Abstract de la publi numéro 16263

Scholarly publications reify fruitful collaborations between co-authors. A branch of research in the Science Studies focuses on analyzing the co-authorship networks of established scientists. Such studies tell us about how their collaborations developed through their careers. This paper updates previous work by reporting a transversal and a longitudinal studies spanning the lifelong careers of a cohort of researchers from the DBLP bibliographic database. We mined 3,860 researchers' publication records to study the evolution patterns of their co-authorships. Two features of co-authors were considered: 1) their expertise, and 2) the history of their partnerships with the sampled researchers. Our findings reveal the ephemeral nature of most collaborations: 70% of the new co-authors were only one-shot partners since they did not appear to collaborate on any further publications. Overall, researchers consistently extended their co-authorships 1) by steadily enrolling beginning researchers (i.e., people who had never published before), and 2) by increasingly working with confirmed researchers with whom they already collaborated.