Optimizing Multi-Time Notifications Using Q-Learning

Abstract

In this research, we propose time optimization for notifications to assist users in remembering their actions by taking into consideration the amount of time it takes for them to respond and react. Using the Q-Learning algorithm, this proposal calculates when the best time to send notifications to users' smartphones in order to remind them of something important. The time at which the message is sent will be adjusted depending on the replies of prior users, which may be transformed into feedback at any time that is convenient. Notifications will be sent out, either repetitively or not, depending on the appropriate time for each individual, with the goal of ensuring that users do not forget about activities that they have planned. The results of testing our technique using the dataset show that it may be used to improve the time at which notifications are issued to recipients. It is possible to experiment with a variety of different times for the delivery of alerts in order to determine which of these periods is most successful for prompting users to take action. As a consequence of this, the algorithm is able to accommodate specific characteristics of individuals and find solutions to problems using a variety of standard operating procedures. Our proposal has the potential to successfully maintain the notification execution time at the intended level, which will prevent users from becoming concerned about the volume of notifications. Users who do not see the notification initially have the opportunity to do so at a later time step, which guarantees that activity data will still be collected.