Date of Award
5-2010
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Legacy Department
School of Computing
Committee Chair/Advisor
Wang, Zijun
Committee Member
Srimani , Pradip
Committee Member
McGregor , John
Committee Member
Luo , Feng
Abstract
Medical Subject Headings (MeSH) is a controlled vocabulary used by the National Library of Medicine to index medical articles, abstracts, and journals contained within the MEDLINE database. Although MeSH imposes uniformity and consistency in the indexing process, it has been proven that using MeSH indices only result in a small increase in precision over free-text indexing. Moreover, studies have shown that the use of controlled vocabularies in the indexing process is not an effective method to increase semantic relevance in information retrieval.
To address the need for semantic relevance, we present an ontology-based information retrieval system for the MEDLINE collection that result in a 37.5% increase in precision when compared to free-text indexing systems. The presented system focuses on the ontology to: provide an alternative to text-representation for medical articles, finding relationships among co-occurring terms in abstracts, and to index terms that appear in text as well as discovered relationships. The presented system is then compared to existing MeSH and Free-Text information retrieval systems.
This dissertation provides a proof-of-concept for an online retrieval system capable of providing increased semantic relevance when searching through medical abstracts in MEDLINE.
Recommended Citation
Taylor, William II, "CREATING A BIOMEDICAL ONTOLOGY INDEXED SEARCH ENGINE TO IMPROVE THE SEMANTIC RELEVANCE OF RETREIVED MEDICAL TEXT" (2010). All Dissertations. 535.
https://open.clemson.edu/all_dissertations/535