US 12,169,850 B1
Enhancing content and layout control with generative systems
Timur Luguev, Toronto (CA); Zakaria Patel, Hamilton (CA); Hantang Li, Toronto (CA); and Max Sinclair, Toronto (CA)
Assigned to ECOMTENT INC., Toronto (CA)
Filed by Ecomtent Inc., Toronto (CA)
Filed on Mar. 22, 2024, as Appl. No. 18/613,518.
Int. Cl. G06Q 30/02 (2023.01); G06Q 30/0204 (2023.01); G06Q 30/0242 (2023.01); G06Q 30/0251 (2023.01)
CPC G06Q 30/0253 (2013.01) [G06Q 30/0204 (2013.01); G06Q 30/0246 (2013.01)] 13 Claims
OG exemplary drawing
 
1. A method performed on a server comprising a processor, a memory, a data storage, and a network interface device connected to a network, the method comprising:
importing listing data using Application Programming Interface (API) connections over the network;
analyzing the listing data to calculate a multimodal vector embedding;
generating content elements for e-commerce display comprising product images, textual descriptions, and infographics based on the listing data, the multimodal vector embedding, and real-time market data metrics comprising click-through rates, time spent on pages, and conversion rates;
wherein the generating comprises:
applying a controlled generation algorithm through a text-to-image diffusion model to create the product images; and
wherein the controlled generation algorithm integrates loss-guidance and attention injection within the diffusion model to produce a controlled layout of the product images;
storing the generated content elements in the data storage;
re-generating the content elements at predetermined intervals or in response to changes in the real-time market data metrics; and
wherein changes in the real-time market data metrics are determined based on a threshold value set for variations in market demand determined by product views and search volume, competitor pricing derived from real-time price comparisons, and consumer behaviour trends based on said click-through rates, time spent on pages, and conversion rates.