Cross-Category Product Recommendation via Evaluation Value Integration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing product recommendation systems fail to suggest appropriate combinations of products across different categories, leading to suboptimal recommendations.
Innovation Solution
An image retrieval apparatus that identifies product images from different categories that have optimal evaluation value patterns, allowing for the generation of combined product images that can be used to recommend suitable product combinations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If products are recommended individually for each category, then the recommendation coverage is improved, but the combination appropriateness deteriorates
Solution Approach 1:
The patent merges individual category recommendations by introducing a combination recommendation unit that integrates products from multiple categories. The recommendation engine combines evaluation values across categories (upper garments, lower garments, shoes, hats) to generate coordinated outfit suggestions, ensuring that recommended products form appropriate combinations rather than isolated items.
Solution Approach 2:
The recommendation system achieves multi-functionality by enabling both individual category recommendations and cross-category combination recommendations through a unified engine. The same evaluation value calculation mechanism serves both purposes, allowing the system to adapt between recommending single-category products and multi-category outfits based on user needs.
2Device complexity
If product recommendations are generated without considering cross-category relationships, then the processing complexity is reduced, but the recommendation quality deteriorates
Solution Approach 1:
The patent segments the recommendation process into distinct functional units: an evaluation value calculation unit that computes compatibility scores for individual products, and a combination recommendation unit that integrates these scores across categories. This segmentation allows the system to handle complex cross-category relationships through modular processing, managing complexity while maintaining recommendation quality.
Solution Approach 2:
The system uses parameter changes by introducing evaluation values as a quantitative measure of product compatibility. By converting qualitative assessment into numerical evaluation values that can be calculated and compared across different product categories, the system enables precise cross-category matching without excessive processing complexity.
Data Source
AI summary
An image retrieval apparatus acquires an image to be processed, the image to be processed including at least one article. The image retrieval apparatus specifies, as an association image, a product image of a product having an evaluation value that has a predetermined relationship with an evaluation value of the article included in the image to be processed, from among product images of a plurality of products each belonging to a particular category, the particular category being different from a category to which the article included in the image to be processed belongs. The image to be processed and the specified association image may be output in association with each other.