Extended Analysis of Topological-Pattern-Based Ontology Enrichment

Area of Science:

• Biomedical Informatics
• Ontology Engineering
• Knowledge Representation

Background:

• Biomedical ontology maintenance is complex and resource-intensive.
• Existing methods for identifying concepts for ontology extension require significant expert input.
• The Unified Medical Language System (UMLS) provides a framework for integrating diverse biomedical terminologies.

Purpose of the Study:

• To quantitatively analyze cross-ontology topological patterns for concept insertion.
• To estimate the expert resources required for extending target ontologies from reference ontologies.
• To confirm the impact of 'alternative classification' on concept import feasibility.

Main Methods:

• Developed a topological-pattern-based method for identifying candidate concepts from a reference ontology for a target ontology.
• Quantitatively described the universe of cross-ontology topological patterns.
• Performed theoretical analysis on potential concept placements and synonym creation within target ontology paths.
• Utilized the National Cancer Institute thesaurus (NCIt) as a target ontology.

Main Results:

• Quantified cross-ontology topological patterns, providing a basis for resource estimation.
• Confirmed that 'alternative classification' frequently prevents concept importation.
• Demonstrated the method's applicability to hierarchical ontologies within the UMLS.

Conclusions:

• The topological-pattern method offers a quantitative approach to managing biomedical ontology integration.
• Accurate estimation of expert resources is crucial for efficient ontology extension.
• Understanding topological patterns and classification constraints is key to successful ontology merging.