Generative AI Content Carousels for Relevance and Novelty

The patent transitions from a single-dimension relative relevance scoring system to a multi-dimensional absolute relevance scoring system that incorporates multiple criteria including revenue potential and novelty. This is achieved by introducing separate scoring dimensions (relevance score, revenue score, novelty score) that can be independently optimized and then combined, allowing the system to handle multiple optimization objectives simultaneously while maintaining structured control over each dimension.

The patent extracts and separates high-value content items (those meeting predetermined thresholds for relevance, revenue, and novelty scores) from the general content pool and places them in dedicated prominent positions such as promoted slots or featured sections. This extraction allows the majority of standard content to be displayed in conventional layouts without clutter, while the extracted high-value items receive enhanced visibility, thus maintaining interface clarity while preserving content completeness.

The patent applies preliminary absolute relevance scoring and threshold filtering to efficiently identify and promote high-value content items that meet established criteria for revenue and novelty. Simultaneously, the system maintains adaptability by allowing dynamic adjustment of thresholds and incorporating feedback mechanisms that can elevate emerging content items that initially fall below thresholds but show promise for future value, thus balancing reliable optimization with adaptability to emerging trends.