A quantitative measure to compare the disciplinary profiles of research systems and their evolution over time

By modeling research systems as complex systems we generalize similarity measures used in the literature during the last two decades. We propose to use the mathematical tools developed within the spin-glasses literature to evaluate similarity within systems and between systems in a unified manner. Our measure is based on the ‘overlap’ of disciplinary profiles of a set of research systems and can readily be integrated in the framework of traditional bibliometric profile analysis. The investigation of the distribution of the overlaps provides useful insights on the dynamics of the general system, that is whether it converges toward a unique disciplinary structure or to a differentiated pattern.