A rational indicator of scientific creativity

A model is proposed for the creation and transmission of scientific knowledge, based on the network of citations among research articles. The model allows to assign to each article a non-negative value for its creativity, i.e. its creation of new knowledge. If the entire publication network is truncated to the first neighbors of an article (the n references that it makes and the m citations that it receives), its creativity value becomes a simple function of n and m. After splitting the creativity of each article among its authors, the cumulative creativity of an author is then proposed as an indicator of her or his merit of research. In contrast with other merit indicators, this creativity index yields similar values for the top scientists in two very different areas (life sciences and physics), thus offering good promise for interdisciplinary analyses.