Digital Entity Correlation via Multi-Resolution Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing entity matching techniques in knowledge graphs are inefficient due to their reliance on localized attributes and queue-based methods, which fail to utilize holistic characteristics and scale well with large datasets, leading to suboptimal matching and high hardware and memory requirements.
Innovation Solution
The approach involves encoding attribute values using a multi-resolution encoding scheme, constructing a pairs graph to identify seed pairs, expanding to include related entities, and computing similarity metrics to determine matches, allowing for efficient entity correlation and matching across large knowledge graphs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If queue-based entity matching techniques are used, then entity matching can be performed, but hardware and memory requirements become prohibitively high and scalability deteriorates
Solution Approach 1:
The patent segments the entity matching process into multiple stages: candidate generation using localized attributes, candidate filtering using holistic characteristics, and final matching resolution. This segmentation allows the system to process entities in manageable chunks rather than loading all candidates into memory simultaneously, reducing hardware and memory requirements while maintaining matching accuracy.
Solution Approach 2:
The patent introduces a new dimension of processing by computing holistic characteristics (beyond immediate attributes) and using them as an additional filtering dimension. This allows the system to reduce the candidate set efficiently without requiring proportional increases in memory, as the holistic characteristics provide a compact representation that captures essential entity properties.
2Productivity
If localized attributes only are used for entity matching, then processing speed is maintained, but matching accuracy deteriorates due to insufficient information utilization
Solution Approach 1:
The patent performs preliminary computation of holistic characteristics for each entity before the matching process begins. These pre-computed holistic features are then used as efficient filters during candidate generation and filtering stages, allowing the system to quickly eliminate non-matching candidates without performing expensive detailed comparisons, thus maintaining processing speed while improving accuracy.
Solution Approach 2:
The patent computes holistic characteristics that capture more information than strictly necessary for basic matching (excessive action), but uses them selectively to filter candidates. This partial application of comprehensive feature analysis allows the system to achieve high matching accuracy by focusing detailed analysis only on promising candidate pairs rather than all possible pairs, maintaining overall processing efficiency.
Data Source
AI summary
Techniques for using digital entity correlation to generate a composite knowledge graph from constituent graphs. In an aspect, digital attribute values associated with primary entities may be encoded into primitives, e.g., using a multi-resolution encoding scheme. A pairs graph may be constructed, based on seed pairs calculated from correlating encoded primitives, and further expanded to include subjects and objects of the seed pairs, as well as pairs connected to relationship entities. A similarity metric is computed for each candidate pair to determine whether a match exists. The similarity metric may be based on summing a weighted landing probability over all primitives associated directly or indirectly with each candidate pair. By incorporating primitive matches from not only the candidate pair but also from pairs surrounding the candidate pair, entity matching may be efficiently implemented on a holistic basis.