Utilizing a genetic algorithm in applying objective functions to determine distribution times for electronic communications

Patent metadata

Field	Value
Publication number	US-11645542-B2
Application number	US-201916384558-A
Country	US
Kind code	B2
Filing date	Apr 15, 2019
Priority date	Apr 15, 2019
Publication date	May 9, 2023
Grant date	May 9, 2023

Abstract

The present disclosure relates to systems, methods, and non-transitory computer readable media for generating a target distribution schedule for providing electronic communications based on predicted behavior rates by utilizing a genetic algorithm and one or more objective functions. For example, the disclosed systems can generate predicted behavior rates by training and utilizing one or more behavior prediction models. Based on the predicted behavior rates, the disclosed systems can further utilize a genetic algorithm to apply objective functions to generate one or more candidate distribution schedules. In accordance with the genetic algorithm, the disclosed systems can select a target distribution schedule for a particular user/client device. The disclosed systems can thus provide one or more electronic communications to individual users based on respective target distribution schedules.

First claim

What is claimed is:

1. In a digital medium environment for distributing electronic communications across computer networks, a computer-implemented method of determining electronic message send times for individual recipient devices, the computer-implemented method comprising:
determining, by at least one processor for a user over a target time horizon and based on previously distributed electronic communications, a predicted open rate using an open behavior machine learning model trained to predict open rates, a predicted click rate using a click behavior machine learning model trained to predict click rates, and a predicted fatigue rate using a fatigue behavior machine learning model trained to predict fatigue rates;
utilizing a genetic algorithm by the at least one processor to generate, based on the predicted open rate, the predicted click rate, and the predicted fatigue rate, a distribution schedule for providing electronic communications to the user by:
determining, by the at least one processor utilizing an objective function in relation to the predicted open rate, the predicted click rate, and the predicted fatigue rate, a set of candidate distribution schedules;
generating, by the at least one processor, a modified set of candidate distribution schedules by adding at least one of a crossover distribution schedule or a mutation distribution schedule to the set of candidate distribution schedules; and
selecting, by the at least one processor, a target distribution schedule from the modified set of candidate distribution schedules based on the objective function; and

providing an electronic communication for display within a user interface on a client device associated with the user based on the target distribution schedule.

2. The computer-implemented method of claim 1 , further comprising receiving input from an administrator device to set the target time horizon and the objective function utilized by the genetic algorithm.

3. The computer-implemented method of claim 1 , wherein the objective function utilized by the genetic algorithm comprises at least one of: a rate maximization objective function, an open rate decay objective function, or a threshold interaction probability objective function.

4. The computer-implemented method of claim 1 , wherein the open behavior machine learning model, the click behavior machine learning model, and the fatigue behavior machine learning model are each trained based on historical user behavior in relation to the previously distributed electronic communications.

5. The computer-implemented method of claim 1 , wherein the distribution schedule comprises a plurality of individual distribution times for providing the electronic communication and further comprises, for each of the plurality of individual distribution times, an indication of whether to provide the electronic communication.

6. A non-transitory computer readable medium comprising instructions that, when executed by at least one processor, cause a computer device to:
determine, by the at least one processor for a user over a target time horizon and based on previously distributed electronic communications, a predicted open rate using an open behavior machine learning model trained to predict open rates, a predicted click rate using a click behavior machine learning model trained to predict click rates, and a predicted fatigue rate using a fatigue behavior machine learning model trained to predict fatigue rates;
utilize a genetic algorithm by the at least one processor to generate, based on the predicted open rate, the predicted click rate, and the predicted fatigue rate, a distribution schedule for providing electronic communications to the user by:
determining, by the at least one processor utilizing an objective function in relation to the predicted open rate, the predicted click rate, and the predicted fatigue rate, a set of candidate distribution schedules;
generating, by the at least one processor, a modified set of candidate distribution schedules by adding at least one of a crossover distribution schedule or a mutation distribution schedule to the set of candidate distribution schedules; and
selecting, by the at least one processor, a target distribution schedule from the modified set of candidate distribution schedules based on the objective function; and

provide an electronic communication for display within a user interface on a client device associated with the user based on the target distribution schedule.

7. The non-transitory computer readable medium of claim 6 , further comprising instructions that, when executed by the at least one processor, cause the computer device to generate the crossover distribution schedule by combining two candidate distribution schedules from the set of candidate distribution schedules.

8. The non-transitory computer readable medium of claim 6 , further comprising instructions that, when executed by the at least one processor, cause the computer device to generate the mutation distribution schedule by modifying a candidate distribution schedule from the set of candidate distribution schedules.

9. The non-transitory computer readable medium of claim 6 , further comprising instructions that, when executed by the at least one processor, cause the computer device to utilize a fitness function comprising the objective function to select the set of candidate distribution schedules from among a plurality of candidate distribution schedules in accordance with one or more distribution parameters.

10. The non-transitory computer readable medium of claim 6 , further comprising instructions that, when executed by the at least one processor, cause the computer device to select the target distribution schedule by repeatedly applying the genetic algorithm to generate modified sets of candidate distribution schedules until one or more stop criteria are satisfied.

11. The non-transitory computer readable medium of claim 6 , wherein the objective function comprises one of a rate maximization objective function, an open rate decay objective function, or a threshold interaction probability objective function.

12. The non-transitory computer readable medium of claim 11 , further comprising instructions that, when executed by the at least one processor, cause the computer device to receive an input from an administrator device to select the objective function from among the rate maximization objective function, the open rate decay objective function, or the threshold interaction probability objective function.

13. The non-transitory computer readable medium of claim 11 , wherein:
the rate maximization objective function weights the predicted open rate and the predicted click rate to improve the weighted predicted open rate and the weighted predicted click rate over the target time horizon in accordance the predicted fatigue rate;
the open rate decay objective function applies an open rate decay to the weighted predicted open rate of the rate maximization objective function; and
the threshold interaction probability objective function increases a probability that the user will either open or click a distributed electronic communication a threshold number of times within the target time horizon.

14. A system comprising:
a memory device comprising an open behavior machine learning model, a click behavior machine learning model, a fatigue behavior machine learning model, a genetic algorithm, and an objective function; and
one or more processors coupled to the memory device, the one or more processors configured to:
determine, for a user over a target time horizon and based on previously distributed electronic

Assignees

• Adobe Inc

Inventors

• Zhang Lei
• He Jun
• Yan Zhenyu
• Dai Wuyang
• Pani Abhishek

Classifications

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N3/086 Primary

  using evolutionary algorithms, e.g. genetic algorithms or genetic programming · CPC title

• G06F17/18

  for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)} · CPC title