Suggesting biomedical topics for unseen research articles based on MeSH descriptors

Abstract

Due to the huge number of research articles in the biomedical domain, it becomes more and more important to develop methods to find relevant articles of our specific research interests. Keyword extraction is a useful method to find important topics from documents and summarize their major information. Unfortunately, it is hard to select appropriate keywords extracted by traditional method of keyword extraction from specific research fields such as biomedical domain. Although human experts can support to understand details of the keywords, extra time should be required to read contents of the documents. In this paper, we propose a method for suggesting keyword-based topics for unseen biomedical research articles from PubMed. Our method uses MeSH descriptors to summarize each document by obtaining frequencies of them. The list of frequencies is used to make keyword suggestions for given documents based on the MeSH. In the experiments, we evaluate the performance of the method by measuring the accuracy of keyword suggestions for a given set of unseen documents.

Automated Recommendation of Research Keywords from PubMed That Suggest the Molecular Mechanism Associated with Biomarker Metabolites

“…MeSH terms have been used to develop informatic methods (i.e., semantic similarity, MeSH-Gram [8,9]), to visualize research trends (i.e., hierarchical structure, MeSH Sim [10]), and to estimate relationships among terms (i.e., establishment of disease-related MeSH terms [11]). Moreover, methods have been reported for suggesting keyword-based topics for unseen biomedical research articles from PubMed [12] and for annotation of scientific data with keywords from a controlled vocabulary [13]. However, an automated method to find MeSH terms associated with a biological molecule and the existing knowledge remains to be explored.…”