Science of Science Amin Mazloumian Chair of Sociology: in particular of Modeling and Simulation Chair of Sociology, in particular of Modeling and Simulation
Scientometrics: The science of measuring science Derek J. de Solla Price The pattern of bibliographic references indicates the nature of the scientific research front. Science (1965) 1922-1983 2
Scientometrics: The science of measuring science Derek J. de Solla Price A mathematical theory of Cumulative- Advantage: Success breeds success 3
Scientometrics: The science of measuring science Quantitative studies of the network of citations between scientific papers Derek J. de Solla Price The discovery of power-law distributions in citation networks A mathematical theory of preferential attachment 1922-1983 4
Citations Citations are mainly a measure of usability. Derek J. de Solla Price Paper and citations counts are the official currency in science and are widely used to assess the productivity and impact of authors, institutions, and scientific fields. 1922-1983 5
Distribution of citations for papers N(x) number of papers with at least x citations Redner, EPJ B (1998) 6
A map of science based on citation flows M Rosvall & CT Bergstrom, PNAS (2008) 7
Citation Counts Measures for individual publications: Total number of citations Average annual citations Half-life Maximum annual citations M.Amin, M.A. Mabe, Impact factors: use and abuse, Medicina 2003 8
Citation Counts Measures for individual publications: Total number of citations Average annual citations Half-life Maximum annual citations M.Amin, M.A. Mabe, Impact factors: use and abuse, Medicina 2003 9
Citation Counts Measures for individual scientists: Total number of citation Average number of citations per year Average number of citations per publication h-index m-index 10
Citation Counts: Hirsch Total number of papers: Advantage: measures productivity. Disadvantage: does not measure importance or impact of papers. 11
Citation Counts: Hirsch Total number of citations Advantage: measures total impact. Disadvantages: - hard to find - may be inflated by a small number of big hits 12
Citation Counts: Hirsch Citation per paper Advantage: allows comparison of scientists of different ages Disadvantage: - hard to find - rewards low productivity, and penalizes high productivity. 13
Citation Counts: Hirsch Number of significant papers Advantage: gives an idea of broad and sustained impact Disadvantage: its threshold is arbitrary and randomly favor or disfavor individuals 14
Citation Counts: Hirsch Number of citations of q most cited papers Advantage: gives an idea of broad and sustained impact Disadvantage: -q is arbitrary and randomly favor or disfavor individuals -difficult to compare 15
h-index A scientist has index h if h of his or her N p papers have at least h citations each, and the other (N p h) papers have no more than h citations each. [J.E. Hirsch, PNAS 2005] 16
h-index Example 1 2 1 0 17
Citation Counts: h-index Example 2 2000 2 1 18
h-index Example 3 10 8 7 7 6 6 5 1 19
h-index Hirsch, PNAS (2005) 20
h-index Assume a scientist publishes p papers each year and his papers receive c citations per year. 21
h-index Total life time of his papers after n year: ½ n (n+1) p Total number of citations after n years: ½ n (n+1) pc 22
h-index Assuming: papers up to year y contribute to h-index a) (n-y) c = h b) p y = h 23
h-index and m-index h-index grows linearly with time. 24
h-index and m-index 25
h-index Variants of h-index can be calculated for: - Journals - Universities - Countries 26
Citation boosts 27
Citation boosts 28
Citation boosts 29
Predicting future success Are citation counts reliable predictors of future scientific success? 30
Predicting future success 31
Predicting future success 32
Predicting future success By analyzing careers of 150,000 scientists: 1- Among all citation indicators, the annual citations at the time of prediction is the best predictor of future citations. 2- Future citations of scientists published papers can be predicted accurately. 3- Future citations of future work are hardly predictable. 33
34
Global Multi-level Analysis of Scientific Food Web Cij citations from location i to location j Cj total citations to location j Ri total citation from location i Pi papers published by location i 35
Global Multi-level Analysis of Scientific Food Web If the references listed in papers of entity i would cite the papers published by entity j in a proportional way, the expected number of citations from i to j would be 36
Global Multi-level Analysis of Scientific Food Web Excess citations per reference from i to j 37
Global Multi-level Analysis of Scientific Food Web Knowledge flow from j to i (removing self citations) 38
39
Global Multi-level Analysis of Scientific Food Web Scientific fitness: the sum of excess knowledge flows 40
Global Multi-level Analysis of Scientific Food Web 41
Global Multi-level Analysis of Scientific Food Web 42
Global Multi-level Analysis of Scientific Food Web 43