Start Date: | 4/27/2015 | Start Time: | 3:00 PM |
End Date: | 4/27/2015 | End Time: | 4:00 PM |
|
Event Description Alexander Yong, University of Illinois at Urbana-Champaign
Abstract: The h-index is a widely discussed metric for evaluating the productivity of academics. It was introduced by the physicist J.E. Hirsch in 2005 as an easy and useful supplement to the raw citation count. It now appears, for example, in Google Scholar and Web of Science profiles. However, there has been significant controversy about its use: this sits within a broader debate about numerics in job, promotion and award decisions.
We suggest a simple model for the h-index that assumes a citation profile behaves like an integer partition chosen uniformly at random. We thereby reinterpret a result of E.R. Canfield-S. Corteel-C. Savage, about the Durfee square of a partition, as the rule-of-thumb:
h-index ~ 0.54 x sqrt(citations)
We discuss evidence for and against the model as well as competing interpretations of the approximation. We compute additional predictions of the model using textbook generating series techniques. The key result used is a classic identity of L. Euler and C.F. Gauss from combinatorial number theory. |
|
Location: Korman Center, Room 245, 15 South 33rd Street, Philadelphia, PA 19104 |
|