Knowledge Graph Representation Learning via Subgraph Serialization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Designing a knowledge graph is time-consuming and labor-intensive, and suffers from data sparseness, which hinders effective representation learning for entities and relationships in knowledge bases.
Innovation Solution
A method involving sampling a sub-graph from a knowledge graph, serializing it into a text, and using a pre-trained language model to learn knowledge representations, addressing data sparseness and improving representation learning for entities and relationships.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a knowledge graph is manually designed to store and represent knowledge bases, then the knowledge can be structured and queried, but the process becomes time-consuming and labor-intensive
Solution Approach 1:
The patent uses neural networks to automatically copy and transform knowledge from unstructured text sources into structured knowledge graph representations, eliminating manual knowledge graph construction while preserving knowledge accuracy through learned semantic mappings
Solution Approach 2:
The patent replaces the mechanical manual process of knowledge graph design with an automated neural network-based system that learns to extract and structure knowledge representations from text, substituting human labor with intelligent algorithms
2Reliability
If a knowledge graph is manually designed, then knowledge structure can be established, but data sparseness problems occur
Solution Approach 1:
The patent performs preliminary action by pre-training neural networks on large corpora to learn general knowledge representations and semantic relationships before applying them to specific knowledge extraction tasks, enabling the system to handle sparse data effectively through transferred knowledge
Solution Approach 2:
The patent changes parameters by transforming knowledge representations into different vector space dimensions and formats, allowing the neural network to capture semantic relationships in a dense continuous space that overcomes the sparsity inherent in traditional knowledge graph structures
3Loss of information
If traditional representation learning is used for entities and relationships, then semantic information can be extracted, but the performance of knowledge acquisition and reasoning is limited
Solution Approach 1:
The patent combines multiple representation learning techniques and neural network architectures into a composite system that integrates entity representations, relationship representations, and contextual information, creating a more powerful and versatile knowledge reasoning framework
Data Source
AI summary
A method, an apparatus, a device and a storage medium for learning a knowledge representation are provided. The method can include: sampling a sub-graph of a knowledge graph from a knowledge base; serializing the sub-graph of the knowledge graph to obtain a serialized text; and reading using a pre-trained language model the serialized text in an order in the sub-graph of the knowledge graph, to perform learning to obtain a knowledge representation of each word in the serialized text. The knowledge representation learning in this embodiment is performed for entity and relationship representation learning in the knowledge base.