Generating personas of users of networked services based on activities

Patent metadata

Field	Value
Publication number	US-11863643-B1
Application number	US-202318193891-A
Country	US
Kind code	B1
Filing date	Mar 31, 2023
Priority date	Mar 31, 2023
Publication date	Jan 2, 2024
Grant date	Jan 2, 2024

Abstract

Clusters of users of networked services are defined based on tasks performed by such users during such networked services. Activities of the users during sessions of the networked services are tracked, and representations of such users or such activities are used to train a model to predict activities of users in the future, including but not limited to services utilized by such users, or pages visited by such users. Subsequently, when a user accesses a networked service during a session, activities of the user may be determined, and a representation of the session is provided as an input to the model, along with contextual information such as an identifier of the persona of the user. A next action, e.g., a service or a page utilized by the user, may be predicted based on outputs received from the model.

First claim

What is claimed is:

1. A first computer system comprising:
one or more computer processors;
one or more memory components; and
one or more data stores,
wherein the first computer system is programmed with one or more sets of instructions that, when executed, cause the first computer system to perform a method comprising:
determining data regarding a plurality of user sessions, wherein the data identifies, for each of the plurality of user sessions, at least one of a plurality of networked services accessed by a user during one of the plurality of user sessions and a page visited by the user during the one of the plurality of user sessions;
generating sequences for each of the plurality of user sessions, wherein each of the sequences comprises:
a text-based identifier of a networked service and a page of the networked service;
a text-based identifier of another networked service and a page of the other networked service; and
a text-based token provided between the text-based identifiers;

providing each of the sequences to a model as inputs;
receiving outputs from the model, wherein each of the outputs comprises a representation of a user;
defining a set of clusters of a plurality of representations of the users, wherein each of the set of clusters comprises at least one representation of a user;
mapping each of the set of clusters to one of a set of personas, wherein each of the personas is defined to include at least one task unique to a persona and at least one task shared with another persona;
determining that a first user is associated with a first cluster of the set of clusters; and
selecting at least one of a networked service or a page for the first user based at least in part on a first persona to which the first cluster is mapped.

2. The first computer system of claim 1 , wherein mapping each of the set of clusters to one of the set of personas comprises:
determining probabilities that tasks of each of the set of personas of the set are included in one of the representations of the clusters; and
for each of the clusters,
identifying a persona having a greatest probability that a task of the persona is included in the cluster; and
mapping the cluster to the persona.

3. The first computer system of claim 1 , wherein mapping each of the plurality of clusters to one of the set of personas comprises:
determining probabilities that tasks of each of the set of personas of the set are included in one of the representations of the clusters; and
determining that a first probability that a task of a second persona of the set of personas is included in a first cluster is approximately equal to a second probability that a task of a third persona of the set of personas is included in the first cluster, wherein the first cluster is one of the plurality of clusters; and
in response to determining that the first probability is approximately equal to the second probability,
defining a fourth persona based at least in part on at least one of the tasks of the second persona and at least one of the tasks of the third persona;
adding the fourth persona to the set of personas; and
mapping the first cluster to the fourth persona.

4. The first computer system of claim 1 , wherein the model comprises one of:
a bidirectional encoder representations from transformers having a plurality of layers;
a principal component analysis; or
a singular value decomposition.

5. The first computer system of claim 1 , wherein each of the representations is an embedding comprising a continuous vector representing at least some of a sequence of activity of a user.

6. A method comprising:
providing data representing sequences of activities of a plurality of users of a networked service provider as a first set of inputs to a model, wherein each of the sequences of activities comprises a text-based identifier of at least one of a plurality of services provided by the networked service provider to one of the users or a page operated by the one of the users in accordance with one of the plurality of services, and wherein the model is trained to generate a representation of a user based at least in part on a sequence of activity of the user;
receiving a first set of outputs from the model in response to the first set of inputs;
determining representations of the plurality of users based at least in part on the first set of outputs;
generating a plurality of clusters of the representations;
identifying a plurality of tasks performed by the users of at least one of the plurality of services provided by the networked service provider;
defining a set of personas based at least in part on the plurality of tasks, wherein each of the plurality of personas is defined to include at least one task unique to a persona and at least one task shared with another persona;
mapping each of the plurality of clusters to one of the set of personas;
identifying a first sequence of activity of a first user of the networked service provider;
determining that the first user is associated with a first persona of the set of personas based at least in part on the first sequence of activity; and
in response to determining that the first user is associated with the first persona,
selecting at least one of a service provided by the networked service provider or a page associated with the service for the first user based at least in part on a task of the first persona.

7. The method of claim 6 , wherein the model comprises one of:
a bidirectional encoder representations from transformers having a plurality of layers;
a principal component analysis; or
a singular value decomposition.

8. The method of claim 6 , wherein the first sequence of activity comprises:
a first activity of the first user during a first user session;
a first page accessed by the first user during the first user session;
a second activity of the first user during the first user session; and
a second page accessed by the first user during the first user session.

9. The method of claim 8 , wherein determining that the first user is associated with the first persona of the set of personas based at least in part on the first sequence of activity comprises:
providing information regarding the first sequence of activity as a second set of inputs to the model, wherein the information regarding the first sequence of activity comprises:
a first text-based identifier of the first activity;
a second text-based identifier of the second activity; and
a text-based token provided between the first text-based identifier and the second text-based identifier;

receiving a second set of outputs from the model in response to the second set of inputs; and
determining a representation of the first user based at least in part on the second set of outputs,
wherein that the first user is associated with the first persona of the set of personas is determined based at least in part on the representation of the first user.

10. The method of claim 6 , wherein each of the representations is an embedding comprising a continuous vector representing at least some of a sequence of activity of a user.

11. The method of claim 6 , wherein mapping each of the plurality of clusters to one of the set of personas comprises:
determining probabilities that tasks of each of the set of personas of the set are included in one of the representations of the clusters; and
for each of the clusters,
identifying a persona having a greatest probability that a task of the persona is included in the cluster; and
mapping the cluster to the persona.

12. The method of claim 6 , wherein mapping each of the plurality of clusters to one of the s

Assignees

• Amazon Tech Inc

Inventors

• Seyeditabari Narjessadat
• Bannihatti Kumar Vinayshekhar
• Gangadharaiah Rashmi
• Nadig Deepak Seetharam
• Kapoor Ankit
• Luo Fayun

Classifications

• H04L67/535 Primary

  Tracking the activity of the user (network monitoring arrangements H04L43/00; recording of computer activity G06F11/34) · CPC title

• G06F40/284

  Lexical analysis, e.g. tokenisation or collocates · CPC title

• H04L67/306 Primary

  User profiles · CPC title