Systems and methods for predicting actionable tasks using contextual models

Patent metadata

Field	Value
Publication number	US-11263241-B2
Application number	US-201916569449-A
Country	US
Kind code	B2
Filing date	Sep 12, 2019
Priority date	Oct 19, 2018
Publication date	Mar 1, 2022
Grant date	Mar 1, 2022

Abstract

The present disclosure relates to an intelligent user interface that predicts tasks for users to complete using a trained machine-learning model. In some implementations, when a user accesses the intelligent user interface, the available tasks and a user profile can be inputted into the trained machine-learning model to output a prediction of one or more tasks for the user to complete. Advantageously, the trained machine-learning model outputs a prediction of tasks that the user will likely need to complete, based at least in part on the user's profile and previous interactions with applications.

First claim

What is claimed is:

1. A computer-implemented method, comprising:
collecting a plurality of log records stored at one or more servers, each log record of the plurality of log records representing one or more attributes of an interaction between a user device and an application, the application facilitating performance of one or more tasks, and the interaction being associated with a task previously selected by a user using the application;
generating contextual user data for each user of a plurality of users, the contextual user data for each user of the plurality of users being generated by aggregating the one or more attributes associated with the user across the plurality of log records;
performing, by a processor, a clustering operation on the contextual user data of the plurality of users using a machine-learning technique, the performance of the clustering operation causing a formation of one or more clusters of users, and each cluster of the one or more clusters representing one or more users that share a common attribute;
determining a set of tasks performable using one or more applications, each task of the set of tasks including one or more actions performable using an application of the one or more applications;
identifying a plurality of selection models, each selection model of the plurality of selection models indicating a protocol for selecting one or more tasks from the set of tasks, and each selection model of the plurality of selection models being associated with an accuracy; and
for each cluster of the one of more clusters:
selecting a selection model from the plurality of selection models, the selection being based on the accuracy of the selection model as compared to an accuracy of remaining selection models of the plurality of selection models;
evaluating, using the selected selection model, the contextual user data of the one or more users included in the cluster and the set of tasks; and
in response to the evaluation, executing the protocol associated with the selected selection model to select one or more tasks from the set of tasks, the selection of the one or more tasks corresponding to a recommendation of selectable tasks for presentation to the one or more users included in the cluster.

2. The computer-implemented method of claim 1 , wherein the plurality of selection models includes at least one multi-armed bandit machine learning algorithm.

3. The computer-implemented method of claim 1 , further comprising:
selecting one or more additional tasks from the set of tasks, each of the one or more additional tasks being selected as part of an exploration strategy to optimize the one or more tasks predicted to be taken by a particular user device, and wherein the one or more additional tasks are not determined based on a previous task performed in association with the particular user device.

4. The computer-implemented method of claim 1 , further comprising:
receiving an indication that a task of the one or more tasks was selected by a user;
in response to receiving the indication; generating a positive feedback signal indicating that the selected task was selected; and
updating the accuracy associated with the selection model used to select the selected task for presentation to the user, wherein the positive feedback signal causes the selection model to bias predicted tasks towards the selected task.

5. The computer-implemented method of claim 1 , further comprising:
determining that a task of the one or more tasks presented to the user was unselected;
generating a negative feedback signal indicating that the task was unselected; and
updating the accuracy associated with the selection model used to select the unselected task for presentation to the user, wherein the negative feedback signal causes the selection model to bias predicted tasks away from the unselected task.

6. The computer-implemented method of claim 1 , further comprising:
calculating the accuracy associated with each selection model of the plurality of selection models, the accuracy indicating a success rate of predicted tasks, and the accuracy being recalculated each instance the selection model is executed to select the one or more tasks for presenting to the user.

7. The computer-implemented method of claim 1 , further comprising:
displaying an interface that presents the one or more tasks selected using the selection model;
receiving input at the interface, the input corresponding to a selection of a presented task of the presented one or more tasks; and
in response to receiving the input, accessing the application associated with the selected task, the application associated with the selected task being configured to perform the selected task;
identifying, using the application, one or more actions associated with the selected task; and
automatically performing the one or more actions by triggering the application to perform the one or more actions in response to the received input.

8. A system, comprising:
one or more data processors; and
a non-transitory computer-readable storage medium containing instructions which, when executed on the one or more data processors, cause the one or more data processors to perform operations including:
collecting a plurality of log records stored at one or more servers, each log record of the plurality of log records representing one or more attributes of an interaction between a user device and an application, the application facilitating performance of one or more tasks, and the interaction being associated with a task previously selected by a user using the application;
generating contextual user data for each user of a plurality of users, the contextual user data for each user of the plurality of users being generated by aggregating the one or more attributes associated with the user across the plurality of log records;
performing a clustering operation on the contextual user data of the plurality of users using a machine-learning technique, the performance of the clustering operation causing a formation of one or more clusters of users, and each cluster of the one or more clusters representing one or more users that share a common attribute;
determining a set of tasks performable using one or more applications, each task of the set of tasks including one or more actions performable using an application of the one or more applications;
identifying a plurality of selection models, each selection model of the plurality of selection models indicating a protocol for selecting one or more tasks from the set of tasks, and each selection model of the plurality of selection models being associated with an accuracy; and
for each cluster of the one of more clusters:
selecting a selection model from the plurality of selection models, the selection being based on the accuracy of the selection model as compared to an accuracy of remaining selection models of the plurality of selection models;
evaluating, using the selected selection model, the contextual user data of the one or more users included in the cluster and the set of tasks; and
in response to the evaluation, executing the protocol associated with the selected selection model to select one or more tasks from the set of tasks, the selection of the one or more tasks corresponding to a recommendation of selectable tasks for presentation to the one or more users included in the cluster.

9. The system of claim 8 , wherein the plurality of selection models includes at least one multi-armed bandit machine learning algorithm.

10. The system of claim 8 , wherein the operations further comprise:
selecting one or more additional tasks from the set of tasks, each of the one or more additional tasks being

Assignees

• Oracle Int Corp

Inventors

• Rezaeian Amir Hossein
• Polleri Alberto
• Parkinson Stacy Paige
• Roberts Tara U

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N20/00 Primary

  Machine learning · CPC title

• G06N5/04

  Inference or reasoning models · CPC title

• G06F3/048

  Interaction techniques based on graphical user interfaces [GUI] · CPC title

• G06F16/2365

  Ensuring data consistency and integrity · CPC title