Automated Ontology Building via Parse Tree Dependency Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The manual intervention required for building ontologies is labor-intensive and inefficient, leading to increased costs and reduced accuracy in text analytics and cognitive computing due to the exponential growth of data, which may result in missing critical concepts.
Innovation Solution
An automated ontology building method that calculates dependency graphs from parsed text, generates concept instance candidates, and arranges them into a tree structure, providing a user interface for editing and refining the ontology, leveraging state-of-the-art Natural Language Processing techniques to extract and represent semantic relationships.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual intervention is used to build ontologies by reading domain specific documents, then accuracy of concept discovery can be maintained, but labor intensity and costs increase significantly
Solution Approach 1:
The system enables automated ontology building where the computer system performs concept discovery, extraction, and relationship identification autonomously through NLP techniques, parse tree analysis, and dependency graph calculation, eliminating the need for manual reading and annotation of domain documents while maintaining comprehensive concept coverage
Solution Approach 2:
The patent replaces the mechanical manual process of reading and analyzing documents with automated computational methods including natural language processing, parse tree generation, dependency graph calculation, and algorithmic concept extraction, substituting human cognitive labor with machine-based automated analysis
2Measurement precision
If more manual refinement is performed on the ontology, then accuracy improves, but productivity decreases due to increased manpower requirements
Solution Approach 1:
The system performs preliminary automated concept extraction, relationship identification, and ontology structure generation before any human intervention is needed, pre-processing the domain documents to create an initial ontology framework that can then be reviewed or refined, thereby accelerating the overall process while maintaining quality
Solution Approach 2:
The patent incorporates feedback mechanisms where the automated system generates ontology candidates that can be evaluated and refined iteratively, with the ability to learn from corrections and improvements, allowing rapid cycling between automated generation and human review to achieve high accuracy without linearly increasing time requirements
3Productivity
If automated NLP techniques are used to extract concepts, then productivity increases, but measurement precision may decrease due to missing critical concepts
Solution Approach 1:
The patent segments the ontology building process into distinct automated stages including tokenization, parse tree generation, dependency graph calculation, concept candidate identification, and relationship extraction, allowing each stage to be optimized independently and enabling comprehensive analysis that reduces missed concepts while maintaining high throughput
Solution Approach 2:
The system employs universal NLP techniques and multiple analysis methods (parse trees, dependency graphs, pattern matching) that can handle diverse domain documents and concept types uniformly, ensuring comprehensive concept extraction across different domains and document styles without requiring domain-specific manual tuning for each case
4Reliability
If comprehensive manual review of all documents is performed, then all critical concepts are captured, but loss of time increases due to the exponential increase in data
Solution Approach 1:
The patent replaces time-consuming manual document review with automated computational analysis that can process exponential amounts of text data rapidly through algorithmic parsing, dependency analysis, and concept extraction, achieving comprehensive concept coverage without linearly increasing review time
Solution Approach 2:
The system creates structured representations (parse trees, dependency graphs) that copy and organize information from the original documents in a machine-analyzable format, allowing comprehensive analysis of all documents through systematic processing of these structured copies rather than repeated human reading of the full text
Data Source
AI summary
A computer-implemented method is provided for automated ontology building. The method calculates, using a parse tree obtained by parsing text, a dependency graph across contextual tokens created from the text. The method generates concept instance candidates and parent-child relationships based on pattern matching and transformation of the parse tree. The method groups concept instance candidates into concept candidates and arranges the concept candidates into a tree having tree nodes. The method provides an ontology based on the tree and text and formed as an output graph of nodes. The method provides a user interface for editing the ontology by selectively adding a new node to the output graph, removing an existing node, moving a node or sub-graph across a parent-child hierarchy in the output graph, creating a new relation across the nodes, and removing an existing relation edge.