Methods and systems for using machine learning to determine times to send message notifications

1. A system for using machine learning to determine a time to send a message notification to a user device, the system comprising:

one or more processors configured to execute computer program instructions that, when executed, cause the one or more processors to perform operations comprising:

receiving a message comprising text and metadata indicating a sender of the message, the user device intended to receive the message, and a timestamp;

generating, via a message embedding model, a vector representation of the message, wherein the vector representation is indicative of the text and the metadata of the message;

inputting the vector representation into a sentiment detection model to obtain a sentiment identifier associated with the message;

inputting the vector representation into an urgency detection model to obtain an urgency level associated with the message;

inputting an indication of the sentiment identifier, the urgency level, the timestamp, and user device information into a response prediction model to obtain a predicted response time for the message indicative of a quantity of time predicted to transpire between a first time at which a notification is received at the user device and a second time at which a response to the message is predicted to be sent;

determining a time to send the message notification to the user device;

sending the notification to the user device;

receiving, from the user device, feedback information indicating a preferred time for receiving the message notification and a second sentiment identifier of the message indicating an interpretation of the message by a user; and

training, based on the feedback information, the sentiment detection model, the urgency detection model, and the response prediction model.

2. The system of claim 1 , wherein the determining a time to send the message notification to the user device further comprises:

determining, for a first time period, a first sentiment identifier predicted to correspond to a response from the user device, if the notification is sent during the first time period; and

based on a determination that the sentiment identifier matches a target sentiment identifier, determining the time to send the message notification based on the first time period.

3. The system of claim 1 , wherein the determining a time to send the message notification to the user device comprises:

comparing the predicted response time with a first threshold of a plurality of thresholds; and

based on a determination that the predicted response time fails to satisfy the first threshold, determining the time to send the message notification based on a user preference, wherein the user preference indicates a time at which non-urgent message notifications should be sent.

4. The system of claim 1 , wherein the user device information comprises schedule information associated with a user of the user device, the schedule information indicating a time when the user is available.

5. A method for using machine learning, the method comprising:

receiving a message comprising text and metadata indicating a sender of the message, a user device intended to receive the message, and a timestamp; and

generating, via one or more machine learning models, a sentiment identifier and an urgency level associated with the message, wherein the generating a sentiment identifier and an urgency level comprises:

generating a vector representation of the message, wherein the vector representation is indicative of the text and the metadata of the message;

inputting the vector representation into a sentiment detection model to obtain a sentiment identifier associated with the message; and

inputting the vector representation into an urgency detection model to obtain an urgency level associated with the message; and

inputting an indication of the sentiment identifier, the urgency level, the timestamp, and user device information into a response prediction model to obtain a predicted response time for the message.

6. The method of claim 5 , further comprising:

determining, based on the predicted response time, a time to provide a message notification to a user; and

providing a notification to the user.

7. The method of claim 5 , wherein the predicted response time for the message is indicative of a quantity of time predicted to transpire between a first time at which a notification is received at the user device and a second time at which a response is predicted to be sent to the message.

8. The method of claim 5 , further comprising:

receiving, from the user device, feedback information indicating a preferred time for receiving a message notification and a second sentiment identifier of the message indicating an interpretation of the message by a user; and

training, based on the feedback information, the one or more machine learning models.

9. The method of claim 6 , wherein the determining a time to send the message notification to the user device further comprises:

determining, for a first time period, a first sentiment identifier predicted to correspond to a response from the user device, if the notification is sent during the first time period; and

based on a determination that the sentiment identifier matches a target sentiment identifier, determining the time to send the message notification based on the first time period.

10. The method of claim 6 , wherein the determining a time to send the message notification to the user device comprises:

comparing the predicted response time with a first threshold of a plurality of thresholds; and

based on a determination that the predicted response time fails to satisfy the first threshold, determining the time to send the message notification based on a user preference, wherein the user preference indicates a time at which non-urgent message notifications should be sent.

11. The method of claim 5 , wherein the user device information comprises schedule information associated with a user of the user device, the schedule information indicating a time when the user is available.

12. The method of claim 5 , wherein the information indicating a preferred time comprises a timestamp indicating a time at which a response to the message was sent.

13. A non-transitory, machine-readable medium storing instructions that when executed by one or more processors effectuate operations comprising:

receiving a message comprising text and metadata indicating a sender of the message, a user intended to receive the message, and a timestamp; and

generating, via one or more machine learning models, a sentiment identifier and an urgency level associated with the message, wherein the generating a sentiment identifier and an urgency level comprises:

generating a vector representation of the message, wherein the vector representation is indicative of the text and the metadata of the message;

inputting the vector representation into a sentiment detection model to obtain a sentiment identifier associated with the message; and

inputting the vector representation into an urgency detection model to obtain an urgency level associated with the message; and

inputting an indication of the sentiment identifier, the urgency level, the timestamp, and user device information into a response prediction model to obtain a predicted response time for the message.

14. The non-transitory, machine-readable medium of claim 13 , wherein the instructions, when executed by the one or more processors, further effectuate operations comprising:

determining, based on the predicted response time, a time to provide a message notification to a user; and

providing a notification to the user.

15. The non-transitory, machine-readable medium of claim 13 , wherein the predicted response time for the message is indicative of a quantity of time predicted to transpire between a first time at which a notification is received at a user device and a second time at which a response is predicted to be sent to the message.

16. The non-transitory, machine-readable medium of claim 13 , wherein the instructions, when executed by the one or more processors, further effectuate operations comprising:

receiving, from a user device, feedback information indicating a preferred time for receiving a message notification and a second sentiment identifier of the message indicating an interpretation of the message by the user; and

training, based on the feedback information, the one or more machine learning models.

17. The non-transitory, machine-readable medium of claim 14 , wherein the instructions for determining a time to send the message notification to a user device, when executed by the one or more processors, further effectuate operations comprising:

determining, for a first time period, a first sentiment identifier predicted to correspond to a response from the user device, if the notification is sent during the first time period; and

based on a determination that the sentiment identifier matches a target sentiment identifier, determining the time to send the message notification based on the first time period.

18. The non-transitory, machine-readable medium of claim 14 , wherein the instructions for determining a time to send the message notification to a user device comprises, when executed by the one or more processors, further effectuate operations comprising:

comparing the predicted response time with a first threshold of a plurality of thresholds; and

based on a determination that the predicted response time fails to satisfy the first threshold, determining the time to send the message notification based on a user preference, wherein the user preference indicates a time at which non-urgent message notifications should be sent.

19. The non-transitory, machine-readable medium of claim 13 , wherein a user device information comprises schedule information associated with a user of a user device, the schedule information indicating a time when the user is available.

20. The non-transitory, machine-readable medium of claim 13 , wherein the information indicating a preferred time comprises a timestamp indicating a time at which a response to the message was sent.