You are here

Link prediction in a MeSH co-occurrence network: preliminary results.

Printer-friendly versionPrinter-friendly version
Kastrin A, Rindflesch TC, Hristovski D
Stud Health Technol Inform. 2014;205:579-83.
Abstract: 

Literature-based discovery (LBD) refers to automatic discovery of implicit relations from the scientific literature. Co-occurrence associations between biomedical concepts are commonly used in LBD. These co-occurrences can be represented as a network that consists of a set of nodes representing concepts and a set of edges representing their relationships (or links). In this paper we propose and evaluate a methodology for link prediction of implicit connections in a network of co-occurring Medical Subject Headings (MeSH®). The proposed approach is complementary to, and may augment, existing LBD methods. Link prediction was performed using Jaccard and Adamic-Adar similarity measures. The preliminary results showed high prediction performance, with area under the ROC curve of 0.78 and 0.82 for the two similarity measures, respectively.

Kastrin A, Rindflesch TC, Hristovski D. Link prediction in a MeSH co-occurrence network: preliminary results. Stud Health Technol Inform. 2014;205:579-83.