Related Experiment Videos
An enriched unified medical language system semantic network with a multiple subsumption hierarchy
Li Zhang1, Yehoshua Perl, Michael Halper
1CS Department, NJIT, University Heights, Newark, NJ 07102, USA. lxz1853@njit.edu.
Journal of the American Medical Informatics Association : JAMIA
|February 7, 2004
Summary
The Unified Medical Language System
Area of Science:
- Medical Informatics
- Knowledge Representation
- Ontology Engineering
Background:
- The Unified Medical Language System's (UMLS) Semantic Network (SN) uses a restrictive two-tree structure.
- This structure limits semantic types to a single parent, hindering complex relationships.
Purpose of the Study:
- To expand the UMLS Semantic Network into a multiple subsumption structure.
- To allow semantic types to have multiple parents, creating a more flexible knowledge representation.
Main Methods:
- Developed two methodologies for identifying and adding new IS-A links.
- Method 1: Connectivity-based transformations on SN partitions.
- Method 2: String matching on semantic type names and definitions, with expert review.
Main Results:
- Added 19 new IS-A links and created 4 new semantic types.
- Developed the Enriched Semantic Network (ESN) with a directed acyclic graph (DAG) hierarchy.
- Derived a partition of the ESN with 19 connected groups.
Conclusions:
- The Enriched Semantic Network (ESN) offers a more comprehensive abstraction of the UMLS.
- Its multiple subsumption hierarchy better represents complex semantic relationships.
- Provides enhanced access to broader subsumption knowledge.