Text Classification Data from 15 Drug Class Review SLR Studies

These datasets are derived from the data provided and originally used by: Cohen A.M., Hersh W.R., Peterson K., Yen P.-Y. (2006): Reducing workload in systematic review preparation using automated citation classification. Journal of the American Medical Informatics Association, 13: 206–219. The original data file is available from https://dmice.ohsu.edu/cohenaa/systematic-drug-class-review-data.html and contains PubMed IDs of articles labeled as relevant or irrelevant to 15 study topics, based on human evaluation of abstracts and full texts. The derived data provided here contains 15 datasets corresponding to these study topics and contain: article PubMed IDs, abstracts retrieved from MEDLINE-exported files for the TREC 2004 Genomics Track, available from https://dmice.ohsu.edu/trec-gen/2004data.html, relevant/irrelevant class labels based on human abstract evaluation. Articles for which no matching abstracts could be found are ommitted. The size and class distribution of the datasets differ from those presented by Cohen et al. (2006) due to possible changes of the MEDLINE records or differences in the matching procedure.