Entity Resolution Using Self-Information Scores
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for entity resolution in entity graphs are inefficient, as they require pairwise comparisons of database records, leading to suboptimal performance in determining whether different records represent the same real-world entity.
Innovation Solution
The method involves assigning self-information scores to vertices in an entity graph, allowing vertices to update their entity information based on the self-information score of directly connected vertices with higher scores, thereby optimizing entity resolution by directly linking related records.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pairwise comparison of database records is performed to determine entity relationships, then entity resolution accuracy is improved, but processing time and computational complexity increase
Solution Approach 1:
The patent pre-calculates self-information scores for each vertex based on its record content before entity resolution begins. This preliminary computation of information content allows the system to quickly assess vertex quality without performing full pairwise comparisons, reducing processing time while maintaining resolution accuracy through informed selection of representative vertices
Solution Approach 2:
The patent extracts only the most informative vertices from the entity graph by evaluating self-information scores and selecting vertices with highest information content. By taking out and processing only these key vertices rather than all vertices in pairwise comparisons, the system achieves accurate entity resolution with significantly reduced computational overhead and processing time
2Reliability
If pairwise comparison of all record pairs is performed, then complete entity relationship detection is improved, but computational complexity increases
Solution Approach 1:
The patent applies local quality by assigning different self-information scores to different vertices based on the information content of their associated records. Vertices with higher information content (more distinctive or informative records) are identified and prioritized, allowing the system to focus computational resources on the most relevant entity relationships rather than uniformly processing all pairs
Solution Approach 2:
The patent changes the parameter from comparing all vertex pairs to comparing only selected vertices based on self-information score thresholds. By transforming the problem from a complete graph traversal to a selective sampling approach based on information content parameters, the system maintains reliable entity relationship detection while dramatically reducing computational complexity
3Reliability
If entity resolution is performed on all vertices in the entity graph, then resolution completeness is improved, but processing efficiency decreases
Solution Approach 1:
The patent performs entity resolution on a selective subset of vertices rather than all vertices in the graph. By applying partial action focused on vertices with highest self-information scores and those meeting specific selection criteria, the system achieves sufficient resolution completeness for practical purposes while significantly improving processing efficiency through reduced scope
Data Source
AI summary
Provided are a computer program product, system, and method to determine whether to perform entity resolution on vertices in an entity graph. A determination is made of pairs of records in a database having a relationship value satisfying a threshold. An entity relationship graph has a vertex for each of the records of the pairs and an edge between two vertices. Each vertex has a self-information score based on content in the record, an initial unique entity identifier, and an entity information score. For each subject vertex of the vertices, a determination is made of a target vertex directly connected to the subject vertex that has a highest entity information score and whether to set the subject vertex entity identifier and entity information score to the entity identifier and entity information score of the target vertex based on the target vertex self-information score.