Knowledge Graph Embedding via Relation Path Feature Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for knowledge graph embedding are inefficient and inaccurate, particularly due to the complexity of neural network models and the symbolic nature of knowledge graphs, which hinders effective manipulation and application in tasks like link prediction and relation extraction.
Innovation Solution
A computer-implemented method for knowledge graph embedding that selects node pairs, identifies direct relation paths, collects and counts occurrences of alternative relation paths to form feature vectors, and uses these vectors to construct a prediction model for accurately predicting unknown direct relation paths between target nodes, thereby improving efficiency and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If neural network models are used for knowledge graph embedding, then prediction accuracy can be improved, but model complexity and computational resource requirements increase
Solution Approach 1:
The patent segments the knowledge graph into multiple subgraphs or clusters based on entity relationships and properties. Each subgraph is processed independently or with reduced complexity, allowing accurate predictions without requiring a single complex global model. This divides the computational burden while maintaining prediction accuracy through localized analysis.
Solution Approach 2:
The patent dynamically adjusts model parameters such as embedding dimensions, network depth, and processing granularity based on the specific characteristics of the knowledge graph being analyzed. This allows the system to achieve high accuracy when needed while reducing complexity for simpler tasks, resolving the contradiction between accuracy and model complexity.
2Measurement precision
If comprehensive relation path analysis is performed, then embedding accuracy improves, but computational time and resources increase
Solution Approach 1:
The patent pre-computes and stores relation path features, entity embeddings, and relationship statistics during an offline preprocessing phase. This preliminary action allows the system to quickly retrieve and use pre-analyzed data during online prediction, achieving high accuracy without incurring excessive computational time during actual operations.
Solution Approach 2:
The patent performs comprehensive relation path analysis only when necessary, using simplified or partial analysis for common queries. The system selectively applies full analysis depth based on query complexity and confidence thresholds, achieving good enough accuracy for most cases while reserving comprehensive analysis for challenging predictions, thus balancing accuracy with computational efficiency.
Data Source
AI summary
A computer-implemented method, system and computer program product for knowledge graph embedding. A node pair for each triple of a knowledge graph is selected and a direct relation path between the selected node pair is identified. Furthermore, for each triple, a set of relation paths between the selected node pair is collected except for a path representing the direct relation path. The number of occurrences of each relation path for each triple in the collected set of relation paths is counted thereby forming a feature vector set for each triple, where the feature vector set includes a set of occurrences of each relation path for a node pair along with a corresponding direct relation path. A prediction model is then constructed using the feature vector set for each triple to predict an unknown direct relation path between two target nodes in the knowledge graph.