Published: January 2014
Type: Research Article
Mery Piedad Zamudio Igami*, José Carlos Bressiani, Rogerio Mugnaini1
Department of Library, Nuclear and Energy Research Institute, (IPEN–CNEN/SP), Superintendence, Nuclear and Energy Research Institute, (IPEN-CNEN/SP), São Paulo, Brazil,
1Department of Librarianship and Documentation School of Communication and Arts (ECA), University of São Paulo, São Paulo – SP, Brazil
A thesis defense should be considered as not the end but the starting point for scientific communication flow. How many articles truly extend doctoral research? This article proposes a new model to automatically identify the productivity of theses in terms of article publications. We evaluate the use of the co‑word analysis technique to establish relationships among 401 doctoral theses and 2,211 articles journal articles published by students in a graduate program at a Brazilian National Nuclear Research Institution (IPEN‑CNEN/SP).To identify the relationship between a thesis and an article published by the same author, we used co‑descriptor pairs from a controlled vocabulary. To validate the proposed model, a survey was applied to a random sample of theses authors (n = 128, response rate of 79%), thus establishing a minimum threshold of three coincident co‑descriptors to identify the relationship between theses and articles. The agreement level between an author’s opinion and the automatic method was 86.9%, with a sampling error of 7.36%, which indicates an acceptable level of accuracy. Differences between the related or nonrelated distributions of articles were also demonstrated, as was a reduction in the median lag time to publication and the supervisor’s influence on student productivity.