Classification of topics-related documents. In this classification task, the algorithm ranked documents with a given MeSH term annotation with respect to random documents using the references cited in the document and present in the full document or its sections. The median ROC area obtained when training only with the document of interest (S, the baseline) is indicated (dotted line). T: total (or full document), I: introduction, M: methods, R: results, D: discussion, C: conclusion, and I+D: introduction and discussion, ROC: receiver operating characteristic.
Ortuño et al. BMC Bioinformatics 2013 14:113 doi:10.1186/1471-2105-14-113