Search Result Ranking with Conversion-Value Score Boosting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Online systems face challenges in ranking search results based on relevance and value, leading to user frustration and reduced revenue due to irrelevant or expensive items being prioritized over relevant and affordable options, making it difficult for users to compare brands and find desired items.
Innovation Solution
An online system ranks items using a machine learning model that considers item category, retailer type, and user segment, boosting scores based on historical data to predict conversion probability and expected value, ensuring items are ranked based on relevance, affordability, and user preferences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If items are ordered based on value (price) to maximize revenue, then revenue is improved, but user satisfaction deteriorates because relevant items are not prominently displayed
Solution Approach 1:
The patent applies different ranking criteria to different segments of search results. High-value items receive boosted positions in the ranking, while maintaining relevance-based ordering for other items. This local differentiation allows the system to maximize revenue from high-value items without completely disrupting the relevance-based ordering that satisfies users.
Solution Approach 2:
The system dynamically adjusts ranking parameters by introducing a boost factor that modifies the position of items based on their value. This parameter change allows flexible control over the balance between revenue optimization and user satisfaction, enabling the system to prioritize high-value items while maintaining overall relevance.
2Ease of operation
If items are ordered based on relevance to improve user satisfaction, then user satisfaction is improved, but revenue deteriorates because high-value items are not prominently displayed
Solution Approach 1:
The patent introduces localized adjustments to the relevance-based ranking by applying value-based boosts to specific items. This allows the system to maintain overall relevance-based ordering for user satisfaction while creating local exceptions for high-value items to increase revenue.
Solution Approach 2:
The system pre-calculates boost factors for items based on their value attributes before final ranking. This preliminary action allows high-value items to be automatically prioritized in the ranking without requiring real-time intervention, balancing revenue optimization with user satisfaction.
3Loss of energy
If items with varying prices are scattered throughout search results to maximize revenue exposure, then revenue is improved, but ease of comparison deteriorates because similar items are not grouped together
Solution Approach 1:
The patent applies value-based boosting within category groups rather than across all items. This allows similar items to remain grouped together for easy comparison while enabling high-value items within each group to receive prominent positioning for revenue optimization.
Data Source
AI summary
An online system receives a search query from a client device associated with a user and queries a database including item data for a set of items matching the query, in which the set of items is at a retailer location associated with a retailer type and each item is associated with an item category. For each item of the set, a machine learning model is applied to predict a probability of conversion for the user and item and a score is computed based on an expected value, in which the expected value is based on a value associated with the item and the probability. The score for each item is boosted based on the item category, retailer type, or a user segment that is based on the user's historical order data. The items are ranked based on the boosted scores and the ranking is sent to the client device.