Complementary Product Recommendations with Multi-Modal Embeddings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining complementary product recommendations often require large amounts of baseline data or manual definition, making it difficult to automatically identify complementary items based on a relatively small amount of data.
Innovation Solution
A system and method using a Siamese network with Bidirectional LSTM components to calculate similarity scores based on text and image features, specifically RGB color histograms, to recommend complementary products.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If existing methods are used for determining complementary product recommendations, then manual definition or large amounts of baseline data are required, but this makes it difficult to automatically identify complementary items with limited data
Solution Approach 1:
The patent replaces manual definition mechanisms with an automated machine learning system. A neural network model automatically learns complementary relationships from product data without requiring manual annotation or large pre-established baseline datasets, enabling automatic identification of complementary items through pattern recognition in product features and user behavior
Solution Approach 2:
The patent transforms the approach by changing from traditional data quantity requirements to feature-based representation. Instead of requiring large amounts of baseline data, the system uses transformed product features (such as category, price, visual characteristics) and learns complementary relationships through machine learning models that operate effectively with limited data samples
2Extent of automation
If manual definition methods are used for complementary items, then automation is reduced, but this increases the complexity and time required for system setup
Solution Approach 1:
The system performs self-service by automatically learning complementary relationships from product data and user interactions without requiring manual configuration or definition. The machine learning model continuously adapts to product catalogs and user behavior patterns, enabling the system to generate recommendations autonomously based on learned patterns rather than pre-programmed rules
Solution Approach 2:
The patent replaces manual definition mechanisms with an automated machine learning system. A neural network model automatically learns complementary relationships from product data without requiring manual annotation or large pre-established baseline datasets, enabling automatic identification of complementary items through pattern recognition in product features and user behavior
Data Source
AI summary
Systems and methods for providing suggestions of complementary products responsive to an anchor product are disclosed. The method includes receiving a selection of an anchor product. A similarity score between text embeddings of the anchor product and text embeddings of a plurality of products in a product database is calculated. A similarity score between an image feature of the anchor product and an image feature of the plurality of products in the product database is calculated. A weighted score between the two similarity scores as calculated for the anchor product and the plurality of products in the product database is calculated. At least one of the products from the product database having a highest weighted score is selected and returned responsive to the selection of the anchor product.