The elements shown in the drawings are for illustrative purposes only. It will be appreciated that additional, fewer and / or different elements may be used in non-limiting embodiments or aspects. Figure 5 As shown in , input data table 500 includes rows 502 and 504 and columns 506 , 508 , 510 , 512 , 514 , 516 , 518 , 520 , 522 , and 524 .

[0093] In some non-limiting embodiments or aspects, rows 502 and 504 may each include RTP transaction data for one of the plurality of RTP transactions. For example, the data in row 502 may include data from a first RTP transaction, and the data in row 504 may include data from a second RTP transaction. In some non-limiting embodiments or aspects, the number of rows may be equal to the number of the plurality of RTP transactions.

[0094] In some non-limiting embodiments or aspects, columns 506, 508, 510, 512, 514, 516, 518, 520, 522, and 524 may include multiple fields and / or multiple attributes of each RTP transaction in the multiple RTP transactions. For example, each column 506, 508, 510, 512, 514, 516, 518, 520, 522, and 524 may represent one of the multiple fields corresponding to the corresponding attribute in the multiple attributes, including a creditor bank column 506 (e.g., a recipient), a creditor routing number column 508, a creditor account number 510, a column a 512, a column b 514, a miscellaneous column 516, a debtor bank column 518 (e.g., a sender), a debtor routing number column 520, a debtor account column 522, and a transaction amount column 524. In some non-limiting embodiments or aspects, the number of columns may be equal to the number of multiple fields and / or multiple attributes. In some non-limiting embodiments or aspects, data in a column can be selected as a target attribute. For example, data in column 524, transaction amount (e.g., Amt), can be selected as a target attribute.

[0095] Reference now Fig. 6A and Figure 6B , respectively, show an exemplary field distribution bar chart 600 showing the distribution (e.g., frequency) of a target attribute and a bar chart 600 using the Figure 5 An example of frequent negative sampling 610 of the fields of the input data table. Fig. 6AAs shown in , the field distribution bar chart 600 may include categories 604, 606, and 608. In some non-limiting embodiments or aspects, each category 604, 606, and 608 may represent one of the multiple categories. The vertical axis may represent the number of target values ​​in each category and / or the frequency of target values. In some non-limiting embodiments or aspects, each category 604, 606, and 608 may represent a value range for a continuous value field. For example, the first machine learning model system 104 may generate multiple categories 604, 606, and 608 (e.g., amt1, amt2, and amt3), wherein each category 604, 606, and 608 represents a value range for a transaction amount (e.g., amt1 = $0 to $500, amt2 = $501 to $1,000, and amt3 = $1,001 to $1,500). In some non-limiting embodiments or aspects, each target attribute can be stored in one of the categories 604, 606, and 608 based on the value of the target attribute. For example, for category 604 (e.g., amt1 = $0 to $500), category 606 (amt2 = $501 to $1,000), and category 608 (amt3 = $1,001 to $1,500), the first machine learning model system 104 will store the transaction amount of $1,222.88 in category 608 (e.g., amt3 = $501 to $1,000).

[0096] like Figure 6B As shown in , field frequent negative sample 610 includes field frequent negative sample input vector 612 and false target 614. In some non-limiting embodiments or aspects, field frequent negative sample input vector 612 can use input data from real RTP transactions. For example, field frequent negative sample input vector 612 can use input data from Figure 5 , as described above at step 312. In some non-limiting embodiments or aspects, the false target 614 can be a category of the plurality of categories that is different from the true target attribute value of the corresponding row from the input data table 500. For example, the false target 614 can be Fig. 6A The illustrated category is one of the categories 604, 606, and 608, rather than the actual category. Fig. 6A and Figure 6BAs shown in , the target attribute appears more frequently in category 608, amt 3 than in other categories. Therefore, based on the frequency of the target attribute appearing in category 608, the probability of selecting category 608 as a false target 614 for field frequent negative sampling may be higher than selecting category 604 or 606. For example, the probability of selecting amt 3 as a false target value 614 may be proportional to the frequency associated with category 608, and the probability of selecting amt 1 or amt 2 as a false target value 614 may be proportional to the frequency associated with category 604 or category 608, respectively. Thus, performing field frequent negative sampling takes into account the frequency of target values ​​appearing in multiple categories. Therefore, the probability of selecting a specific category among multiple categories is based on the frequency with which the value of the target attribute of each RTP transaction in multiple RTP transactions falls into a specific category.

[0097] supply Fig. 6A and Figure 6B The number and arrangement of the elements shown are provided as examples. Fig. 6A and Figure 6B There may be additional elements, fewer elements, different elements, or a different arrangement of elements than those shown in the drawings.

[0098] Reference now Fig. 7A and Figure 7B , an exemplary first machine learning model 700 and results 710 are shown. Fig. 7A As shown in , the first machine learning model 700 may include an input 702, an embedding layer 704, a hidden layer 706, and an output layer 708. The first machine learning model 700 may receive an input 702. For example, the first machine learning model system 104 may train the first machine learning model to predict a target attribute of each of the multiple RTP transactions based on inputting a plurality of remaining attributes of each of the multiple RTP transactions into the first machine learning model. In some non-limiting embodiments or aspects, the first machine learning model may be a natural language processing (NLP) model. In some non-limiting embodiments or aspects, a continuous bag of attributes model (e.g., a continuous bag of words model) may be used to train the first machine learning model, as described herein. For example, training the first machine learning model to perform the first task may include inputting each of the remaining attributes of each RTP transaction into the first machine learning model as a word of a sentence, and the target attribute may be a target word to be predicted based on the remaining attributes (e.g., words) of the RTP transaction (e.g., a sentence). In some non-limiting embodiments or aspects, the first machine learning model may include an embedding layer 704 and a hidden layer 706. In some non-limiting embodiments or aspects, the first machine learning model 700 can generate an output (e.g., a prediction of a target attribute) at an output layer 708 based on the input propagated forward through the embedding layer 704 and the hidden layer 706.

[0099] like Figure 7B , the result 710 may include a vocabulary column 712 and a vector column 714. In some non-limiting embodiments or aspects, the vocabulary column 712 may include attributes of the RTP data (e.g., input 702). For example, the vocabulary column 712 may include multiple attributes. In some non-limiting embodiments or aspects, the vector column 714 may include a vector generated for each corresponding attribute based on the embedding layer 704. For example, the vector column 714 may include a vector representing each of the multiple attributes generated by the embedding layer 704.

[0100] supply Fig. 7A and Figure 7B The number and arrangement of the elements shown are provided as examples. Fig. 7A and Figure 7B There may be additional elements, fewer elements, different elements, or a different arrangement of elements than those shown in the drawings.

[0101] Reference now Figure 8 , illustrates a non-limiting embodiment or aspect of an environment in which the methods, systems, and / or computer program products described herein may be implemented according to the principles of the presently disclosed subject matter. Figure 8 As shown in , environment 800 includes a transaction service provider system 802, an issuer system 804, a client device 806, a merchant system 808, an acquirer system 810, and a communication network 812. In some non-limiting embodiments or aspects, each of the transaction database 102, the first machine learning system 104, the embedding vector database 106, and the second machine learning model system 108 can be implemented by the transaction service provider system 802 (e.g., a portion of the transaction service provider system). In some non-limiting embodiments or aspects, at least one of the transaction database 102, the first machine learning system 104, the embedding vector database 106, and the second machine learning model system 108 can be implemented by another system, another device, another group of systems or another group of devices, such as the issuer system 804, the merchant system 808, the acquirer system 810, etc., (e.g., a portion thereof), which is separate from the transaction service provider system 802 or includes the transaction service provider system.

[0102] The transaction service provider system 802 may include one or more devices capable of receiving information from an issuer system 804, a client device 806, a merchant system 808, and / or an acquirer system 810 via a communication network 812 and / or transmitting information to the issuer system, the client device, the merchant system, and / or the acquirer system. For example, the transaction service provider system 802 may include a computing device, such as a server (e.g., a transaction processing server), a group of servers, and / or other similar devices. In some non-limiting embodiments or aspects, the transaction service provider system 802 may be associated with a transaction service provider described herein. In some non-limiting embodiments or aspects, the transaction service provider system 802 may communicate with a data storage device, which may be local or remote to the transaction service provider system 802. In some non-limiting embodiments or aspects, the transaction service provider system 802 is capable of receiving information from the data storage device, storing information in the data storage device, transmitting information to the data storage device, or searching for information stored in the data storage device.

[0103] The issuer system 804 may include one or more devices capable of receiving information from the transaction service provider system 802, the client device 806, the merchant system 808, and / or the acquirer system 810 via a communication network 812 and / or transmitting information to the transaction service provider system, the client device, the merchant system, and / or the acquirer system. For example, the issuer system 804 may include a computing device, such as a server, a group of servers, and / or other similar devices. In some non-limiting embodiments or aspects, the issuer system 804 may be associated with an issuer institution described herein. For example, the issuer system 804 may be associated with an issuer institution that issues credit accounts, debit accounts, credit cards, debit cards, etc. to users associated with the client device 806.

[0104] The client device 806 may include one or more devices capable of receiving information from and / or transmitting information to the transaction service provider system 802, the issuer system 804, the merchant system 808, and / or the acquirer system 810 via the communication network 812. Additionally or alternatively, each client device 806 may include a device capable of receiving information from and / or transmitting information to other client devices 806 via the communication network 812, another network (e.g., an ad hoc network, a local network, a private network, a virtual private network, etc.), and / or any other suitable communication technology. For example, the client device 806 may include a client device, etc. In some non-limiting embodiments or aspects, the client device 806 may or may not be capable of communicating via a short-range wireless communication connection (e.g., an NFC communication connection, an RFID communication connection, Communication connection, communication connection, etc.) (e.g., from merchant system 808 or from another client device 806) and / or transmit information (e.g., to merchant system 808) via a short-range wireless communication connection.

[0105] The merchant system 808 may include one or more devices capable of receiving information from and / or transmitting information to the transaction service provider system 802, the issuer system 804, the customer device 806, and / or the acquirer system 810 via the communication network 812. The merchant system 808 may also include a communication connection (e.g., an NFC communication connection, an RFID communication connection, a NFC communication connection, a ...RFID communication connection, a NFC communication connection, a RFID communication connection, a RFID communication connection, a NFC communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, a RFID communication connection, Communication connection, In some non-limiting embodiments or aspects, merchant system 808 may include a computing device, such as a server, a group of servers, a client device, a group of client devices, and / or other similar devices. In some non-limiting embodiments or aspects, merchant system 808 may be associated with merchants described herein. In some non-limiting embodiments or aspects, merchant system 808 may include one or more client devices. For example, merchant system 808 may include a client device that allows merchants to transmit information to transaction service provider system 802. In some non-limiting embodiments or aspects, merchant system 808 may include one or more devices that can be used by merchants to carry out payment transactions with users, such as computers, computer systems, and / or peripheral devices. For example, merchant system 808 may include POS devices and / or POS systems.

[0106] The acquirer system 810 may include one or more devices capable of receiving information from and / or transmitting information to the transaction service provider system 802, the issuer system 804, the client device 806, and / or the merchant system 808 via a communication network 812. For example, the acquirer system 810 may include a computing device, a server, a group of servers, etc. In some non-limiting embodiments or aspects, the acquirer system 810 may be associated with an acquirer described herein.

[0107] The communication network 812 may include one or more wired and / or wireless networks. For example, the communication network 812 may include a cellular network (e.g., a long term evolution (LTE) network, a third generation (3G) network, a fourth generation (4G) network, a fifth generation (5G) network, a code division multiple access (CDMA) network, etc.), a public land mobile network (PLMN), a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), a telephone network (e.g., a public switched telephone network (PSTN)), a private network (e.g., a private network associated with a transaction service provider), an ad hoc network, an intranet, the Internet, a fiber-based network, a cloud computing network, etc., and / or a combination of these or other types of networks.

[0108] In some non-limiting embodiments or aspects, processing a transaction may include generating and / or transmitting at least one transaction message (e.g., an authorization request, an authorization response, any combination thereof, etc.). For example, a client device (e.g., a client device 806, a POS device of a merchant system 808, etc.) may initiate a transaction, for example, by generating an authorization request. Additionally or alternatively, a client device (e.g., a client device 806, at least one device of a merchant system 808, etc.) may transmit an authorization request. For example, the client device 806 may transmit an authorization request to a merchant system 808 and / or a payment gateway (e.g., a payment gateway of a transaction service provider system 802, a third-party payment gateway separated from the transaction service provider system 802, etc.). Additionally or alternatively, the merchant system 808 (e.g., its POS device) may transmit an authorization request to an acquirer system 810 and / or a payment gateway. In some non-limiting embodiments or aspects, the acquirer system 810 and / or a payment gateway may transmit an authorization request to a transaction service provider system 802 and / or an issuer system 804. Additionally or alternatively, the transaction service provider system 802 may transmit the authorization request to the issuer system 804. In some non-limiting embodiments or aspects, the issuer system 804 may determine an authorization decision (e.g., authorization, rejection, etc.) based on the authorization request. For example, the authorization request may cause the issuer system 804 to determine the authorization decision based on the authorization request. In some non-limiting embodiments or aspects, the issuer system 804 may generate an authorization response based on the authorization decision. Additionally or alternatively, the issuer system 804 may transmit the authorization response. For example, the issuer system 804 may transmit the authorization response to the transaction service provider system 802 and / or the payment gateway. Additionally or alternatively, the transaction service provider system 802 and / or the payment gateway may transmit the authorization response to the acquirer system 810, the merchant system 808 and / or the client device 806. Additionally or alternatively, the acquirer system 810 may transmit the authorization response to the merchant system 808 and / or the payment gateway. Additionally or alternatively, the payment gateway may transmit the authorization response to the merchant system 808 and / or the client device 806. Additionally or alternatively, the merchant system 808 may transmit the authorization response to the client device 806. In some non-limiting embodiments or aspects, the merchant system 808 may receive the authorization response (e.g., from the acquirer system 810 and / or the payment gateway). Additionally or alternatively, the merchant system 808 may complete the transaction based on the authorization response (e.g., provide, ship, and / or deliver goods and / or services associated with the transaction; fulfill an order associated with the transaction; any combination thereof; etc.).

[0109] For purposes of illustration, processing a transaction may include generating a transaction message (e.g., an authorization request, etc.) based on an account identifier of a customer (e.g., associated with a customer device 806, etc.) and / or transaction data associated with the transaction. For example, a merchant system 808 (e.g., a client device of the merchant system 808, a POS device of the merchant system 808, etc.) may initiate a transaction, for example, by generating an authorization request (e.g., in response to receiving an account identifier from a portable financial device, etc. of the customer). Additionally or alternatively, the merchant system 808 may transmit the authorization request to an acquirer system 810. Additionally or alternatively, the acquirer system 810 may transmit the authorization request to a transaction service provider system 802. Additionally or alternatively, the transaction service provider system 802 may transmit the authorization request to an issuer system 804. The issuer system 804 may determine an authorization decision (e.g., authorization, rejection, etc.) based on the authorization request, and / or the issuer system 804 may generate an authorization response based on the authorization decision and / or the authorization request. Additionally or alternatively, the issuer system 804 may transmit the authorization response to the transaction service provider system 802. Additionally or alternatively, the transaction service provider system 802 may communicate the authorization response to the acquirer system 810 , which may communicate the authorization response to the merchant system 808 .

[0110] For purposes of illustration, clearing and / or settlement of a transaction may include generating a message (e.g., a clearing message, a settlement message, etc.) based on an account identifier of a customer (e.g., associated with a customer device 806, etc.) and / or transaction data associated with the transaction. For example, the merchant system 808 may generate at least one clearing message (e.g., a plurality of clearing messages, a batch of clearing messages, etc.). Additionally or alternatively, the merchant system 808 may transmit the clearing message to an acquirer system 810. Additionally or alternatively, the acquirer system 810 may transmit the clearing message to the transaction service provider system 802. Additionally or alternatively, the transaction service provider system 802 may transmit the clearing message to the issuer system 804. Additionally or alternatively, the issuer system 804 may generate at least one settlement message based on the clearing message. Additionally or alternatively, the issuer system 804 may transmit the settlement message and / or funds to the transaction service provider system 802 (and / or a settlement bank system associated with the transaction service provider system 802). Additionally or alternatively, the transaction service provider system 802 (and / or the settlement bank system) may transmit the settlement message and / or funds to the acquirer system 810, which may transmit the settlement message and / or funds to the merchant system 808 (and / or an account associated with the merchant system 808).

[0111] Provided as an example Figure 8The number and arrangement of systems, devices, and / or networks shown in FIG. Additional systems, devices, and / or networks, fewer systems, devices, and / or networks, different systems, devices, and / or networks, and / or in a similar manner may exist. Figure 8 The systems, devices and / or networks shown in the drawings are arranged in different ways. In addition, the systems, devices and / or networks can be implemented in a single system and / or device. Figure 8 two or more of the systems or devices shown in, or Figure 8 A single system or device shown in the environment 800 may be implemented as multiple distributed systems or devices. Additionally or alternatively, a group of systems (e.g., one or more systems) and / or a group of devices (e.g., one or more devices) of the environment 800 may perform one or more functions described as being performed by another group of systems or another group of devices of the environment 800.

[0112] Reference now Fig. 9 , a diagram showing example components of a device 900 according to a non-limiting embodiment or aspect. For example, the device 900 may correspond to Figure 1 In some non-limiting embodiments or aspects, such a system or device may include at least one device 900 and / or at least one component of device 900. The number and arrangement of the components shown are provided as examples. In some non-limiting embodiments or aspects, Fig. 9 Compared to the components shown in , device 900 may include additional components, fewer components, different components, or components arranged in a different manner. Additionally or alternatively, a group of components (e.g., one or more components) of device 900 may perform one or more functions described as being performed by another group of components of device 900.

[0113] like Fig. 9As shown in , the device 900 may include a bus 902, a processor 904, a memory 906, a storage component 908, an input component 910, an output component 912, and a communication interface 914. The bus 902 may include components that permit communication between components of the device 900. In some non-limiting embodiments or aspects, the processor 904 may be implemented in hardware, software, or a combination of hardware and software. For example, the processor 904 may include a processor (e.g., a central processing unit (CPU), a graphics processing unit (GPU), an accelerated processing unit (APU), etc.), a microprocessor, a digital signal processor (DSP), and / or any processing component that can be programmed to perform a function (e.g., a field programmable gate array (FPGA), an application-specific integrated circuit (ASIC), etc.). The memory 906 may include a random access memory (RAM), a read-only memory (ROM), and / or another type of dynamic or static storage device (e.g., a flash memory, a magnetic memory, an optical memory, etc.) storing information and / or instructions for use by the processor 904.

[0114] Continue to refer Fig. 9 , the storage component 908 may store information and / or software related to the operation and use of the device 900. For example, the storage component 908 may include a hard disk (e.g., a magnetic disk, an optical disk, a magneto-optical disk, a solid-state disk, etc.) and / or another type of computer-readable medium. The input component 910 may include a component that allows the device 900 to receive information, for example, through user input (e.g., a touch screen display, a keyboard, a keypad, a mouse, a button, a switch, a microphone, etc.). In addition or alternatively, the input component 910 may include a sensor for sensing information (e.g., a global positioning system (GPS) component, an accelerometer, a gyroscope, an actuator, etc.). The output component 912 may include a component that provides output information from the device 900 (e.g., a display, a speaker, one or more light-emitting diodes (LEDs), etc.). The communication interface 914 may include a transceiver-type component (e.g., a transceiver, a separate receiver and a transmitter, etc.) that enables the device 900 to communicate with other devices, for example, through a wired connection, a wireless connection, or a combination of wired and wireless connections. The communication interface 914 may allow the device 900 to receive information from another device and / or provide information to another device. For example, the communication interface 914 may include an Ethernet interface, an optical interface, a coaxial interface, an infrared interface, a radio frequency (RF) interface, a universal serial bus (USB) interface, interface, cellular network interface, and / or the like.

[0115] The device 900 can perform one or more processes described herein. The device 900 can perform these processes based on the processor 904 executing software instructions stored by computer-readable media such as memory 906 and / or storage component 908. The computer-readable medium may include any non-transient memory device. The memory device includes a memory space located inside a single physical storage device or a memory space spread over multiple physical storage devices. The software instructions can be read from another computer-readable medium or from another device to the memory 906 and / or storage component 908 through the communication interface 914. When executed, the software instructions stored in the memory 906 and / or storage component 908 can enable the processor 904 to perform one or more processes described herein. Additionally or alternatively, hard-wired circuits can be used in place of or in combination with software instructions to perform one or more processes described herein. Therefore, the embodiments or aspects described herein are not limited to any specific combination of hardware circuits and software. The term "programmed or configured" used herein refers to the arrangement of software, hardware circuit systems, or any combination thereof on one or more devices.

[0116] Although embodiments have been described in detail for purposes of illustration, it is to be understood that such details are for that purpose only and that the present disclosure is not limited to the disclosed embodiments or aspects, but on the contrary is intended to cover modifications and equivalent arrangements within the spirit and scope of the appended claims. For example, it is to be understood that the present invention contemplates that, to the extent possible, one or more features of any embodiment may be combined with one or more features of any other embodiment.

Claims

1. A computer-implemented method comprising: receiving, using at least one processor, real-time payment RTP data, the real-time payment data being associated with a plurality of RTP transactions, wherein the real-time payment RTP data of each RTP transaction in the plurality of RTP transactions comprises a plurality of attributes, the plurality of attributes comprising a sender and a receiver; selecting, using the at least one processor, one of the plurality of attributes as a target attribute, wherein the target attribute comprises a continuous-valued field; wherein the plurality of attributes other than the target attribute include a plurality of remaining attributes; generating a plurality of categories using the at least one processor; using the at least one processor, classifying each target attribute into one of the plurality of categories based on the value of the target attribute, wherein the plurality of categories each represent a range of values ​​for the continuous-valued field; training a first machine learning model using the at least one processor to perform a first task, the first task comprising predicting the target attribute based on inputting the plurality of remaining attributes of each of the plurality of RTP transactions into the first machine learning model, wherein the first machine learning model comprises at least one embedding layer and at least one hidden layer; After training the model, using the at least one processor, converting each of the plurality of remaining attributes into a first vector using the at least one embedding layer of the first machine learning model to form a first set of vectors; storing, using the at least one processor, the first set of vectors; as well as inputting the first set of vectors into a second machine learning model using the at least one processor to perform at least one second task different from the first task; wherein the first machine learning model comprises a natural language processing (NLP) model, and wherein the training further comprises inputting, using the at least one processor, each of the plurality of attributes of each RTP transaction as a word of a sentence into the natural language processing (NLP) model; Wherein training the first machine learning model to perform the first task further comprises: Using the at least one processor, creating a negative sample of the target attribute based on an equal probability of selecting any single target attribute classified into one of the plurality of categories, wherein creating the negative sample comprises: determining, using the at least one processor, a frequency with which the target attribute occurs in each of the plurality of categories; and Using the at least one processor, negative sampling of the target attribute is performed by selecting a false class for the target attribute based on the frequency with which the target attribute appears in each of the plurality of classes.

2. A computer-implemented method as described in claim 1, wherein the real-time payment RTP data of each RTP transaction in the multiple RTP transactions includes multiple fields, and each corresponding field in the multiple fields corresponds to a corresponding attribute in the multiple attributes.

3. The computer-implemented method of claim 1 , wherein the second machine learning model comprises a fraud detection model.

4. The computer-implemented method of claim 1 , further comprising: A graph comprising a plurality of nodes is generated based on the first set of vectors using at least one processor, wherein each node of the plurality of nodes represents an RTP transaction of the plurality of RTP transactions.

5. A system comprising at least one processor, the at least one processor being programmed or configured to: receiving real-time payment RTP data, the real-time payment data being associated with a plurality of RTP transactions, wherein the real-time payment RTP data of each RTP transaction in the plurality of RTP transactions comprises a plurality of attributes, the plurality of attributes comprising a sender and a receiver; selecting one of the plurality of attributes as a target attribute, wherein the target attribute comprises a continuous value field; wherein the plurality of attributes other than the target attribute comprises a plurality of remaining attributes; Generate multiple categories; classifying each target attribute into one of the plurality of categories based on the value of the target attribute, wherein the plurality of categories each represent a range of values ​​for the continuous-valued field; training a first machine learning model to perform a first task, the first task comprising predicting the target attribute based on inputting the plurality of remaining attributes of each of the plurality of RTP transactions into the first machine learning model, wherein the first machine learning model comprises at least one embedding layer and at least one hidden layer; After training the model, converting each of the plurality of remaining attributes into a first vector using the at least one embedding layer of the first machine learning model to form a first set of vectors; storing the first set of vectors; as well as Inputting the first set of vectors into a second machine learning model to perform at least one second task different from the first task; wherein the first machine learning model comprises a natural language processing (NLP) model, and wherein the training further comprises inputting each of the plurality of attributes of each RTP transaction into the natural language processing (NLP) model as a word of a sentence; Wherein training the first machine learning model to perform the first task further comprises: Based on an equal probability of selecting any single target attribute classified into one of the multiple categories, creating a negative sample of the target attribute, wherein creating the negative sample comprises: determining a frequency with which the target attribute occurs in each of the plurality of categories; and Based on the frequency at which the target attribute appears in each of the plurality of categories, a false category is selected for the target attribute, and negative sampling of the target attribute is performed.

6. The system of claim 5, wherein the real-time payment RTP data of each of the plurality of RTP transactions comprises a plurality of fields, each respective field of the plurality of fields corresponding to a respective attribute of the plurality of attributes.

7. The system of claim 5, wherein the second machine learning model comprises a fraud detection model.

8. The system of claim 5, wherein the at least one processor is further programmed or configured to: A graph including a plurality of nodes is generated based on the first set of vectors, wherein each node of the plurality of nodes represents an RTP transaction of the plurality of RTP transactions.

9. A computer program product comprising at least one non-transitory computer readable medium, the at least one non-transitory computer readable medium comprising one or more instructions which, when executed by at least one processor, cause the at least one processor to: receiving real-time payment RTP data, the real-time payment data being associated with a plurality of RTP transactions, wherein the real-time payment RTP data of each RTP transaction in the plurality of RTP transactions comprises a plurality of attributes, the plurality of attributes comprising a sender and a receiver; selecting one of the plurality of attributes as a target attribute, wherein the target attribute comprises a continuous value field; wherein the plurality of attributes other than the target attribute comprises a plurality of remaining attributes; Generate multiple categories; classifying each target attribute into one of the plurality of categories based on the value of the target attribute, wherein the plurality of categories each represent a range of values ​​for the continuous-valued field; training a first machine learning model to perform a first task, the first task comprising predicting the target attribute based on inputting the plurality of remaining attributes of each of the plurality of RTP transactions into the first machine learning model, wherein the first machine learning model comprises at least one embedding layer and at least one hidden layer; After training the model, converting each of the plurality of remaining attributes into a first vector using the at least one embedding layer of the first machine learning model to form a first set of vectors; storing the first set of vectors; as well as Inputting the first set of vectors into a second machine learning model to perform at least one second task different from the first task; wherein the first machine learning model comprises a natural language processing (NLP) model, and wherein the training further comprises inputting each of the plurality of attributes of each RTP transaction into the natural language processing (NLP) model as a word of a sentence; Wherein training the first machine learning model to perform the first task further comprises: Based on an equal probability of selecting any single target attribute classified into one of the multiple categories, creating a negative sample of the target attribute, wherein creating the negative sample comprises: determining a frequency with which the target attribute occurs in each of the plurality of categories; and Based on the frequency at which the target attribute appears in each of the plurality of categories, a false category is selected for the target attribute, and negative sampling of the target attribute is performed.

10. The computer program product of claim 9, wherein the real-time payment RTP data of each of the plurality of RTP transactions comprises a plurality of fields, each respective field of the plurality of fields corresponding to a respective attribute of the plurality of attributes.

11. The computer program product of claim 9, wherein the second machine learning model comprises a fraud detection model.

12. A computer program product as described in claim 9, wherein the one or more instructions that cause the at least one processor to train the first machine learning model to perform the first task also cause the at least one processor to: generate a graph comprising a plurality of nodes based on the first set of vectors, wherein each of the plurality of nodes represents an RTP transaction among the plurality of RTP transactions.

Images