Relation Graph Optimization via Inconsistent Cycle Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In large text data sets, relationship information extracted from different subsets often exhibits logical inconsistencies, such as conflicting probabilities regarding the order of events or relationships, which becomes more pronounced as data set size increases, posing a challenge in systematically addressing these inconsistencies while extracting useful information.
Innovation Solution
The development of algorithms that analyze text sentences to determine relationship probabilities between entities, construct relationship graphs, identify inconsistent cycles, and prune edges based on a multi-factor optimization function considering inconsistencies and confidence levels, resulting in a consistent probabilistic view of relationships.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If relationship information is extracted from larger data sets, then more relationship signals can be identified, but logical inconsistencies increase
Solution Approach 1:
The patent segments the relationship extraction process into distinct phases: initial graph construction from multiple data subsets, cycle detection to identify inconsistencies, and optimization to resolve conflicts. This segmentation allows the system to handle large datasets by processing them in manageable stages while maintaining consistency through systematic conflict resolution.
Solution Approach 2:
The patent implements feedback mechanisms where the optimization process uses cycle detection results to adjust relationship probabilities. The system continuously refines the relationship graph by using detected inconsistencies as feedback to prune edges and resolve conflicts, thereby improving logical consistency while preserving valid relationship signals from large datasets.
2Loss of information
If all relationship signals are retained, then comprehensive relationship information is captured, but inconsistent cycles are introduced
Solution Approach 1:
The patent extracts and removes inconsistent relationship signals through cycle detection and edge pruning. By identifying cycles that represent logical inconsistencies and selectively removing edges with lower confidence scores, the system eliminates contradictory relationships while preserving the majority of valid relationship information from the dataset.
Solution Approach 2:
The patent changes the probability parameters of relationships during optimization. By adjusting confidence scores based on cycle detection results and pruning edges below threshold values, the system dynamically modifies relationship parameters to eliminate inconsistencies while maintaining high-confidence relationships, thus resolving the contradiction between completeness and consistency.
3Reliability
If relationship graphs are optimized to remove inconsistencies, then logical consistency improves, but some high-confidence edges may be pruned
Solution Approach 1:
The patent applies partial optimization by pruning only those edges that contribute to inconsistent cycles and fall below confidence thresholds. Rather than removing all potentially conflicting relationships, the system selectively prunes edges involved in detected cycles, thereby achieving logical consistency improvement while minimizing loss of valid relationship information through targeted rather than comprehensive pruning.
Data Source
AI summary
A first graph is generated from a text data set, with graph nodes representing named entities in the data set and edges representing relationships between the named entities, and with edge weights indicating confidence levels. At least one cycle of the graph may be designated as inconsistent using a rule set. An edge may be selected for deletion from the first graph based on its presence in an inconsistent cycle, the cycle's weight, and/or on the edge weight. A representation of relationships indicated in the modified graph is provided programmatically.