Attribute Redundancy Removal for Accurate Item Similarity Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional similarity scoring approaches overestimate the similarity of items due to attribute redundancy, failing to account for duplicative or non-discriminatory attribute values, leading to inaccurate similarity calculations and manual intervention challenges in large and complex systems.
Innovation Solution
A tree-based apparatus with pruning logic identifies and removes redundant attribute values, using incremental discriminatory power metrics like Entropy and Gini Index to generate a pruned tree for accurate similarity scoring, thereby preventing inflated similarity scores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional similarity scoring approaches are used that match attribute values, then the similarity calculation is simple to implement, but the similarity score is overestimated due to attribute redundancy
Solution Approach 1:
The patent extracts and removes redundant attribute values from the item attribute sets before calculating similarity. The pruning logic identifies and eliminates duplicate attributes and attributes with single values, keeping only the non-redundant attributes for similarity computation. This extraction approach directly addresses the overestimation problem by removing the harmful redundant information while maintaining the essential discriminative attributes.
Solution Approach 2:
The patent segments the attribute processing into distinct phases: first identifying redundant attributes, then pruning them separately, and finally calculating similarity on the pruned attributes. The pruning logic is further segmented to handle different types of redundancy (duplicate attributes vs. single-value attributes) through separate processing steps, allowing for more precise control over the attribute reduction process.
2Measurement precision
If manual intervention is used to clean product attributes, then attribute redundancy can be reduced, but the process becomes impractical for large and complex systems
Solution Approach 1:
The patent implements self-service by automatically detecting and pruning redundant attributes without requiring manual intervention. The pruning logic autonomously analyzes the attribute structure, identifies redundant patterns (duplicates and single-values), and removes them automatically. This self-service capability enables the system to handle large and complex attribute sets efficiently, making the process scalable to modern retail systems with thousands of products and attributes.
Solution Approach 2:
The patent replaces the mechanical manual cleaning process with an automated computational system. Instead of requiring human operators to manually review and clean attributes, the pruning logic uses algorithmic rules to automatically identify and remove redundant attributes. This substitution of mechanical human labor with automated computing enables high-throughput processing of large attribute datasets.
3Loss of information
If all attributes including redundant ones are used for similarity calculation, then more information is available for comparison, but the similarity score becomes inflated and less discriminatory
Solution Approach 1:
The patent applies local quality by differentiating between useful and useless attributes at the local level of individual attributes. Rather than uniformly keeping or removing all attributes, the pruning logic selectively removes only the redundant ones (duplicates and single-values) while preserving the non-redundant attributes that provide discriminatory power. This local differentiation ensures that information retention is optimized for each attribute based on its actual contribution to similarity discrimination.
Data Source
AI summary
Systems, methods, and other embodiments associated with attribute redundancy removal are described. In one embodiment, a method includes identifying redundant attribute values in a group of attributes that describe two items. The example method also includes generating a pruned group of attributes having the redundant attribute values removed. The similarity of the two items is calculated based, at least in part, on the pruned group of attribute values.