Bibliometric indices for the assessment of the citation curve tail

Abstract

Scientometrics is an area recently attracting greater research interest. To a great extent this area has been fertilized by the proposal of the h-index (2005), which represents a measure for the quality and quantity of a researcher's impact. The Perfectionism Index has been recently proposed aiming at differentiating between 'influentials' and 'mass producers' (2015); the former category produces articles, which are (almost all) with high impact, whereas the latter category produces a lot of articles with moderate or no impact at all. In this paper, we record a number of metrics that are of similar nature, i.e. they shed light into these publishing patterns (influentials vs. mass producers). We carry out a correlation analysis to reveal which metrics are describing the above phenomenon in a similar way and, thus, retain the most descriptive features. Finally, we report the results of an experiment with a dataset consisting of the academic staff of Greek Computer Science/Engineering departments.

Gazing at the skyline for star scientists

“…Including Hirsch's a parameter (which is essentially the metric1 from Table 1) we can safely deduce that this group also expresses information about the tail of the citation curve. The tail-related metrics are found to penalize the citation curves that are highly skewed and have a long and thick tail (Gogoglou et al, 2015). These are the indices constituting cluster1.…”

“…In Figs. 2 and 3 we present the 2D skyline set for the years 1995 and 2005 respectively, with dimensions corresponding to the h-index and the Perfectionism index, PI (Gogoglou, Sidiropoulos, Katsaros, & Manolopoulos, 2015;Sidiropoulos et al, 2015). Perfectionism index is a recently proposed indicator which aims at differentiating among laconic and mass-producing scientists.…”

“…By choosing one index from each cluster, we ensure that all different features of scientific performance are represented as attributes of the skyline operator. The first combination we contemplated is the set of the three cluster representatives (see Table 4) comprised by R (Jin et al, 2007), h-rationalized (Ruane & Tol, 2008) (hrat) and metric2 (Gogoglou et al, 2015). For the second set, we decided to incorporate a set of the most popular indices that belong to each cluster, namely e-index (Zhang, 2009) (excess index), the h-index and the hnor (Sidiropoulos et al, 2007).…”

Análise de componentes principais em data warehouses

“…Porém, sua eficácia para a classificação de pesquisadores é discutível, já que a curva de citações do pesquisador não é analisada por inteiro. Algumas das desvantagens do índice-H podem ser encontradas nos trabalhos [GSKM15,ACHVH09], bem como alternativas para a caracterização de pesquisadores em relação à sua produção científica.…”

Complex networks and machine learning in scientometrics

“…Perfectionism Index (PI) [Sidiropoulos et al, 2015]: The Perfectionism Index was introduced to account for the tail of the citation curve, meaning the publications that received less than h citations. It is defined as:…”