Dynamic content selection and optimization
AssigneeAmazon Technologies, Inc.
InventorsMichael Schmutz, Aigars Reiters
Life of this patent
Estimating the term…
Abstract
Various embodiments of a framework which allow dynamic testing of many creative content and other messages simultaneously using metrics-based optimization. A “multi-armed bandit” algorithmic approach employed, as an alternative to limited AB-type testing, to automatically select a set of content parameters based on the content parameters' respective probabilities, render the selected parameters to generate content sent to a user, and, after obtaining feedback in the form of user interaction data, update the parameters for future, iterative selection of content parameters. This framework can be used in essentially any setting to allow for the provision of feedback, including user interaction data.
Claims15
A computer-implemented method, comprising:
- obtaining creative content including a plurality of image and textual elements, the image and textual elements having a plurality of variable parameters;
- selecting initial permutations of the creative content, based at least in part on application of a random number;
- displaying, on a computing device display, the initial permutations of the creative content;
- generating a plurality of creative content permutations, corresponding to the creative content, using the plurality of image and textual elements and parameter settings from among the variable parameters, the creative content permutations having respective selection probabilities;
- automatically selecting display content, from the plurality of creative content permutations, based on a highest selection probability for a corresponding creative content permutation, the highest selection probability based at least in part on one or more multi-armed bandit algorithms processing the content permutations' respective selection probabilities;
- rendering the display content for display;
- obtaining user interaction data corresponding to the display content;
- recalculating at least one of the respective selection probabilities based at least in part upon the user interaction data;
- generating updated display content, based at least in part on the recalculating at least one of the respective selection probabilities;
- and
- displaying, on the computing device display, the display content.
A computer-implemented method, comprising:
- obtaining content including a plurality of elements, the elements having a plurality of variable parameters;
- selecting initial permutations of the content, based at least in part on application of a random number;
- displaying, on a computing device display, the initial permutations of the content;
- generating a plurality of content permutations, corresponding to the content, using parameter settings from among the variable parameters, the content permutations having respective selection probabilities;
- selecting display content, from the plurality of content permutations, based on the selection probability for a corresponding content permutation, the selection probability for the corresponding content permutation determined at least in part on one or more multi-armed bandit algorithms processing the content permutations' respective selection probabilities;
- rendering the display content for display;
- and
- displaying, on the computing device display, the display content.
A computing system, comprising:
- at least one processor;
- and
- memory including instructions that, when executed by the at least one processor, cause the computing system to:
- obtain content including a plurality of elements, the elements having a plurality of variable parameters;
- selecting initial permutations of the content, based at least in part on application of a random number;
- displaying, on a computing device display, the initial permutations of the content;
- generate a plurality of content permutations, corresponding to the content, using parameter settings from among the variable parameters, the content permutations having respective selection probabilities;
- select display content, from the plurality of content permutations, based on the selection probability for a corresponding content permutation, the selection probability for the corresponding content permutation determined at least in part on one or more multi-armed bandit algorithms processing the content permutations' respective selection probabilities;
- render the display content for display;
- and
- displaying, on the computing device display, the display content.
Dependencies are inferred from the claim text.