Home // DBKDA 2014, The Sixth International Conference on Advances in Databases, Knowledge, and Data Applications // View article

Learning Links in MeSH Co-occurrence Network: Preliminary Results

Authors:
Andrej Kastrin
Dimitar Hristovski

Keywords: network analysis, link prediction, literature-based discovery

Abstract:
Literature-based discovery (LBD) is focusing on automatically generating scientific hypotheses by uncovering hidden, previously unknown relations between existing knowledge. Co-occurrences between biomedical concepts can be represented by a network that consists of a set of nodes representing concepts and a set of edges representing their relationships. In this work we propose a method for link prediction of implicit connections between Medical Subject Headings (MeSH®) descriptors. Our approach is complementary to standard LBD. Link prediction was performed using Jaccard and Adamic-Adar similarity measures. Preliminary results showed high prediction performance with area under the ROC curve of 0.78 and 0.82 for Jaccard and Adamic-Adar coefficient, respectively.

Pages: 161 to 164

Copyright: Copyright (c) IARIA, 2014

Publication date: April 20, 2014

Published in: conference

ISSN: 2308-4332

ISBN: 978-1-61208-334-6

Location: Chamonix, France

Dates: from April 20, 2014 to April 24, 2014