SearcharxivSearch

arXiv subjects

Branislav Dragovic

Publications and source records attributed to Branislav Dragovic.

2 recordsLinked to original sources

How to estimate the total number of citations of a researcher using his h index and his h core?

So far, many researchers have investigated the following question: Given total number of citations, what is the estimated range of the h index? Here we consider the converse question. Namely, the aim of this paper is to estimate the total number of citations of a researcher using only his h index, his h core and perhaps a relatively small number of his citations from the tail. For these purposes, we use the asymptotic formula for the mode size of the Durfee square when n tends to infinity, which was proved by Canfield, Corteel and Savage (1998), seven years before Hirsch (2005) defined the h index. This formula confirms the asymptotic normality of the Hirsch citation h index. Using this asymptotic formula, in Section 4 we propose five? estimates of a total number of citations of a researcher using his h index and his h core. These estimates are refined mainly using small additional citations from the h tail of a researcher. Related numerous computational results are given in Section 5. Notice that the relative errors delta(B) of the estimate B of a total number of citations of a researcher are surprisingly close to zero for E. Garfield, H.D. White (Table 2), G. Andrews, L. Leydesdorf and C.D. Savage (Table 5).

math.CO

Extensions of Egghe g-index: Improvements of Hirsch h-index

A few new indices to characterize the scientific output of scientists are defined in the paper. These indices are compared with -index and its alternative indices using some proven assertions. The gd-indices are introduced as extensions of the g-index to define H-index as an improvement of the h-index. Numerous computational results which are conducted indicate the good behaviour of defined indices to evaluate the scientific impact of scientists. There exist very good approximations between some pairs of all considered indices in the sense that related ratios are often very close to 1.

cs.DL