Systems and methods for real-time analytics detection for a transaction utilizing synchronously updated statistical aggregation data

Patent metadata

Field	Value
Publication number	US-2020409932-A1
Application number	US-201916509593-A
Country	US
Kind code	A1
Filing date	Jul 12, 2019
Priority date	Jun 26, 2019
Publication date	Dec 31, 2020
Grant date	—

Abstract

A system and method for performing real-time analytics detection where a current profile update is derived from a current incoming transaction for an entity and stored in a profile update table, an up-to-date profile for the entity is derived from the most recent profile stored in a profile table and from all recent profile updates stored in the profile update table for the entity which are more recent than the last time the most recent profile was last updated, the up-to-date profile is utilized to perform the detection, and, the current profile is updated in a consistent manner, without utilizing an update lock on the profile table, by only updating the profile in the profile table based on the recent profile updates which are earlier than a current time minus a synchronization period which is larger than a maximum potential time difference between different clocks in the system.

First claim

1 . A real-time analytics detection system comprising:
a database; a memory; and one or more processors configured to, in response to receiving a current incoming transaction:
retrieve, from a profile table of the database, a most recent profile aggregation data for an entity associated with the current incoming transaction, and
derive a current profile update from the current incoming transaction,
store the current profile update in a profile update table of the database, wherein the profile update includes:
at least one update amount,
a timestamp, and
merging instructions describing how to update the most recent profile aggregation data based on the at least one update amount,

retrieve, from the profile update table of the database all recent profile updates corresponding to the entity, wherein each of the recent profile updates have a timestamp which is more recent than the last time the most recent profile aggregation data was updated,
derive a first up-to-date profile aggregation data from the most recent profile aggregation data and the retrieved recent profile updates, and
store the first up-to-date profile aggregation data in the memory.

2 . The real-time analytics detection system of claim 1 , wherein the one or more processors are configured to update the most recent profile aggregation data in the profile table to be the first up-to-date profile aggregation data asynchronously to the one or more processors performing real-time analytics detection.

3 . The real-time analytics detection system of claim 2 , wherein one or more processors are configured to update the most recent profile aggregation data in a consistent manner without utilizing an update lock on the profile table.

4 . The real-time analytics detection system of claim 3 , wherein the one or more processors are configured to, after performing the real-time analytics detection:
transmit a current profile re-calculation trigger to a message queue, wherein:
the current profile re-calculation trigger is associated with the current profile update,
the re-calculation trigger comprises a primary key which is associated with the entity and which was derived from a previous transaction associated with the entity.

5 . The real-time analytics detection system of claim 4 , wherein the one or more processors are configured to, after performing the real-time analytics detection:
extract a next primary key from a next re-calculation trigger; retrieve a profile row from the profile table which comprises profile aggregation data and the next primary key; retrieve at least one profile update row from the profile update table, wherein each of the at least one profile update rows comprise:
the primary key,
a recent profile update, and
a timestamp which is:
more recent than the last time the most recent profile aggregation data was last updated, and
earlier than a current time in a clock of the database minus a predetermined synchronization period; and

derive a second up-to-date profile aggregation data from the retrieved profile row and the retrieved at least one profile update rows.

6 . The real-time analytics detection system of claim 5 , wherein the one or more processors are configured to update the profile aggregation data in the profile row in the profile table to be the second up-to-date profile aggregation data if a most recent timestamp of the at least one profile update row is more recent than the last time the most recent profile aggregation data was last updated.

7 . The real-time analytics detection system of claim 6 , wherein the profile row further comprises a last update timestamp which indicates when the profile row was last updated, and the one or more processors are configured to update the last update timestamp of the profile row to be the most recent timestamp of the at least one profile update row after updating the profile row in the profile table.

8 . The real-time analytics detection system of claim 5 , wherein the predetermined synchronization period is larger than a maximum potential time difference between different clocks in the real-time analytics detection system and is in an order of milliseconds.

9 . The real-time analytics detection system of claim 1 , wherein the most recent profile aggregation data comprises statistical aggregators derived from previously received transactions associated with the entity.

10 . The real-time analytics detection system of claim 1 , wherein the one or more processors are configured to perform real-time analytics detection on the current incoming transaction by utilizing the first up-to-date profile aggregation data to determine to determine if the current incoming transaction is valid, should be blocked, or if the one or more processors should generate an alert.

11 . A method for performing real-time analytics detection, the method comprising:
in response to receiving a current incoming transaction:
deriving, using the one or more processors, a current profile update from the current incoming transaction;
storing the current profile update in a profile update table of a database, wherein the profile update includes:
at least one update amount,
a timestamp, and
merging instructions describing how to update the most recent profile aggregation data based on the at least one update amount;

retrieving a most recent profile aggregation data for an entity associated with the current incoming transaction from a profile table of a database;
retrieving all recent profile updates corresponding to the entity from the profile update table, wherein each recent profile update has a timestamp which is more recent than the last time the most recent profile aggregation data was updated; and
deriving, using the one or more processors, a first up-to-date profile aggregation data from the most recent profile aggregation data and the retrieved recent profile updates; and
storing the first up-to-date profile aggregation data in a memory of the one or more processors.

12 . The method of claim 11 , further comprising updating, using the one or more processors, the most recent profile aggregation data in the profile table, in a consistent manner without utilizing an update lock on the profile table, asynchronously to the one or more processors performing real-time analytics detection.

13 . The method of claim 12 , further comprising, after performing the real-time analytics detection, the one or more processors transmitting a current profile re-calculation trigger to a message queue, wherein:
the current profile re-calculation trigger is associated with the current profile update, and the re-calculation trigger comprises a primary key which is associated with the entity and which was derived from a previous transaction associated with the entity

14 . The method of claim 13 , further comprising, after performing the real-time analytics detection:
extracting, using the one or more processors, a next primary key form a next re-calculation trigger; retrieving a profile row from the profile table which comprises profile aggregation data and the next primary key; retrieving at least one profile update row from the profile update table, wherein each of the at least one profile update rows comprise: the primary key,
a recent profile update, and
a timestamp which is:
more recent than the last time the most recent profile aggregation data was last updated, and
earlier than a current time in a clock of the database minus a predetermined synchronization period; and

deriving, using the one or more processors, a second up-to-date profile aggregation data from the retrieved profi

Assignees

• Nice Ltd

Inventors

• Weizman Shlomi

Classifications

• G06F16/288

  Entity relationship models · CPC title

• G06F16/2343 Primary

  Locking methods, e.g. distributed locking or locking implementation details · CPC title

• G06F16/27

  Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor · CPC title

• G06F16/2365 Primary

  Ensuring data consistency and integrity · CPC title

• G06F9/466

  Transaction processing · CPC title