Theme recommendation engine

A system comprising:

• a processor;
• and
• a theme recommendation engine configured to store a set of instructions that, when executed by the processor, cause the system to:
• identify, by the theme recommendation engine, at least a first plurality of images in a first theme comprising two or more images, wherein the first theme is being accessed from a library of themes by a computing device, the first theme being one of a plurality of themes accessible from the library of themes, each of the plurality of themes configured to present at least one of a product and service, wherein the first theme presents at least one image of the first plurality of images with a first layout characteristic;
• determine a first plurality of vector representations of the first plurality of images using an image processing model, wherein the image processing model has been trained at least in part on images used in relation to themes and comprises a layered matrix of nodes wherein nodes and layers of nodes are interconnected to other layers and nodes and wherein the interconnections are assigned a weight;
• identify, by the theme recommendation engine, at least a second plurality of images in a second theme comprising two or more images, the second theme being one of the plurality of themes accessible from the library of themes, wherein the second theme presents at least one image of the second plurality of images with a second layout characteristic that is different from the first layout characteristic;
• determine a second plurality of vector representations of the second plurality of images using the image processing model;
• compute a first vector difference between the first plurality of vector representations and the second plurality of vector representations;
• determine a recommendation based on the computed first vector difference;
• communicate the recommendation to the computing device;
• receive feedback from the computing device that indicates whether the recommendation was accepted;
• and
• map the feedback to the weightings of the interconnections to improve a future recommendation.

A non-transitory computer-readable storage medium storing a set of instructions that, when executed by a processor of a computing device, cause the computing device to:

• identify, by a theme recommendation engine, at least a first plurality of images in a first theme comprising two or more images, wherein the first theme is being accessed from a library of themes by the computing device, the first theme being one of a plurality of themes accessible from the library of themes, each of the plurality of themes configured to present at least one of a product and service, wherein the first theme presents at least one image of the first plurality of images with a first layout characteristic;
• determine a first plurality of vector representations of the first plurality of images using an image processing model, wherein the image processing model has been trained at least in part on images used in relation to themes and comprises a layered matrix of nodes wherein nodes and layers of nodes are interconnected to other layers and nodes and wherein the interconnections are assigned a weight;
• identify, by the theme recommendation engine, at least a second plurality of images in a second theme comprising two or more images, the second theme being one of the plurality of themes accessible from the library of themes, wherein the second theme presents at least one image of the second plurality of images with a second layout characteristic that is different from the first layout characteristic;
• determine a second plurality of vector representations of the second plurality of images using the image processing model;
• compute a first vector difference between the first plurality of vector representations and the second plurality of vector representations;
• determine a recommendation based on the computed first vector difference;
• communicate the recommendation to the computing device;
• receive feedback from the computing device that indicates whether the recommendation was accepted;
• and
• map the feedback to the weightings of the interconnections to improve a future recommendation.