Industrial Knowledge Graph Construction from Mixed Data Sources
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The industrial field faces challenges in constructing knowledge graphs due to the messiness and disordered nature of its information, which often comes from unstructured texts, making it difficult to perform semantic analysis and extract knowledge units effectively.
Innovation Solution
A method involving machine learning to extract knowledge from unstructured data, combined with structured and semi-structured data, to build a knowledge graph by integrating multiple sub-fields, represented in the form of triples, while considering the proportion of unstructured data and industrial category/sub-category similarity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual semantic analysis is performed on unstructured industrial data to extract knowledge units, then knowledge accuracy is improved, but construction time and labor cost increase significantly
Solution Approach 1:
The patent applies preliminary action by pre-training machine learning models on annotated industrial data before actual knowledge extraction. The model is prepared in advance with learned patterns from training data, enabling efficient automatic extraction during deployment without manual analysis for each new dataset.
Solution Approach 2:
The patent replaces the mechanical manual semantic analysis process with an automated machine learning-based system. The ML model automatically performs entity recognition, relation extraction, and knowledge graph construction, substituting human manual work with computational processes that scale efficiently.
2Loss of information
If knowledge graph construction is performed manually on messy unstructured industrial data, then knowledge completeness is improved, but implementation complexity increases
Solution Approach 1:
The patent creates a universal knowledge graph construction system that handles multiple data types (structured, semi-structured, and unstructured data) through a single integrated ML-based platform. The system performs multiple functions including data cleaning, entity recognition, relation extraction, and graph construction within one framework, reducing implementation complexity.
Solution Approach 2:
The patent applies parameter changes by adjusting model parameters and processing strategies based on data characteristics. The system dynamically adapts its extraction parameters and processing depth according to the specific industrial domain and data quality, optimizing between completeness and complexity automatically.
3Productivity
If machine learning models are trained on annotated data to automate knowledge extraction, then construction efficiency is improved, but data preparation requirements increase
Solution Approach 1:
The patent applies preliminary action by creating annotated training datasets in advance and using them to pre-train the ML model. This preparatory work is performed once, enabling efficient automatic extraction for multiple subsequent projects without repeating the annotation process.
Solution Approach 2:
The patent implements self-service by enabling the ML model to automatically perform knowledge extraction without requiring manual annotation for each new dataset. The trained model serves itself by processing new data autonomously, reducing the need for ongoing data preparation and manual intervention.
Data Source
AI summary
Various embodiments of the teachings herein include a method for creating a knowledge graph in the industrial field. An example includes: obtaining unstructured data from a first source in a sub-field of the industrial field, with knowledge annotations; performing machine learning on the unstructured data to generate a first model adapted to extract knowledge; extracting knowledge from second unstructured data provided by the first source based on the first model, without knowledge annotations; obtaining first structured data and first semi-structured data from a second source in a second sub-field; extracting second knowledge from the first structured data; extracting third knowledge from the first semi-structured data; and building a knowledge graph integrating the first and second sub-field based on the first, second, and third knowledge, represented in the form of triples.