Knowledge Graph Filling via Adaptive Model Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for filling knowledge graphs are not effective in handling unknown data points, as they lack a suitable model for classification and distribution of data points based on labels, leading to inefficient clustering and relationship definition.
Innovation Solution
A method is developed to train a classification model using subsets of data points, where data points are associated with clusters based on label distribution conditions, and a quality factor is used to select the best model for defining relationships in the knowledge graph, with cluster centers determined to optimize intra-cluster density.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If data points are clustered based on simple distance metrics, then clustering speed is improved, but clustering accuracy deteriorates
Solution Approach 1:
The patent applies preliminary action by performing multiple exchange operations between data points and cluster centers before final classification. Data points are repeatedly exchanged with cluster centers based on distance calculations, allowing the clustering structure to optimize itself before making final assignments. This preliminary optimization improves clustering accuracy without significantly impacting overall processing speed.
2Device complexity
If a single classification model is used, then model complexity is reduced, but classification accuracy for unknown data points deteriorates
Solution Approach 1:
The patent segments the classification task by training multiple classification models, each specialized for different clusters or data subsets. Instead of using one general model, the system divides the classification problem into multiple specialized models that can be selectively applied based on the input data characteristics, thereby improving accuracy for unknown data points while maintaining manageable complexity through modular architecture.
Solution Approach 2:
The patent implements dynamics by making the model selection adaptive rather than static. The system dynamically selects which classification model to apply based on the characteristics of the input data and the clustered structure. This dynamic approach allows the system to adapt to different types of data points and improve overall classification reliability without requiring a single overly complex model.
3Ease of operation
If data points are uniformly distributed across clusters, then distribution simplicity is improved, but label distribution balance deteriorates
Solution Approach 1:
The patent applies local quality by allowing different clusters to have different numbers of data points and different label distributions based on their local characteristics. Rather than enforcing uniform distribution across all clusters, the system permits each cluster to develop its own optimal structure based on the data it contains, while still maintaining overall balance through the exchange mechanism that prevents any single cluster from becoming overly dominant.
Data Source
AI summary
A method for filling a knowledge graph. A first and second subset of data points are determined. A data point to which a label is assigned is associated with a cluster from among a set of clusters, depending on whether a distribution of labels from data points that are already associated with the cluster satisfies a condition. Data points that are associated with the cluster are associated with the first or second subset. Models for classification are trained depending on data points from the first subset. For at least one of the models, a value of a quality factor is determined depending on data points from the second subset. A model for classification is selected from the models depending on the value. A classification that defines a relationship, node, or type of node in the knowledge graph for the sentence is determined using the selected model.