Knowledge Graph Node Labeling via Deep Language Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing knowledge bases are manually constructed, making them time-consuming and expensive to scale and maintain, and they struggle with efficient retrieval of knowledge in a scalable and intuitive manner.
Innovation Solution
A computer-implemented method for updating a knowledge base with topic types by storing a knowledge graph, accessing a topic type hierarchy computed from a corpus of text documents, and labelling nodes with topic types using a deep language model or template matching, thereby improving the knowledge base's accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If knowledge bases are manually constructed, then accuracy and completeness can be maintained, but construction time and cost increase significantly
Solution Approach 1:
The system performs automated topic type discovery and node labeling using deep language models and template matching algorithms. The knowledge base construction process serves itself by automatically inferring topic types from text documents and assigning labels to nodes without human intervention, thereby maintaining accuracy while dramatically reducing construction time and cost
Solution Approach 2:
The patent replaces manual mechanical labeling processes with automated computational systems. Deep language models and template matching algorithms substitute human experts in identifying topic types and labeling knowledge graph nodes, achieving both high precision and scalability simultaneously
2Measurement precision
If knowledge bases are manually constructed with detailed labels, then query accuracy improves, but retrieval efficiency decreases
Solution Approach 1:
The system segments the query processing into two stages: first, automated topic type discovery and node labeling during knowledge base construction; second, efficient template-based retrieval during query execution. This segmentation allows detailed labeling to be performed once during construction while enabling fast retrieval through pre-structured templates
Solution Approach 2:
The system performs preliminary action by automatically discovering topic types and labeling all nodes during the knowledge base construction phase. This advance preparation creates a pre-structured knowledge base with inferred topic types, enabling efficient retrieval operations without requiring detailed manual labeling at query time
3Productivity
If automated labeling methods are used, then construction speed increases, but labeling precision decreases
Solution Approach 1:
The system introduces templates as an intermediary between automated labeling and topic types. Template matching acts as a mediator that guides the deep language model to infer topic types according to predefined patterns, ensuring that automated labeling achieves both high speed and high precision by constraining predictions to valid topic type structures
Solution Approach 2:
The system changes the parameter of labeling precision by using deep language models with configurable confidence thresholds and template matching with adjustable similarity criteria. These parameter adjustments allow the system to optimize the balance between construction speed and labeling precision based on specific application requirements
Data Source
AI summary
In various examples there is a computer-implemented method of database construction. The method comprises storing a knowledge graph comprising nodes connected by edges, each node representing a topic. Accessing a topic type hierarchy comprising a plurality of types of topics, the topic type hierarchy having been computed from a corpus of text documents. One or more text documents are accessed and the method involves labelling a plurality of the nodes with one or more labels, each label denoting a topic type from the topic type hierarchy, by, using a deep language model; or for an individual one of the nodes representing a given topic, searching the accessed text documents for matches to at least one template, the template being a sequence of words and containing the given topic and a placeholder for a topic type; and storing the knowledge graph comprising the plurality of labelled nodes.