Echeverria, MercedesStuart, DavidBlanke, Tobias2015-03-112015-03-112015-03-12978-93-81232-05-7http://hdl.handle.net/1944/1856This paper reports the results obtained on the predictability of references for the identification of derivative articles from doctoral theses, based on a sample of 68 medical theses and 334 articles published by the same theses authors. The study performs an analysis of the common references shared by theses and articles through a text similarity approach. A textual similarity comparison is carried out with the discursive sections of articles (Introduction, Methodology, Results and Discussion) based on the full-text of theses and articles. The results suggest that the Reference section has a high sensitivity to detect true positives cases and a low specificity to identify negative cases, corresponding to a high recall a low precision in the detection of derivative articles.en-USDerivative ArticlesDoctoral ThesesCluster Analysis MethodologyArticle