Relationship Detector for Unpredictable Connections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing relationship mining technologies find it difficult to search for unpredictable relationships by considering multiple standpoints, as they primarily use independent relationships and lack structures for mutual comparison, making it hard to detect unexpected connections between elements from different perspectives.
Innovation Solution
A relationship detector that calculates first and second distances between elements within different relationships and calculates an unpredictability score through a predetermined rule, allowing for the identification of unexpected relationships by comparing these distances, thereby displaying combinations with high unpredictability scores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If independent relationships are used for mining, then the structure is simple and easy to implement, but the ability to search unpredictable relationships from multiple standpoints is insufficient
Solution Approach 1:
The patent introduces a new dimension by calculating distances in multiple relationship standpoints (first relationship and second relationship) simultaneously. Instead of using a single relationship metric, the system computes distance scales from different perspectives and compares them to identify unpredictable relationships, thereby enhancing adaptability without excessive complexity increase
Solution Approach 2:
The patent segments the relationship analysis into distinct components: first distance calculation based on first relationship, second distance calculation based on second relationship, and unpredictability calculation comparing these distances. This segmentation allows the system to handle multiple standpoints independently while maintaining overall structure clarity
2Measurement precision
If linear sum of multiple relationships is used, then the similarity search capability is improved, but the ability to compare relationships from different standpoints is lost
Solution Approach 1:
Instead of summing multiple relationships to get a single similarity value (which loses comparative information), the patent inverts the approach by calculating the ratio or difference between distances in different relationships. This inversion preserves the individual relationship characteristics while enabling mutual comparison, thus maintaining measurement precision without losing comparative information
3Measurement precision
If only close relationships in graph structure are searched, then the detection accuracy for similar interests is high, but unexpected relationships and serendipity are missed
Solution Approach 1:
The patent changes the parameter for relationship evaluation from单纯的 graph distance to a composite parameter that includes distance ratios across multiple relationships. By incorporating unpredictability calculation that compares first and second distances, the system can identify elements that are far in one relationship but close in another, thus discovering unexpected relationships while maintaining detection accuracy through multi-dimensional validation
Data Source
AI summary
For a group of elements defined with a first relationship between elements stored in a first data memory unit and a second relationship therebetween different from the first relationship stored in a second data memory unit, a relationship detector includes a first distance calculating unit that calculates a predetermined first distance between the two elements belonging to the group in the first relationship, a second distance calculating unit that calculates a predetermined second distance between the two elements belonging to the group in the second relationship, and an unpredictability calculating unit that calculates a dissociation level between the first distance and the second distance between the two elements belonging to the group through a predetermined rule.