BOAT : automatic alignment of biomedical ontologies using term informativeness and candidate selection

The biomedical sciences is one of the few domains where ontologies are widely being developed to facilitate information retrieval and knowledge sharing, but there still remains the problem that applications using different ontologies cannot share knowledge without explicit references between overlap...

Full description

Saved in:
Bibliographic Details
Main Authors: Chua, Watson Wei Khong, Kim, Jung-jae
Other Authors: School of Computer Engineering
Format: Article
Language:English
Published: 2013
Subjects:
Online Access:https://hdl.handle.net/10356/105389
http://hdl.handle.net/10220/17512
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:The biomedical sciences is one of the few domains where ontologies are widely being developed to facilitate information retrieval and knowledge sharing, but there still remains the problem that applications using different ontologies cannot share knowledge without explicit references between overlapping concepts. Ontology alignment is the task of identifying such equivalence relations between concepts across ontologies. Its application to the biomedical domain should address two open issues: (1) determining the equivalence of concept-pairs which have overlapping terms in their names, and (2) the high run-time required to align large ontologies which are typical in the biomedical domain. To address them, we present a novel approach, named the Biomedical Ontologies Alignment Technique (BOAT), which is state-of-the-art in terms of F-measure, precision and speed. A key feature of BOAT is that it considers the informativeness of each component word in the concept labels, which has significant impact on biomedical ontologies, resulting in a 12.2% increase in F-measure. Another important feature of BOAT is that it selects for comparison only concept pairs that show high likelihoods of equivalence, based on the similarity of their annotations. BOAT’s F-measure of 0.88 for the alignment of the mouse and human anatomy ontologies is on par with that of another state-of-the-art matcher, AgreementMaker, while taking a shorter time.