Analysis platform for actionable insight into user interaction data

Patent metadata

Field	Value
Publication number	US-11610206-B2
Application number	US-202017070714-A
Country	US
Kind code	B2
Filing date	Oct 14, 2020
Priority date	Oct 14, 2020
Publication date	Mar 21, 2023
Grant date	Mar 21, 2023

Abstract

There are provided systems and methods for actionable insight into user interaction data. A service provider server can access user interaction data associated with an interaction between a first communication device and the service provider server, and generates feature representations of the user interaction data, in which the feature representations respectively correspond to extracted features that include textual data features or audio data features. The service provider server can determine an intent of the interaction from the feature representations using a machine learning-trained classifier, in which the intent corresponds to a first actionable insight category. The interaction is mapped to a first cluster based on the intent, and the service provider server issues a remedial action for the interaction based on the mapping of the interaction to the first cluster, in which the remedial action is associated with a particular type of activity in the first actionable insight category.

First claim

What is claimed is:

1. A method, comprising:
obtaining, by one or more hardware processors, user interaction data corresponding to a communication from a first communication device of a user to a service provider server, wherein the communication is regarding a user account with a service provider associated with the service provider server;
extracting, by the one or more hardware processors from the user interaction data, a plurality of features comprising one or more of textual data features or audio data features;
determining, by the one or more hardware processors, an intent of the communication using a machine learning-trained classifier based on the plurality of features;
grouping, by the one or more hardware processors, the communication with a first plurality of past communications that is associated with the intent;
assigning, by the one or more hardware processors, each of the first plurality of past communications and the communication to one or more clusters from a first plurality of clusters using one or more first clustering algorithms based on types of activity associated with the first plurality of past communications and the communication;
identifying, by the one or more hardware processors and from the first plurality of clusters, a first particular cluster to which the communication is assigned;
assigning, by the one or more hardware processors, each of a second plurality of past communications within the first cluster and the communication to one or more clusters from a second plurality of clusters using one or more second clustering algorithms based on activity patterns associated with the second plurality of past communications and the communication;
identifying, by the one or more hardware processors and from the second plurality of clusters, a second cluster to which the communication is assigned;
deriving, by the one or more hardware processors, one or more common attributes shared by past communications within the second cluster;
detecting, by the one or more hardware processors, an anomaly associated with the second cluster based on the one or more common attributes;
determining, by the one or more hardware processors, a risk level for the communication based on the anomaly; and
restricting, by the one or more hardware processors, the user from accessing one or more services of the service provider server through the user account based on the risk level.

2. The method of claim 1 , wherein the first communication device was authenticated to access the user account via a first authentication process, and wherein the method further comprises:
in response to detecting the anomaly, performing a second authentication process with the user via the first communication device for accessing the user account.

3. The method of claim 1 , wherein the detecting the anomaly comprises:
determining whether a unique pattern of activities in the second cluster corresponds to a malicious activity.

4. The method of claim 1 , further comprising:
determining one or more anomalies in the second plurality of clusters;
determining whether the one or more anomalies exceed a threshold number of anomalies;
generating a report indicating the one or more anomalies when the one or more anomalies exceed the threshold number of anomalies; and
transmitting the report to a second communication device.

5. The method of claim 1 , wherein the obtaining the user interaction data comprises:
obtaining textual data associated with a first type of interaction from a first data structure in a data repository communicably coupled to the service provider server;
obtaining audio data associated with a second type of interaction from a second data structure in the data repository; and
generating the user interaction data based on combining the textual data with the audio data,
wherein the textual data and the audio data correspond to different portions of the user interaction data.

6. The method of claim 1 , wherein the user interaction data comprises textual data and audio data, and wherein the extracting the plurality of features comprises:
selecting, from a plurality of feature extraction algorithms, a feature extraction algorithm based on a comparison of performance metrics associated with the plurality of feature extraction algorithms;
applying the selected feature extraction algorithm to the user interaction data;
extracting a plurality of textual data features from the textual data; and
extracting a plurality of audio data features from the audio data.

7. The method of claim 1 , further comprising:
selecting one of a plurality of machine learning-trained classifiers based on a comparison of performance metrics associated with the plurality of machine learning-trained classifiers,
wherein the intent of the communication is determined using the selected one of the plurality of machine learning-trained classifiers.

8. The method of claim 1 , further comprising:
generating a plurality of machine learning-based networks corresponding to a plurality of actionable insight categories.

9. The method of claim 8 , further comprising:
training each of the plurality of machine learning-based networks with a respective training dataset, the respective training dataset comprising labeled interaction data indicating what information pertains to which of the plurality of actionable insight categories; and
selecting, from the plurality of machine learning-based networks, a particular machine learning-based network to be the machine learning-trained classifier for determining the intent.

10. The method of claim 1 , further comprising:
accessing a plurality of user interaction datasets associated with respective ones of a plurality of interactions between a plurality of communication devices and the service provider server,
determining a corresponding intent of each of the plurality of interactions from extracted features associated with each of the plurality of interactions using the machine learning-trained classifier; and
classifying each of the plurality of interactions as a respective category of the plurality of actionable insight categories based at least in part on the corresponding intent of the interaction.

11. A system, comprising:
a non-transitory memory; and
one or more hardware processors coupled with the non-transitory memory and configured to read instructions from the non-transitory memory to cause the system to perform operations comprising:
obtaining user interaction data corresponding to a communication from a first communication device of a user to a service provider server, wherein the communication is regarding a user account with a service provider associated with the service provider server;
extracting, from the user interaction data, a plurality of features comprising one or more of textual data features or audio data features;
determining an intent of the communication using a machine learning-trained classifier based on the plurality of features;
grouping the communication with a first plurality of past communications that is associated with the intent;
assigning each of the first plurality of past communications and the communication to one or more clusters from a first plurality of clusters using one or more first clustering algorithms based on types of activity associated with the first plurality of past communications and the communication;
identifying, from the first plurality of clusters, a first cluster to which the communication is assigned;
assigning each of a second plurality of past communications within the first cluster and the communication to one or more clusters from a second plurality of clusters using one or more second cluster

Assignees

• Paypal Inc

Inventors

• Nunes Eric
• Yadav Meethil Vijay
• Johnson Raoul

Classifications

• H04L63/08 Primary

  for authentication of entities (cryptographic mechanisms or cryptographic arrangements for entity authentication H04L9/32) · CPC title

• G06Q20/4016 Primary

  involving fraud or risk level assessment in transaction processing · CPC title

• G06N20/00

  Machine learning · CPC title

• H04L63/102

  Entity profiles · CPC title

• G06Q20/386

  using messaging services or messaging apps (using mobile network messaging services for payment, e.g. SMS G06Q20/3255) · CPC title