Back to results
US 12,217,298 B2Granted
Systems and methods for integration of calendar applications with task facilitation services
AssigneeYohana LLC
InventorsYoky Matsuoka, Nitin Viswanathan
Life of this patent
Estimating the term…
Abstract
Integration of an external calendar application with a task facilitation service includes mechanisms for creating tasks within the task facilitation service based on calendar data of the calendar application received by the task facilitation service and processed using various dynamic models and algorithms. Further examples of integration include the task facilitation service generating recommendations for new calendar items and modifications to existing calendar items by leveraging the data and models available to the task facilitation service.
Claims21
A computer-implemented method comprising:
- receiving calendar data for a particular user of a task facilitation service through an external application programming interface (API), wherein the calendar data is associated with a calendar of a calendar application;
- accessing a user model corresponding to the particular user, wherein the user model is updated based on historic activity of the particular user;
- processing the calendar data and the user model using a natural-language processing (NLP) model to generate a task recommendation, wherein the task recommendation indicates one or more recommended tasks for delegation by the particular user, wherein the NLP model was initially trained with a training dataset using unsupervised training and without user supervision, and wherein the training dataset includes training data associated with other users;
- transmitting an indication corresponding to the task recommendation, wherein, when the indication is received by a computing device, the computing device is enabled to approve the task recommendation;
- receiving an approval to proceed with performing the one or more recommended tasks;
- accessing task-execution data associated with the one or more recommended tasks, wherein the task-execution data identifies performance statuses associated with the one or more recommended tasks;
- and
- updating the NLP model based on the task-execution data, wherein updating includes adjusting one or more weights of the NLP model using the unsupervised training and without user supervision, and wherein the one or more weights of the NLP model are adjusted until a corresponding logarithmic loss exceeds a predetermined threshold.
5 more elements
A system comprising:
- one or more processors;
- and
- a non-transitory computer-readable storage medium that stores instructions that, when executed by the one or more processors, cause the one or more processors to perform operations of:
- receiving calendar data for a particular user of a task facilitation service through an external application programming interface (API), wherein the calendar data is associated with a calendar of a calendar application;
- accessing a user model corresponding to the particular user, wherein the user model is updated based on historic activity of the particular user;
- processing the calendar data and the user model using a natural-language processing (NLP) model to generate a task recommendation, wherein the task recommendation indicates one or more recommended tasks for delegation by the particular user, wherein the NLP model was initially trained with a training dataset using unsupervised training and without user supervision, and wherein the training dataset includes training data associated with other users;
- transmitting an indication corresponding to the task recommendation, wherein, when the indication is received by a computing device, the computing device is enabled to approve the task recommendation;
- receiving an approval to proceed with performing the one or more recommended tasks;
- accessing task-execution data associated with the one or more recommended tasks, wherein the task-execution data identifies performance statuses associated with the one or more recommended tasks;
- and
- updating the NLP model based on the task-execution data, wherein updating includes adjusting one or more weights of the NLP model using the unsupervised training and without user supervision, and wherein the one or more weights of the NLP model are adjusted until a corresponding logarithmic loss exceeds a predetermined threshold.
8 more elements
A non-transitory computer-readable storage medium that stores instructions that, when executed by one or more processors, cause the one or more processors to perform operations of:
- receiving calendar data for a particular user of a task facilitation service through an external application programming interface (API), wherein the calendar data is associated with a calendar of a calendar application;
- accessing a user model corresponding to the particular user, wherein the user model is updated based on historic activity of the particular user;
- processing the calendar data and the user model using a natural-language processing (NLP) model to generate a task recommendation, wherein the task recommendation indicates one or more recommended tasks for delegation by the particular user, wherein the NLP model was initially trained with a training dataset using unsupervised training and without user supervision, and wherein the training dataset includes training data associated with other users;
- transmitting an indication corresponding to the task recommendation, wherein, when the indication is received by a computing device, the computing device is enabled to approve the task recommendation;
- receiving an approval to proceed with performing the one or more recommended tasks;
- accessing task-execution data associated with the one or more recommended tasks, wherein the task-execution data identifies performance statuses associated with the one or more recommended tasks;
- and
- updating the NLP model based on the task-execution data, wherein updating includes adjusting one or more weights of the NLP model using the unsupervised training and without user supervision, and wherein the one or more weights of the NLP model are adjusted until a corresponding logarithmic loss exceeds a predetermined threshold.
5 more elements
Dependencies are inferred from the claim text.