# US 2020/0402058 A1 · Systems and methods for real-time processing of data streams

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2020/0402058 A1
- Document ID: US20200402058A1
- Status: Published application
- Kind: A1
- Application number: 16/447,904
- Filed: 2019-06-20
- Published: 2020-12-24
- Earliest claimed: 2019-06-20
- Assignees: Coupang Corp.
- Applicants: Coupang Corp.
- Inventors: Yan ZHOU; Yonghui CHEN
- Web page: https://patentgrep.com/patent/US20200402058A1

## CPC

- G06Q 20/385: {using an alias or single-use codes}
- G06Q 20/4016: {involving fraud or risk level assessment in transaction processing}
- G06F 9/54: Interprogram communication
- G06F 9/542: {Event management; Broadcasting; Multicasting; Notifications}
- G06F 16/24568: {Data stream processing; Continuous queries}
- G06F 17/15: Correlation function computation {including computation of convolution operations (arithmetic circuits for sum of products per se, e.g. multiply-accumulators G06F7/5443; digital filters, e.g. FIR, IIR, adaptive filters H03H17/00)}
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/047: Probabilistic or stochastic networks
- G06N 3/09: Supervised learning
- G06Q 20/3276: {using a pictured code, e.g. barcode or QR-code, being read by the M-device}
- G06Q 20/40145: {Biometric identity checks}
- G06N 3/045: Combinations of networks
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06N 20/20: Ensemble learning
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A system for generating alerts including processors and storage devices. The instructions configure the one or more processors to perform operations, which include receiving an event from a data stream, extracting keys from the event, associating the event with at least one account based on the extracted keys, identifying a state variable associated with the at least one account, updating the state variable by accumulating the event in the state variable, registering a time stamp for the event in the state variable, and retiring expired events from the state variable. The operations may also include determining whether the state variable is above a threshold level and generating an alert for the account when the state variable is above the threshold level.

## Claims

### Claim 1

A system for generating alerts, comprising:
one or more processors; and
one or more storage devices storing instructions that, when executed, configure the one or more processors to perform operations comprising:
receiving an event from a data stream, the event comprising transaction information;
extracting keys from the event based on a group of pre-defined selected topics;
associating the event with at least one account based on the extracted keys;
identifying a state variable associated with the at least one account, the state variable being configured to be accessed with a constant time complexity, the state variable being a single value variable in an array with a unique element;
updating the state variable by:
accumulating the event in the state variable by adding an influence of the event to a single value stored in the state variable, the influence of the event being proportional to a transaction amount associated with the event;
registering a time stamp for the event; and
retiring expired events from the state variable by subtracting, from the single value, influence of events outside an influence window;
determining whether the state variable is above a threshold level;
retrieving a plurality of state variables associated with the account from a non-relational database, the plurality of state variables being arrays with unique elements;
assigning weights to each state variable based on the extracted keys;
computing a probability of fraud using a predictive model; and
generating an alert signal when the probability of fraud is above a probability threshold.

### Claim 2 (depends on 1)

The system of claim 1, wherein associating the event with the at least one account comprises:
retrieving from a non-relational database a lookup table associating keys with accounts; and
identifying an account in the lookup table that is correlated with one or more of the extracted keys.

### Claim 3 (depends on 1)

The system of claim 1, wherein
the operations further comprise determining a length of time for the influence window; and
retiring expired events comprises:
retrieving a time stamp for existing events accumulated in the state variable; and
deleting existing events with a time stamp older than a lower bound of the influence window.

### Claim 4 (depends on 3)

The system of claim 3, wherein
a single copy of the state variable is stored in a single memory unit; and
the time window is of 24 hours.

### Claim 5

(canceled)

### Claim 6 (depends on 1)

The system of claim 1, wherein
the operations further comprise extracting values from the event;
the extracted keys include at least one of an IP address, an email address, or a phone number; and
the extracted values include a transaction value.

### Claim 7 (depends on 6)

The system of claim 6, wherein accumulating the event in the state variable further comprises:
assigning a score to the event proportional to the transaction value; and
increasing the value of the state variable based on the assigned score.

### Claim 8 (depends on 1)

The system of claim 1, wherein the operations further comprise:
receiving a transaction request from an API interfacing client systems;
identifying an account associated with the transaction request;
declining the transaction request when an alert has been generated for the account.

### Claim 9 (depends on 1)

The system of claim 1, wherein
receiving an event from a data stream comprises receiving a plurality of events from multiple client systems; and
extracting keys from the event comprises:
configuring one or more stream capture applications;
assigning one or more brokers to each capture application; and
defining in each broker to convert event information into variables accessible with O(1) operators.

### Claim 10 (depends on 1)

The system of claim 1, wherein receiving the event comprises:
identifying events that do not include an associated event time;
deleting events without associated event time; and
normalizing event information of undeleted events by normalizing time zone of event times and currency of the events.

### Claim 11

A computer-implemented method for generating alerts, the method comprising:
receiving an event from a data stream, the event comprising transaction information;
extracting keys from the event based on a group of pre-defined selected topics;
associating the event with at least one account based on the extracted keys;
identifying a state variable associated with the at least one account, the state variable being configured to be accessed with a constant time complexity, the state variable being a single value variable in an array with a unique element;
updating the state variable by:
accumulating the event in the state variable by adding an influence of the event to a single value stored in the state variable, the influence of the event being proportional to a transaction amount associated with the event;
registering a time stamp for the event; and
retiring expired events from the state variable by subtracting, from the single value, influence of events outside an influence window;
determining whether the state variable is above a threshold level;
generating an alert for the account when the state variable is above the threshold level;
retrieving a plurality of state variables associated with the account from a non-relational database, the plurality of state variables being arrays with unique elements;
assigning weights to each state variable based on the extracted keys;
computing a probability of fraud using a predictive model; and
generating an alert signal when the probability of fraud is above a probability threshold.

### Claim 12 (depends on 11)

The method of claim 11, wherein associating the event with the at least one account comprises:
retrieving from a non-relational database a lookup table associating keys with accounts; and
identifying an account in the lookup table that is correlated with one or more of the extracted keys.

### Claim 13 (depends on 11)

The method of claim 11, further comprising determining a length of time for the influence window; and
wherein retiring expired events comprises:
retrieving a time stamp for existing events accumulated in the state variable; and
deleting existing events with a time stamp older than a lower bound of the influence window.

### Claim 14 (depends on 13)

The method of claim 13, wherein
a single copy of the state variable is stored in a single memory unit; and
the time window is of 24 hours.

### Claim 15

(canceled)

### Claim 16 (depends on 11)

The method of claim 11, further comprising:
extracting a transaction value associated with the event; and
wherein the extracted keys include at least one of an IP address, an email address, a phone number.

### Claim 17 (depends on 16)

The method of claim 16, wherein accumulating the event in the state variable further comprises:
assigning a score to the event proportional to the transaction value; and
increasing the value of the state variable based on the assigned score.

### Claim 18 (depends on 11)

The method of claim 11, further comprising:
receiving a transaction request from an API interfacing client systems;
identifying an account associated with the transaction request; and
declining the transaction request when an alert has been generated for the account.

### Claim 19 (depends on 11)

The method of claim 11, wherein
receiving an event from a data stream comprises receiving a plurality of events from multiple client systems; and
extracting keys from the event comprises:
configuring one or more stream capture applications;
assigning one or more brokers to each capture application; and
defining in each broker to convert event information into variables accessible with O(1) operators.

### Claim 20

A non-transitory computer-readable medium storing instructions that, when executed by a processor, perform operations to generate real-time alerts based on transactions comprising:
receiving, from a plurality of client systems, a data stream comprising a plurality of events;
filtering out events from the plurality of events that do not include a time stamp;
generating an array of normalized events by normalizing time zone and currency in each one of the non-filtered events;
extracting in real-time a plurality of keys from the normalized events, the plurality of keys comprising IP address, email address, or phone number;
associating the normalized events with at least one account based on corresponding keys;
identifying state variables associated with corresponding accounts, the state variables being configured be accessed with O(1) complexity operators, each of the state variables being single value variables in arrays with a unique position element;
updating the state variables by:
accumulating the non-filtered events in the state variables by adding influence of the events to single values respectively stored in the state variables, the influence of the events being proportional to transaction amounts associated with the events;
registering a timer and a callback for each one of the events; and
retiring expired events from the state variables by subtracting, from the single values, influence of events outside an influence window, the expired events having a time stamp that fall outside a time window;
retrieving a plurality of state variables associated with the account from a non-relational database, the plurality of state variables being arrays with unique elements;
assigning weights to updated state variables and the plurality of state variables based on the extracted keys:
computing a probability of fraud using a predictive model based on the updated variables and the plurality of state variables; and
generating an alert for the account when the probability of fraud is above the threshold level.
