Measuring scientific impact : can citation analysis correctly assess science?

by Prof. Filippo Radicchi (Institute for Scientific Interchange (ISI), Torino, Italy)

Friday 5 Feb 2010, 11:00 → 12:00 Europe/Rome

Aula Corbino (Dip. di Fisica - Edificio E. Fermi)

Seminario organizzato dall'Istituto dei Sistemi Complessi Citation analysis has acquired a central role in modern science for the quantitative evaluation of the scientific relevance of papers, scientists and research institutions. Numerical indicators based on citations are in fact nowadays often used for the evaluation of candidates in scientific contests like those for the assignment of institutional positions, grants and awards. I present two different studies in this context. First, I present a statistical analysis of the distributions of citations received by single publications. The number of citations received by papers strongly depend on their subject. Simple citation counts do not allow fair comparisons between papers about different topics. Here I provide the solution to this problem. When relative citation indicators are used instead of pure citation counts, the possibility to find a paper with a given value of the relative indicator is the same independently of the particular discipline considered. Second, I introduce a new method for the quantitative evaluation of scientists along time. The method is a graph based ranking algorithm.Instead of using only local information as in simple citation count procedures, the method makes use of the whole citation network simulating a diffusion process which mimics the spreading of "scientific credits". The new method seems to be more reliable than standard ones since it has much greater ability to predict the assignment of Nobel Prizes. A website where the algorithm is made available can be found at the address http://www.physauthorsrank.org.