Graph Similarity Analytics via Entropy Indices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing graph similarity analysis methods face challenges in efficiently computing similarity between graph-structured datasets, which is computationally expensive and difficult to quantify, especially when represented as vectors, limiting their application in various graph tasks.
Innovation Solution
The system employs an information distance computation based on entropy measures to determine similarity between graphs, reducing memory consumption and computation complexity, and enabling faster graph similarity computations that can be applied in machine learning and artificial intelligence systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If graph similarity analysis is performed using existing methods, then similarity between graphs can be computed, but the computation is expensive and memory consumption is high
Solution Approach 1:
The patent transforms graph similarity computation from direct structural comparison to entropy-based parameter comparison. By converting graph structures into entropy values (information indices), the system changes the computational parameters from complex structural features to simplified statistical measures, dramatically reducing computation cost while preserving similarity measurement capability
Solution Approach 2:
The patent replaces the mechanical/computational process of direct graph structure comparison with an information-theoretic approach. Instead of mechanically comparing nodes and edges, the system uses entropy computation to capture structural information, substituting a complex mechanical comparison process with a more efficient information-based calculation
2Measurement precision
If graph similarity analysis is performed using existing methods, then similarity can be determined, but the computation speed is slow
Solution Approach 1:
The patent changes the computational parameters from detailed structural representations to aggregated entropy measures. This parameter transformation reduces the dimensionality of the computation from O(n²) node-pair comparisons to O(n) entropy calculations, significantly improving computation speed while maintaining measurement precision
Solution Approach 2:
The patent segments the graph similarity problem into two independent stages: (1) computing entropy measures for individual graphs, and (2) comparing these entropy values to determine similarity. This segmentation allows each stage to be optimized independently, with the first stage being computationally efficient and the second stage being simple arithmetic comparison
Data Source
AI summary
Techniques that facilitate graph similarity analytics are provided. In one example, a system includes an information component and a similarity component. The information component generates a first information index indicative of a first entropy measure for a first graph-structured dataset associated with a machine learning system. The information component also generates a second information index indicative of a second entropy measure for a second graph-structured dataset associated with the machine learning system. The similarity component determines similarity between the first graph-structured dataset and the second graph-structured dataset based on a graph similarity computation associated with the first information index and the second information index.