Classification and Conceptualization of Clinical Documents using Formal Concept Analysis.
- Author:
Myeng Ki KIM
1
;
Suk Hyung HWANG
;
Hong Gee KIM
;
Yu Kyung KANG
;
Hee Chul CHOI
;
Dong Soon KIM
Author Information
1. DERI Seoul, School of Dentistry, Seoul National University, Korea.
- Publication Type:Original Article
- Keywords:
Ontology;
Formal Concept Analysis;
Concept Hierarchy;
Classification;
Clinical Documents
- MeSH:
Classification*;
Medical Informatics;
Statistics as Topic
- From:Journal of Korean Society of Medical Informatics
2006;12(1):31-43
- CountryRepublic of Korea
- Language:Korean
-
Abstract:
OBJECTIVE: Ontology is becoming a core research field in the realm of medical informatics. The objective of our ongoing research is to explore the potential role of Formal Concept Analysis(FCA) in a context-based ontology building support in a medical domain. The concept hierarchy plays an important role as the backbone of ontology, but its construction is a complex and time-consuming process. We present a novel approach to the automatic acquisition of taxonomies or concept hierarchies from clinical documents. METHODS: Our approach is based on FCA, a mathematical tool used in data analysis and knowledge engineering. It provides methods to group objects and attributes into concepts, pairs of object-sets(clinical documents) and attribute-sets(fields contained in the clinical documents), such that the binary relation can be presented in a concept lattice. Based on the FCA, we have applied out approach for 8 clinical documents used in a university hospital. As a result of our experiments, we can extract 15 concepts with 7 common fields that can be shared with 8 clinical documents. RESULTS: We show how FCA can be used to classify clinical documents and acquire a concept hierarchy for the medical domain out of the clinical documents with maximal property factorization. CONCLUSION: The whole of our work is based on the concept lattice of which allows to construct a "well defined" ontological concept hierarchy. As an application of this approach, we presented some results of classification of clinical documents with maximally factorized common fields. We have shown that FCA can be useful method to classify and analyze various medical data by constructing concept hierarchy. From that concept hierarchy, we can acquire well-structured facts and knowledges in medical domain.