Bipartite Graph Topology for Entity Resolution Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing entity resolution techniques fail to incorporate entity relationships, leading to decreased accuracy and increased processing time due to their reliance solely on demographic information without considering relationships between entities.
Innovation Solution
The use of bipartite graph topology to explore relationships between entities, where similarity edges connect potential duplicates, and iterative neighborhood expansion to refine similarity scores and identify mergeable entities, incorporating vertex weights and PageRank for weighted embodiments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing entity resolution techniques rely solely on demographic information without considering entity relationships, then the processing is simpler and faster, but the accuracy of entity resolution decreases
Solution Approach 1:
The patent transitions from examining entities in isolation (one-dimensional approach using only demographic attributes) to analyzing entities within the context of their relationships (multi-dimensional approach using graph topology). By incorporating the structural dimension of entity relationships into the resolution process, the system achieves higher accuracy without proportionally increasing complexity.
Solution Approach 2:
The patent introduces graph topology metrics and neighborhood information as intermediary elements that mediate between raw demographic data and entity resolution decisions. These intermediaries (relationship patterns, connectivity metrics) provide additional contextual information that enhances accuracy while maintaining manageable processing complexity through structured computation.
2Measurement precision
If existing entity resolution techniques examine only individual entities, then the processing duration is shorter, but the accuracy of entity resolution decreases
Solution Approach 1:
The patent performs preliminary computation of graph topology metrics and neighborhood information before the actual entity resolution matching process. By pre-computing relationship-based features and storing them in the graph structure, the system avoids repeated expensive computations during matching, thus achieving higher accuracy without proportionally increasing total processing time.
Solution Approach 2:
The patent maintains continuous utilization of computed graph topology information throughout the entity resolution process. Rather than computing relationship metrics only once or repeatedly from scratch, the system continuously leverages the pre-established graph structure and its derived metrics across multiple resolution operations, optimizing the balance between accuracy and processing duration.
3Measurement precision
If entity relationships are incorporated into entity resolution calculation, then the accuracy is improved, but the processing duration needed to achieve fixed accuracy is increased
Solution Approach 1:
The patent segments the entity resolution process into distinct phases: graph topology computation, neighborhood information extraction, and final matching decision. By dividing the complex process of incorporating relationship information into manageable segments, the system achieves high accuracy while improving processing efficiency through specialized optimization at each stage rather than uniform complex processing throughout.
Solution Approach 2:
The patent dynamically adjusts parameters such as neighborhood radius, graph traversal depth, and metric computation granularity based on the specific resolution task and data characteristics. This adaptive parameter adjustment allows the system to incorporate relationship information effectively for accuracy while avoiding unnecessary computational overhead, thus maintaining high productivity.
Data Source
AI summary
In an embodiment, a computer stores a bipartite graph that consists of a source subgraph and a target subgraph. Each vertex in the bipartite graph represents an entity. The source subgraph and the target subgraph are connected by many similarity edges. Each similarity edge indicates an original amount of similarity between the entity of a source vertex in the source subgraph and the entity of a target vertex in the target subgraph. For each similarity edge, the computer determines: a set of neighbor source vertices that are reachable from the source vertex of the similarity edge by traversing at most a source radius count of source edges in the source subgraph, a set of neighbor target vertices that are reachable from the target vertex of the similarity edge by traversing at most a target radius count of target edges in the target subgraph, and various amounts based on graph topology. For each similarity edge, the computer calculates a new amount of similarity based on those various amounts.