Facet Cleaner Unit for E-Commerce Search Relevance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing online search engines often display irrelevant product categories or facets to consumers, degrading the user experience and reducing the effectiveness of search results.
Innovation Solution
A system comprising a memory unit, an order unit, a search engine unit, and a facet cleaner unit that uses statistical models based on historical usage data to filter out irrelevant facet values from search results, ensuring only relevant product information is displayed to consumers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all facet values are displayed to customers, then complete product information is provided, but irrelevant information degrades user experience
Solution Approach 1:
The facet cleaner unit extracts and removes irrelevant facet values from the search results data based on statistical models, keeping only the relevant facets for display to customers
Solution Approach 2:
The system uses historical usage information as feedback to train statistical models that predict which facet values are relevant, continuously improving filtering accuracy based on customer behavior patterns
2Ease of operation
If statistical filtering is applied to remove irrelevant facets, then user experience is improved, but system complexity increases
Solution Approach 1:
Statistical models are pre-trained using historical usage information before actual search operations, so that during runtime the facet cleaner unit can efficiently filter results without complex real-time analysis
Solution Approach 2:
The facet cleaner unit acts as an intermediary component between the search engine unit and the commercial product module, handling the complex filtering logic separately to maintain modularity
Data Source
AI summary
A system, method and computer product for displaying product information is described herein. Facet values that may cause bad customer experience(s) are identified and suppressed. The system, method, and computer product may use a randomization scheme to suppress and/or show “bad” facet values occasionally to gather recent data on the facet values have improved. The suppression process works as a robust noise filter on top of the search and browse faceting experience.