Visual Blending Content Recommendations via Embedding Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users face challenges in finding specific content from large online catalogs, particularly when selection is based on aesthetic features, as they may not be aware of relevant keywords, leading to inefficient keyword searches and limited recommendations based solely on the currently displayed item.
Innovation Solution
The system employs visually blending content recommendations by using convolutional neural networks to generate embedding vectors for content items, allowing users to select multiple anchor items and display recommendations in a metric space, where items are visually similar based on their proximity to the anchor items, enabling a seamless transition between recommended items.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If users perform keyword-based searches to find content, then they can locate specific items, but they may not be aware of appropriate search terms leading to inefficient searching
Solution Approach 1:
The system automatically generates visual search results based on the uploaded image without requiring users to manually enter keywords or know appropriate search terms. The convolutional neural network performs the search function autonomously by comparing visual features of the uploaded item with the catalog database.
Solution Approach 2:
The patent replaces the manual keyword-based search mechanism with an automated visual recognition system using convolutional neural networks. Instead of users typing text queries, the system processes images directly to find visually similar items, substituting mechanical text input with automated image analysis.
2Adaptability or versatility
If recommendation systems recommend content based on currently displayed content, then recommendations are generated, but they are limited to single-item based recommendations
Solution Approach 1:
The patent transitions from single-item recommendations to multi-item visual blending recommendations by adding the dimension of visual similarity comparison. Instead of recommending based on one anchor item, the system accepts multiple anchor items and generates recommendations that blend visual characteristics across all anchors, creating a richer recommendation space.
Solution Approach 2:
The recommendation system is enhanced to handle multiple functions: it can process single or multiple anchor items, generate both individual and blended recommendations, and adapt to different user preferences. The system universally handles various input configurations while maintaining a unified recommendation framework.
Data Source
AI summary
Systems and techniques are generally described for generating visually blended recommendation grids. In some examples, a selection of a first item and a second item displayed on a display may be received. In various examples, the first item may be displayed in a first element of a grid and the second item may be displayed in a second element of the grid. In some examples, a third element of the grid that is disposed between the first element and the second element along an axis of the grid may be determined. In various examples, a third item may be determined for display in the third element of the grid based at least in part on a blended representation of an embedding of the first item and an embedding of the second item. The third item may be displayed in the third element of the grid.