Automated Substitute Selection Using Multi-Model Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In e-commerce grocery platforms, identifying substitutes for out-of-stock items that are likely to be accepted by customers is challenging due to the vast number of items and customer preferences, which can lead to high rejection rates and inefficient use of automated resources.
Innovation Solution
A computer-implemented method using multiple trained models to analyze items, identify similar items, determine primary categories, and rank potential substitutes based on rejection probability, thereby minimizing customer rejection and optimizing resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple trained models are used to analyze items and rank substitutes, then the accuracy of substitute selection is improved, but the system complexity increases
Solution Approach 1:
The patent divides the substitute selection task into multiple specialized models, each handling a specific aspect: a first model identifies similar items, a second model determines primary categories, a third model generates potential substitutes, and a fourth model ranks them based on rejection probability. This segmentation allows each model to be optimized for its specific function while collectively achieving high accuracy in substitute selection.
2Productivity
If automated resources are used to pick substitute items, then processing speed is improved, but the rate of customer rejection increases
Solution Approach 1:
The patent incorporates feedback from historical customer data into the ranking model. The fourth trained model uses rejection probability information derived from past customer responses to automatically select substitutes. This feedback mechanism enables the system to learn from previous failures and adjust its selections, maintaining high processing speed while improving customer acceptance rates.
3Reliability
If human pickers are used to determine substitutes, then customer acceptance is improved through subjective judgment, but the processing time increases
Solution Approach 1:
The patent implements a self-service automated system that uses trained models to independently determine substitute items without requiring human intervention. The multi-model approach processes large datasets of customer preferences and item characteristics to automatically generate and rank substitutes, eliminating the time-consuming human judgment process while maintaining high accuracy through data-driven decision-making.
4Reliability
If fixed substitutes are enforced to account for customer preferences, then the rejection rate is reduced, but the system adaptability decreases
Solution Approach 1:
The patent replaces static fixed substitute lists with a dynamic, data-driven ranking system. The fourth trained model continuously processes customer preference data and updates substitute recommendations in real-time. This dynamic approach allows the system to adapt to changing customer preferences, seasonal variations, and individual customer profiles while maintaining high accuracy in predicting customer acceptance.
Data Source
AI summary
Certain aspects of the disclosure provide methods and systems for determining substitutes for an item in a computationally efficient way.