Facilitating time zone prediction based on electronic communication data

Patent metadata

Field	Value
Publication number	US-12021812-B2
Application number	US-202117509885-A
Country	US
Kind code	B2
Filing date	Oct 25, 2021
Priority date	Oct 25, 2021
Publication date	Jun 25, 2024
Grant date	Jun 25, 2024

Abstract

Methods and systems are provided for facilitating time zone prediction using electronic communication data. Electronic message data associated with a message recipient of electronic communications is obtained. The electronic message data includes message delivery data associated with an electronic message and message response data associated with a response, by the message recipient, to a received electronic message. Using a machine learning model and based on the message delivery data and the message response data, a time-zone score is determined for a time zone. Such a time-zone score can indicate a probability the time zone corresponds with the message recipient. Based on the time-zone score, the time zone is identified as corresponding with the message recipient.

First claim

What is claimed is:

1. A computer-implemented method for predicting time zones, the method comprising:
obtaining, via a time zone estimation system, electronic message data associated with a message recipient of electronic communications, the electronic message data including message delivery data associated with a first electronic message delivered as a set of electronic messages to a group of recipients including the message recipient and message response data associated with a response, by the message recipient, to a second electronic message that was received by the message recipient;
determining, using a machine learning model of the time zone estimation system and based on the message delivery data and the message response data, a time-zone score for a time zone, the time-zone score indicating a probability the time zone corresponds with the message recipient based at least on a proportion of messages of the set of electronic messages sent to recipients of the group of recipients associated with the time zone;
based on the time-zone score, identifying, via the time zone estimation system, the time zone as corresponding with the message recipient;
using the time zone identified for the message recipient to schedule at least one subsequent electronic communication delivery; and
initiating one or more electronic communication deliveries to the message recipient, via a network, in accordance with the schedule.

2. The computer-implemented method of claim 1 , wherein the machine learning model comprises a hierarchical set of models including a collaborative-based model that uses the message delivery data associated with the first electronic message delivered as the set of electronic messages to the group of recipients, including the message recipient, to predict a first time-zone score for the time zone, an individual-based model that uses the message delivery data and the message response data to predict a second time-zone score for the time zone, and a prediction model that uses first time-zone score and the second time-zone score to predict the time-zone score for the time zone.

3. The computer-implemented method of claim 1 , further comprising:
generating a first set of feature vectors using the message delivery data;
inputting the first set of feature vectors into a collaborative-based model of the machine learning model to predict a first time-zone score for the time zone;
generating a second set of feature vectors using the message delivery data and the message response data;
inputting the second set of feature vectors into an individual-based model of the machine learning model; and
using output from the collaborative-based model and output from the individual-based model to determine the time-zone score for the time zone.

4. The computer-implemented method of claim 3 , wherein the first set of feature vectors comprises a message feature vector indicating the proportion of messages of the set of electronic messages sent to recipients of the group of recipients in association with a set of time zones and a delivery feature vector indicating percentages of message deliveries to the message recipient in association with the set of time zones.

5. The computer-implemented method of claim 4 , wherein the collaborative-based model is trained using a mean absolute error loss function, a root-mean-square error loss function, multi-class cross entropy, or a Kullback Leibler Divergence loss.

6. The computer-implemented method of claim 3 , wherein the second set of feature vectors comprises an unresponsive feature vector indicating a count of message deliveries without responses, a responsive feature vector indicating a count of message deliveries with responses, and a response time feature vector indicating a count of response times associated with message deliveries.

7. The computer-implemented method of claim 6 , wherein the individual-based model is trained using a cross-entropy loss function, a root-mean-square error loss function, a mean absolute error loss function, or a mean squared logarithmic error loss function.

8. The computer-implemented method of claim 1 further comprising providing the time zone for inclusion in a profile associated with the message recipient.

9. The computer-implemented method of claim 1 , wherein the machine learning model comprises a collaborative-based model that uses the message delivery data to predict a first time-zone score for the time zone and/or an individual-based model that uses the message delivery data and the message response data to predict a second time-zone score for the time zone.

10. One or more non-transitory computer-readable media having a plurality of executable instructions embodied thereon, which, when executed by one or more processors, cause the one or more processors to perform operations comprising:
determining, using a collaborative-based machine learning model, a first set of time-zone scores for a set of time zones in association with a recipient of electronic communications by processing a first set of data associated with the recipient, the first set of data including indications of one or more aggregate proportions of messages sent to groups of recipients associated with each time zone of the set of time zones, wherein an aggregate proportion associated with a particular time zone comprises an aggregate of a first proportion of messages sent to a first group of recipients in the particular time zone in association with a first message delivery and a second proportion of messages sent to a second group of recipients in the particular time zone in association with a second message delivery;
determining, using an individual-based machine learning model, a second set of time-zone scores for the set of time zones in association with the recipient by processing a second set of data associated with the recipient into, the second set of data including indications of response times for messages in association with the set of time zones;
normalizing the first set of time-zone scores and the second set of time-zone scores; and
determining, using a prediction machine learning model, a final set of time-zone scores for the time zones for the set of time zones in association with the recipient by processing the normalized first set of time-zone scores and the normalized second set of time-zone scores, each of the final time-zone scores indicating a probability that the corresponding time zone corresponds with the recipient
based on the final set of time-zone scores for the time zones, identifying a time zone for the recipient;
using the time zone identified for the recipient to schedule at least one subsequent electronic communication delivery; and
initiating one or more electronic communication deliveries to the recipient, via a network, in accordance with the schedule.

11. The media of claim 10 , wherein the first set of data further includes indications of percentages of message deliveries associated with the set of time zones.

12. The media of claim 11 , wherein the collaborative-based machine learning model comprises a linear model, a neural network, a gradient boosting decision tree, or a reinforcement learning model trained to optimize the first set of data using a mean absolute error loss function, a root-mean-square error loss function, or a mean squared logarithmic error loss function.

13. The media of claim 10 , wherein the second set of data further includes indications of delivery times of messages without responses in association with the set of time zones and indications of delivery times of messages with responses in association with the set of time zones.

14. The m

Assignees

• Adobe Inc

Inventors

• Yu Lijun
• Dai Wuyang
• He Jun
• Yang Hsiang-Yu
• Yan Zhenyu

Classifications

• G06F18/24147

  Distances to closest patterns, e.g. nearest neighbour classification · CPC title

• G06F18/217

  Validation; Performance evaluation; Active pattern learning techniques · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06F18/10

  Pre-processing; Data cleansing · CPC title

• G06N20/00

  Machine learning · CPC title