The purpose of this study is to map semiconductor literature by author co-citation analysis in order to highlight major subject
specializations in semiconductors and identify authors and their relationships within these specialties and within the field.
Forty-six of the most productive authors were included in the sample list. Author samples were gathered from the INSPEC database
from 1978 to 1997. The relatively low author co-citation frequencies indicate that there is a low connection among authors
who publish in semiconductor journals and big differences among authors' research areas. Six sets of authors with co-citation
greater than 100 times are M. Cardona and G. Lucovsky; T. Ito and K. Kobayashi; M. Cardona and G. Abstreiter; A. Y. Cho and
H. Morkoc; C. R. Abernathy and W. S. Hobson; H. Morkoc and I. Akasaki. The Pearson correlation coefficient of author co-citation
varies widely, i.e., from -0.17 to 0.92. This shows that some authors with high positive correlations are related in certain
ways and co-cited, while other authors with high negative correlations may be rarely or never related and co-cited. Cluster
analysis and multi-dimensional scaling are employed to create two-dimensional maps of author relationships in the cross-citation
networks. It is found that the authors fall fairly clearly into three clusters. The first cluster covers authors in physics
and its applications. The authors in the second group are experts in electrical and electronic engineering. The third group
includes specialists in materials science. Because of its interdisciplinary nature and diverse subjects, semiconductor literature
lacks a strong group of core authors. The field consists of several specialties around a weak center.
program and further transformed into an original cross-citation matrix as in Table 3 . In the table, rows represent citing authors and columns represent cited authors, the diagonal represents self-citationfrequency. From citationfrequency analysis, self-citation
.g. Scopus or Google Scholar) one would obtain for the same journals other values. Yes, it is evident that this is so. And this is not a drawback of the TRIF—another data set, other citationfrequencies, other IF values (though these values would be quite
represents the overall citationfrequency of journals. The temporal distribution of journals with citations of over 100 times corresponds to three periods: (1) before 1995, (2) 1995–2000, (3) 2001–2005. The journals with most cited frequencies belong to the
Authors:A. Pudovkin, H. Kretschmer, J. Stegmann, and E. Garfield
impact factors (IF) ranging from 0.084 to 47.400, the median and quartiles being 4.226, 2.058, and 6.956. Citationfrequencies range from 0 to 229, the median and quartiles being 9, 2, and 26.
. The basic idea of the present paper fits into this framework. The mathematical nature of the h -index was illuminated by Glänzel ( 2006 ); suggesting also an embarrassingly simple relation between the h -index and—in the original citationfrequency
resources; (iii) cited patents are of equal technological importance, and (iv) cited patents have the same citationfrequency. In reality, these conditions rarely hold: many of the citations at the USPTO are provided by applicants who relate to the firm or
number of referenced journals in each percentile, each corresponding author was isolated from others and the following steps were taken:
We created a list of the author's citing journals with their citationfrequency.
We ranked the