Celebrating Scholarly

Communication Studies
A Festschrift for Olle Persson at his 60th Birthday

special volume of the e-newsletter of the

international society for scientometrics and informetrics

vol. 05-S June 2009

Editorial Board
co-editors of special volume: FREDRIK ÅSTRÖM
RICKARD DANELL
BIRGER LARSEN
JESPER WIBORG SCHNEIDER
technical editors: FREDRIK ÅSTRÖM
BALÁZS SCHLEMMER

Published by ISSI
Celebrating Scholarly Communication Studies
A Festschrift for Olle Persson at his 60th Birthday

Special volume of the e-zine of the ISSI, vol. 05-S June 2009

© 2009, International Society for Scientometrics and Informetrics

Cover: Balázs Schlemmer

Technical editing: Fredrik Åström & Balázs Schlemmer

2
Contents

Foreword.......................................................................................................................5

Articles
How to use Bibexcel for
various types of bibliometric analysis .............................................................9
The Use of Bibliometric Techniques in
Evaluating Social Sciences and Humanities................................................25
Persson’s universe of bibliometrics
– Has his mapping changed the discipline? .................................................39
The most influential editorials...............................................................................47
Publication patterns in all fields............................................................................55
A Webometric Analysis of Olle Persson ...............................................................61
Pennants for Strindberg and Persson.....................................................................71

The Bibliography of Professor Olle Persson ................................................ 85

Addendum ................................................................................................................. 89

3
Foreword

The origin of this Festschrift could be traced to a few March days in

Hundested, Denmark, in 2001. The event taking place in this small fishing
town in the northern parts of Sealand was a PhD course in bibliometrics
organized by Professor Olle Persson together with professors Peter Ingwersen
and Irene Wormell. Apart from being an excellent course, the Hundested event
was also important for bringing together what could be seen as laying the
foundation for ‘the second generation’ of Nordic bibliometrics. As a result, we
could see a similar course given in Umeå in 2008, where the core group of
teachers were participants of the original course; and also, ending up being the
editors of this Festschrift in honour of Olle Persson on his 60th birthday.
Since the 1970s, Olle Persson has been one of the pioneers in Nordic library
and information science (LIS) research, analyzing information retrieval,
information behaviour and scholarly communication; and of course, the line of
research for which he is most closely associated with nowadays: bibliometrics.
However, it is not only through his own research that Olle has become a
central figure in Nordic LIS, he has also been instrumental in establishing LIS
as an academic field of research, where maybe the most tangible effect is the
founding of the Nordic Research School in Library and Information Science
(NORSLIS) on professor Persson’s initiative. And obviously, we must also take
into account the number of PhD students Olle has seen through their
education as academic supervisor/advisor both in sociology and LIS.
As we all know, Olle Perssons impact is not just limited to Nordic academia
and research. Professor Persson has a long list of international publications,
many of them highly cited, and it is a list that continues to grow; and in
addition to the publishing activities, Olle is also a respected member of the
research community, performing duties on editorial boards and program
committees as well as e.g. organizing the 10th ISSI conference in Stockholm in
2005.
One particular aspect of Olle Perssons work that cannot go unmentioned is
the Bibexcel toolbox for bibliometricians (see also the contributions by Meyer
& Glänzel, and Persson, Danell & Schneider in this Festschrift): a software of
huge impact for many of us doing bibliometric analyses; but also something
that reflects both Olle’s flexible mind and generous nature through how the
software has been developed over the years – many times by Olle adding
functionalities at the request of other users. Apart from Olle Perssons strengths

5
as a scholar and a teacher, we should also mention some of those personal
traits making Olle not only a good academic, but also a well esteemed colleague
and a good friend. On one hand, we have that relentless energy and curiosity
that never cease to amaze us, pushing Olle’s activities forward as well as those
by us being fortunate enough to work with him. On the other hand, we also
have the strong drive from Olle to do things for other people.

The festschrift
A few words on the contributions in celebration of Olle Persson. A rare feature
in a Festschrift is an article by the one being celebrated. However, when this
volume started to take shape, one idea we felt was important was to include a
formal description of the Bibexcel software, something that has been missing
for a long time; and, something making it easier for us to acknowledge our
intellectual debt to Olle by having a document to cite when having used
Bibexcel. But, instead of having Olle write it himself, as a birthday gift to him,
we decided to write it, but obviously, have him being first author.
The international and cooperative nature of research being both analyzed and
appreciated by Olle, is reflected in this Festschrift bringing together
contributions from Spain, the UK, Belgium, Norway and the US, as well as
Sweden, Denmark and – to some extent – Australia. The Bibexcel paper is the
obvious starting point, followed an investigation of research evaluation in the
fields of the humanities and social sciences by Isabel Iribarren-Maestro, María
Luisa Lascurain-Sánchez and Elias Sanz-Casado. In the second paper, Martin
Meyer and Wolfgang Glänzel analyze the impact of professor Persson’s
software Bibexcel on the mapping of research fields.
Ronald Rousseau addresses a new topic in his pilot study on the impact of
highly influential editorials in scientific journals, identifying problems related to
the definition of what an editorial is, as well as how they are indexed in the
Web of Science databases. The issue of research evaluation is revisited by
Gunnar Sivertsen, also addressing problems related to how the social sciences
and humanities can be assessed using bibliometric methods; and how that
relates to differences in publication patterns in different research fields.
As well as with Meyer and Glänzel’s article, Mike Thelwall address the impact
of Olle Persson, not by looking at his scholarly articles but by looking at his
online activities, such as the Inforsk Research Group website and the Bibexcel
software. In the last article, Howard White combines relevance theory from
linguistic pragmatics and ideas from informetrics and information retrieval for
producing and interpreting pennant diagrams; and as an example, uses co-
citation analyses on Olle Persson and August Strindberg.

6
Dear Olle!
With this Festschrift, we want to celebrate your 60th birthday, to show our
appreciation for you as colleague and friend, as well as mentor and teacher. We
are many that have you to thank for a lot of what we are doing nowadays,
something that is reflected not only in us wanting to present you with this
volume, but also through the impact evident in the articles analyzing your work
here.

Happy 60th Birthday!

June 15, 2009
Aalborg, Copenhagen, Umeå and Sydney

FREDRIK ÅSTRÖM
Lund University Libraries,
Head Office, P.O. Box 134, SE-22100 Lund (Sweden)
University of Technology Sydney, Faculty of Arts and Social Sciences,
P.O. Box 123, Broadway, NSW 2007 (Australia)
E-mail: fredrik.astrom@lub.lu.se
RICKARD DANELL
Umeå University, Dept. of Sociology, SE-90187 Umeå (Sweden)
E-mail: rickard.danell@soc.umu.se
BIRGER LARSEN
Royal School of Library and Information Science,
Birketinget 6, DK-2300 Copenhagen S (Denmark)
E-mail: blar@db.dk
JESPER WIBORG SCHNEIDER
Royal School of Library and Information Science,
Fredrik Bajers Vej 7K, DK- 9220 Aalborg Ø (Denmark)
E-mail: jws@db.dk

7
How to use Bibexcel for various types of bibliometric
analysis

Olle Persson1, Rickard Danell1 and Jesper Wiborg Schneider2

1 Department of Sociology, Umeå University, SE-90187 Umeå, Sweden
2 Royal School of Library and Information Science, DK-9220, Aalborg

Introduction
Bibexcel is a versatile bibliometric toolbox developed by Olle Persson. In
Bibexcel it is possible to do most types of bibliometric analysis, and Bibexcel
allows easy interaction with other software, e.g. Pajek, Excel, SPSS, etc. The
program offers the user high degree of flexibility in both data management and
analysis and this flexibility is one of the program's real strengths. It is, for
example, possible to use other data sources than Web of Science, and Bibexcel
can in fact deal with data other than bibliographic records. If the user only
learns the basic file structures that Bibexcel requires it possible to import many
different types of data. However, flexibility has its price and the flexibility may
initially cause new users to perceive it as difficult to use. We therefore find this
festschrift an appropriate forum to describe Bibexcel’s basic functions, and we
will describe these functions by analyzing data consisting of Web of Science
articles that cite Olle Persons scientific publications.
The chapter is structured in four main sections. In the first section we
describe how data downloaded from Web of Science must be restructured. In
the second section we will take a closer look at the OUT-file. Bibexcel
produces several types of files; in fact every procedure will give the user a new
file. However, the OUT-file is always created first, and it is this file that is the
starting point for the analysis you want to do in Bibexcel. In the third section
we will give a brief description of basic analytical functions available in
Bibexcel. In the last section we will describe how to export files to Pajek in
order to do visualizations. The aim of this chapter is to introduce readers to
how to use Bibexcel. We will assume that the readers have some basic
knowledge of bibliographic data and basic bibliometric techniques. It should
also be noted that Bibexcel includes far more features than described in this
chapter. However, it is our hope that our basic description will make it possible
for the interested reader to acquire enough knowledge to start using Bibexcel,
and after some experimentation be sufficient self reliant to figure out how to
use functions not described in this chapter.

9
How to prepare and import data
As stated above, it is possible import many different data formats in Bibexcel,
but in this section we will only to comment on how to import data from Web
of Science. The data imported from Web of Science should be saved as plain
text. The plain text file downloaded from Web of Science needs to be
restructured before it can be imported into Bibexcel. The restructuration of
data consists of two simple steps. First, it is necessary to insert carriage return
in the text file we want to import in Bibexcel. Carriage return can be inserted in
the file in two different ways; either we open the file in word and re-save it as a
text file, or we select from Bibexcels menu:
Edit doc file->Replace line feed with carriage return
The next necessary restructuring of the file downloaded from the Web of
Science is to convert the bibliographic records to a DIALOG format. We do
this by first selecting the file with the extension *.tx2 and then choose the
following option from the menu:
Misc->Convert to dialog format->Convert from Web of Science
The procedure tells Bibexcel to create a file with the extension *. doc (we
henceforth refer to this as DOC file). After completing these steps the text file
we downloaded from the Web of Science is ready to be used in Bibexcel and
we can start analyze the data. However, before we start, we may want to
familiarize our self with the structure of the DOC-file. At least it is necessary to
be familiar with the structure of the bibliographic records in the DOC-file.
Bibexcel keeps track of where the bibliographic records begins and ends by
looking for a double-spike, that is | |. Each record is composed of several
bibliographic fields and Bibexcel keeps track of where the bibliographic fields
begins by field tags. For example, the field tag for the author field is "AU". It is
important to keep track of field tags, because we usually have to tell Bibexcel
which bibliographic fields we want to work with. Each bibliographic field ends
with a single spike, i.e. |. In bibliographic fields with multiple units, the units
are separated from each other with some delimiter. For most bibliographic
fields the field delimiter is a semicolon. However, there are other delimiters and
it is necessary to tell Bibexcel how the bibliographic field is delimited.

Creating an OUT-file and calculating frequency distributions

We made the file containing the bibliographic records readable in Bibexcel, and
we are now ready to make an OUT file. Making an OUT-file is always the first
step when analyzing bibliographic data with Bibexcel. When we make the
OUT-file we start by decide which bibliographic field the OUT-file will be
constructed from. We tell Bibexcel which bibliographic field the OUT-file will
be constructed from by entering the field tag in the box marked "Old tag", e.g.

10
if we want the OUT file to be based on the author field we write AU in "Old
tag".
Next we select the DOC-file from the file managing system by clicking on it.
We must know how the field is delimited, and we select an option for how the
units in the field are delimited from the scrollbar marked “Select field to be
analyzed”. In our example we will make an OUT-file based on the author field
and in this field each co-authors are separated from each other with a
semicolon. We therefore select the option “Any ; separated field”.
After we selected the DOC-file, and selected field delimiter, and typed AU in
the box marked “Old tag” we press the button labeled "Prep." Bibexcel now
creates an OUT-file based on the author field, and each line in the OUT-file
will be matched by a unique authorship as a unique author holds. The OUT-
file's structure is simple and it is important that we familiarize ourselves with it.
The OUT-file we produced has the following structure (Table 1):
Table 1 The structure of the OUT-file
Document identification number Authors
1 Levitt JM
1 Thelwall M
2 Hsu PY
2 Shiau WL
2 Su YM
2 Yang SC
3 Anegon FD
3 Guerrero-Bote VP
3 Olmeda-Gomez C
3 Ovalle-Perandones MA
3 Perianes-Rodriguez A
…. ….
…. ….
388 Stefaniak B
389 Vlachy J
390 Ellis D

The left column of the OUT-file consists of a document identity number, and
the right column the authors. In Table 1 we see that in total there are 390
documents citing Olle Persson’s publications and the first document in the
DOC-file has been co-authored by JM Levitt and M. Thelwall. Regardless of
what bibliographic field we select the OUT-file will have this structure. It can
be of interest to note that the OUT-file is a tab-delimited text file, and like all
Bibexcel files, it can be imported in Excel, or other statistical software.

Frequency distributions
Depending on what bibliographic fields we have chosen as a unit when we
created the OUT-file, the frequency calculation function in Bibexcel offers

11
many different options. For example, if the OUT-file consists of cited
document Bibexcel can make a substring search and only count a specified part
of the cited document, e.g. cited journal or cited author. In Bibexcel, we can
also choose between two counting methods when we ask Bibexcel to count
units in the OUT-file: "whole counts" and "fractional counts". If we chose to
check the box marked "Fractionalize" Bibexcel will change counting methods
to "fractional counts" and if this boxed is left unchecked the counting method
will be "whole counts". The method of fractional counting is easy to
understand. For example, if a document is co-authored by two authors each
author attributed half an article and if the document has three co-authors each
author will be attributed a third of the article, etc.
In our example the OUT-file consists of authors who have cited at least one
of Olle Persson’s publications, and we want to know how many times the
authors have cited Olle Persson. To get Bibexcel to calculate the distribution of
authorships we select the OUT-file from the Bibexcel’s file managing system
(under the caption “Select file here”). Next we must tell Bibexcel whether it
should count “whole strings” (i.e. the whole row in the OUT-file) or some
predefined part of the text string. In our example the rows OUT-file consists of
author names and Bibexcel cannot do a substring search in an author name.
Since we want Bibexcel count the whole author name we select the option
"Whole String" from the scrollbar under the caption "Select type of Unit". If
we want the list of authors sorted descending by frequency, we click in the box
marked "Sort descending" and if we want to change counting method to
"fractional counts" we click in the box marked "Fractionlize". We start the
counting procedure by pressing the button marked “Start”, and Bibexcel
creates a frequency distribution and saves it in a file with extension *.cit.
Table 2 Authors citing at least one of Olle Persson’s publications
Author Whole counts Fractional counts
Persson O 15 7,999
Leydesdorff L 14 9
Glanzel W 13 7,416
Meyer M 13 9,833
Melin G 12 7,916
Zitt M 8 3,332
White HD 7 5,5
Rousseau R 6 2,366
Gomez I 6 2,199
Zuccala A 6 4,5
Cronin B 6 4,333
Moya-Anegon F 5 1,999
Morris SA 5 2,083
Bassecoulard E 5 2,166
Herrero-Solana V 5 1,749

12
Creating a new OUT-file
In some situations, it is necessary redefine the units in the OUT-file and to
make a new OUT-file. If we, for example, are interested in examining which
journals that has been cited, Bibexcel enable us to make a new OUT-file in
which an OUT-file containing cited documents are reduced to a OUT-file
containing cited journals, i.e. Bibexcel removes all references that are not
published in a scientific journal, and keeps only the name of the cited journals
in the new OUT-file. I should be noted, in this procedure Bibexcel assumes
that a cited document with a volume number is published in a journal. This
definition is not perfect and we usually need to do some editing of the new
OUT-file. Alternative options to make a new OUT-file is listed in the scrollbar
under the caption "Select type of unit". That is, in the same scrollbar that we
use to select the type of unit when we created the CIT-file.
To create a new OUT-file we start by selecting the “old” OUT-file in the box
for file management. We continue by selecting a type of unit in the scrollbar
under the caption "Select type of unit". In our example, we will select the
option "Cited journal”. Next we decide if we want to Bibexcel to eliminate
duplicates, i.e. remove identical units with the same document identification
number. Since the units in our new OUT-file will be cited journals we should
consider this option; it is possible that several documents in the reference list
has been published in the same journal. Duplicate units in the OUT-file will
cause problems for some types of co-occurrence analysis, e.g. if we use a MDS
algorithm we do not want loops in the matrix. If we click in the box marked
“Remove duplicates” all units with the same document identification number
will be unique. Next we tell Bibexcel to make a new OUT-file by clicking in the
box marked “Make new out-file” and pressing the button labeled “Start”. The
new OUT-file will have the extension *. oux. The OUX-file has the same
structure as the OUT-file, and we can use the OUX-file in the same way we use
the OUT-file, e.g. we can tell Bibexcel to calculate how many times the journals
has been cited. To illustrate the effect of removing duplicates from the OUX-
file Table 3 displays the citation distributions over journals with duplicates
removed and with duplicates included.
Table 3 Scientific journals usually used when citing Olle Persson publications
No. Citations No. Citations
Journal
(no duplicates) (all citations)
SCIENTOMETRICS 344 2691
J AM SOC INFORM SCI 188 988
RES POLICY 151 472
SCI TECHNOL 110 127
SCIENCE 103 198
SOC STUD SCI 101 217
J INFORM SCI 89 179
INFORM PROCESS MANAG 86 201

13
 No. Citations No. Citations
Journal
(no duplicates) (all citations)
J AM SOC INF SCI TEC 85 328
J DOC 79 246
NATURE 56 110
SCI PUBL POLICY 54 78
ANNU REV INFORM SCI 50 111
RES EVALUAT 49 73
AM SOCIOL REV 46 87
AM PSYCHOL 42 60
AM J SOCIOL 37 64
P NATL ACAD SCI USA 37 63
LIBR TRENDS 37 65

Producing a data matrix for export to a statistical software

A useful feature in Bibexcel is the one that enables us to produce data matrices
for export to statistical software. In Bibexcel a data matrix is created by adding
variables to an OUT-file, and the variables are created by selecting fields in the
bibliographic record. However, there is a pitfall that Bibexcel users should be
aware of. We must remember that the first variable in the matrix, i.e. the unit in
the OUT-file, must be present in all bibliographic records. If the bibliographic
field, chosen as unit when the OUT-file was made, is missing in some records
Bibexcel will not be able to create a data matrix. It thus makes sense to choose
a field that is always available, e.g. the field containing the journal name (SO). If
the units in the OUT-file are present in all bibliographic records, it is no
problem to add fields to the OUT-file that are missing in some bibliographic
records; Bibexcel will write "No field" as compensation for the lost variable
value.
If we have created an OUT-file, that contains all the document identification
numbers, the procedure for creating a data matrix quite simple. We begin by
selecting the OUT-file (or an OUX-file, or some other file with the same
structure as an OUT-file), and enter a field tag in the box marked "Old tag". If
we want to add “times cited” to the OUT-file as a variable we write TC in box
labeled "Old tag". By entering the field tag in the "Old tag" we tell which field
we want Bibexcel to use when it adds a new variable to our data matrix. Next,
we press the button labeled "Add fields to units”. The file created by Bibexcel
has the extension *. jn1. If we want to continue to add variables to our matrix
we select the JN1-file and enter a new field tag in "Old tag" and press "Add
fields units”. This makes Bibexcel creates a file with the extension *. jn2. We
continue by selecting the JN2-file and repeat the procedure and Bibexcel
creates a JN3-file. We continue until the matrix contains sufficient information.
Since JN *- file is a tab-delimited text file, it can be imported into Excel or any
other statistical software for further analysis.

14
Table 4 Example of a data matrix produced by Bibexcel
Document identification number Times cited Publication year
1 0 2009
2 0 2009
3 0 2009
4 0 2009
5 0 2008
6 0 2009
7 0 2008
8 0 2009
9 0 2009
… … …
389 3 1987
390 3 1986

Co-occurrence analysis and the COC-file

The present section illustrates how to perform some of the most essential
bibliometric analyses by use of Bibexcel. The previous Bibexcel menus, File,
Edit doc file, and Edit out-file, have focused upon cleaning and setting up data for
analysis. The menu Analyze presented here contains a number of specialized
functions that allow us to perform analyses of citation networks and, perhaps
most important, a range of different co-occurrence analyses. We cannot
demonstrate all the functions in the Analyze menu. We will therefore focus
upon co-occurrence analysis and how to prepare your data and how to perform
co-occurrence analyses. We demonstrate the intrinsic functions and point out
the small differences in performing co-citation analysis, bibliographic coupling,
co-word analysis, and co-author analysis. Finally, we look at some alternatives
in the Analyze menu that enable some more advanced multivariate analyses.

Co-occurrences
A bibliographic record consists of a number of fields used to index the actual
text, its subjects and descriptive data. As demonstrated above, when working
with Bibexcel we usually transforms our initial data to the Dialog-format, more
specific the format for Science Citation Index®. Common data between records
are thus structured in univocal metadata fields, such as publication titles in the
title field, authors in the author field, and references in the reference field. A
co-occurrence relation in a bibliographic record usually means the mutual
occurrence of two units in the same metadata field. Hence, when words x and y
appear together in the title field, or when authors z and w appear together in the
author field. Obviously, one co-occurrence relation between two units is trivial.
What is interesting, on the other, is whether a co-occurrence relation between
two units is frequent over a number of records, for example that the same title
words x and y appears together in a number of records, or the same pair of
authors z and w also appear together in a number records – this is in principle a

15
co-occurrence analysis. Co-occurrence analysis is therefore the study of mutual
appearances of pairs of units over a consecutive number of bibliographic
records. With this in mind, we will now illustrate how we prepare and perform
co-occurrence analysis in Bibexcel.

Preparing data and files for co-occurrence analysis

Basically, the co-occurrence routine in Bibexcel match pairs of units, extracted
from the same metadata field, within and across the records in the DOC-file. In
order for the routine to match pairs of units, we need to indicate 1) on what file
the matching routine is to be performed; and 2) what units should be matched.
The ‘workhorse’ in Bibexcel is the OUT-file. Most analytical routines are
performed on the OUT-file. Remember that the OUT-file is basically an
extraction of units from one metadata field across the records in the DOC-file.
Therefore, the unit of analysis in the OUT-file defines the type of co-
occurrence analysis. For example, an OUT-file that lists the individual authors
from each record in the DOC-file, would be the basis for a co-author analysis.
The matching routine used to match pairs of units must therefore be
performed on the OUT-file. It is the unit pairs in the individual documents and
their frequency cross all documents that must be generated.
Notice, many individual units will have very low frequencies. Such units are
often unimportant in co-occurrence analyse as their mutual relationships will be
trivial due to low frequencies. It is therefore a very good idea to use individual
frequency as an inclusion criterion for the analysis. Further, such a criterion
also speeds up the generation of co-occurrence pairs, since this can be a
resource demanding routine depending on the number of units to match.
Obviously, other criteria can also be used to reduce the number of units for the
analysis. In order to select the units to be included and matched, we need the
CIT-file. Remember, the CIT-file is a frequency file of the units represented in
the OUT-file. What we need to do is to show the CIT-file in The List, which is
done by marking the CIT-file in Select file here and then press the View file
button.
Let us illustrate the preparation procedure. The first thing to do is to actually
select the units to be included in the analysis. Mark the CIT-file in Select file here
and show the CIT-file in The List by pressing the View file button. Next, we
must mark the units from the CIT-file (shown in The List) that we want to
study in the co-occurrence analysis. Pointing the mouse to one unit will mark it.
To mark several consecutive units, mark the first, then press SHIFT and mark
the last unit; to mark a number of non-consecutive units, press CTRL and the
use the mouse to select units. If you regret the selection of a unit, press CTRL
and use the mouse again on the chosen unit. Notice if you need to view and
select many units from The List press the View whole file and scroll down and

16
select units. Below is an example of a CIT-file that contains frequencies of cited
references (represented by author and publication year):
Table 5 The CIT-file
Frequency Cited reference
99 LUUKKONEN T, 1992
79 KATZ JS, 1997
74 MELIN G, 1996
59 LUUKKONEN T, 1993
56 WHITE HD, 1981
50 PERSSON O, 1994
45 NEWMAN MEJ, 2001
44 BEAVER DD, 1979
42 BEAVER DD, 1978
42 WHITE HD, 1998
99 LUUKKONEN T, 1992
79 KATZ JS, 1997

To select all 10 references mark the first unit, press SHIFT and mark the last
unit. Notice, you mark the whole line, including the frequency number.
We are now ready to commence the actual co-occurrence analysis. This is
done by the following procedure from the Bibexel menu:
Analyze -> Co-occurrence -> Select units via listbox
This routine removes non-selected units from the The List so that only the
ones selected for the co-occurrence analysis are kept.
The next step is to identify co-occurrence relations between the selected
units, i.e., the actual co-occurrence analysis. This step is done in two tempi.
First, we need to indicate what file the matching routine is to be performed on,
as indicated above. This is always the OUT-file. We therefore mark the OUT-
file in Select file here. Notice, we only mark the OUT-file, we do not show it in
The List, where we already have the units selected by the previous operation.
Consequently, we have marked the OUT-file in the Select file here and the
selected units are in The List; only the first 17 are visible. The next move is to
run the matching routine on the OUT-file, which is done by the following
procedure from the Bibexcel menu:
Analyze -> Co-occurrence -> Make pairs via listbox
A question pops up immediately after activating the routine, asking whether
one wishes to include individual frequencies for the units in addition to co-
occurrence frequency in the output. Most often, if the purpose of the analysis is
a mapping of some sort, such frequencies should be left out. The outcome of
the co-occurrence routine is the COC-file (abbreviation for co-occurrence),
examples without and with individual frequencies is shown below.

17
Table 6 The COC-files

COC Cited Cited COC Cited Cited

freq. reference #1 reference #2 freq. reference #1 reference #2
42 KATZ JS MELIN G 42 KATZ JS MELIN G
1997 1996 1997$79 1996$74
42 BEAVER DD BEAVER DD 42 BEAVER DD BEAVER DD
1978 1979 1978$42 1979$44
34 WHITE HD WHITE HD 34 WHITE HD WHITE HD
1981 1998 1981$56 1998$42
33 PERSSON O WHITE HD 33 PERSSON O WHITE HD
1994 1981 1994$50 1981$56
33 BEAVER DD LUUKKONEN T 33 BEAVER DD LUUKKONEN T
1978 1992 1978$42 1992$99
31 BEAVER DD LUUKKONEN T 31 BEAVER DD LUUKKONEN T
1979 1992 1979$44 1992$99
30 KATZ JS LUUKKONEN T 30 KATZ JS LUUKKONEN T
1997 1992 1997$79 1992$99
28 MELIN G NEWMAN MEJ 28 MELIN G NEWMAN MEJ
1996 2001 1996$74 2001$45
Example of COC-file, where individual frequencies are not Example of COC-file, where individual frequencies are
included included

The COC-file contains actual co-occurrence frequencies between the units

chosen for the analysis. The co-occurrences frequencies are based on matching
all possible parings of units in the OUT-file, to establish how many times they
appear together across the records.

Opportunities with COC-file

There are several opportunities from here. The COC-file can very easily be
transformed into an input file for network visualization in Pajek. This is
demonstrated in the next section. The COC-file is also the input for the special
clustering algorithm in Bibexcel, called Persson’s Party Clustering. You simply
mark the COC-file in Select file here and then proceed with:
Analyze -> Co-occurrence -> Cluster pairs
The outcome is a PER-file that contains information about the number of
clusters, their members, and links within and between clusters. Finally, the
COC-file can also be used as the basis for construction of square symmetric
matrices. This is done in two steps. First, we need to indicate which units that
will go into the matrix. Most often this is done by marking the COC-file in
Select file here and then proceeds with:
Analyze -> List units in pairs

18
The outcome is a CCC-file. To proceed with the matrix generation, mark the
CCC-file in the Select file here, and from the Analyze menu choose:
Analyze -> Make a matrix for MDS etc.
You have the choice of generation a square symmetric matrix or its
constituent lower left part. The latter is the input for the MDS SYSTAT
algorithm that is compatible with Bibexcel, but not an integrate part of the
software. If you have access to SYSTAT, continue with the routines Make a
map/SYSTAT cmd file and Show map to produce an MDS mapping of the matrix.
Otherwise export the matrix to Excel, SPSS, UCINET or other softwares
containing MDS routines.

Different co-occurrence analyses

Above we have described the intrinsic procedures of co-occurrence analysis in
Bibexcel when using the Analyze -> Co-occurrence routine. Table x below outlines
different types of co-occurrence analyses, what units should be in the OUT-file
to produce them, how to run them, their outcome files, and some of their
special characteristics.
Table 7 Matrix of co-occurrence analyses
Type of co- OUT-file must be based
Bibexcel menu: Special characteristics
occurrence analysis on this metadata field
Co-citation analysis CR or CD; i.e., the cited Analyze -> Co- The focus is upon mapping
reference string, or part occurrence -> Make relations between cited
of if, such as cited author pairs via listbox references
or cited work. Outcome = COC-file
Bibliographic CR or CD; i.e., the cited Analyze -> Shared The focus is upon mapping
coupling reference string, or part units relations between citing
of if, such as cited author Outcome = COU-file publications (the records in
or cited work. (similar format as the the DOC-file)
COC-file)
Co-author analysis AU or AF, and Analyze -> Co-
occurrence -> Make
pairs via listbox
Outcome = COC-file
Co-word analysis TI (title words) , AB Analyze -> Co- In principle every field, if it
(words from abstract), occurrence -> Make makes sense, can be used;
DE (descriptors); SC pairs via listbox even ‘homemade’ fields can be
Subject categories) etc. Outcome = COC-file made if for example data are
imported from other sources,
see Schneider (2004).

For example, to do a document co-citation analysis, you need to generate an

OUT-file with cited references and then proceed as described above. The same
goes for a co-author analysis, where you generate an OUT-file containing

19
authors extracted for example from the AU field in the DOC-file. Co-word
analysis actually covers a number of co-occurrence analyses, depending on the
units in the OUT-file, for example title words, descriptors, subject categories
etc. – you decide.
The only deviance from the intrinsic procedure is bibliographic coupling.
Bibliographic coupling is called Shared units in Bibexcel and the routine is
performed in the following way. Mark the OUT-file in Select file here and
Analyze -> Shared units
The outcome is a COU-file that contains pairs of documents and their
common share of units (references). Notice that documents are indicated by
the record number. You may want to add labels to these numbers, for detailed
description see the Help file in Bibexcel. The COU-file is in principle the same
as a COC-file and can be used as input for clustering and matrix generation, as
described above. [

More advanced co-co-occurrence analysis: The MA5-file

Sometimes we need to do more elaborate multivariate analyses of co-
occurrence relations. For such purposes, Bibexcel provides an extremely
valuable tool, Docs and units matrices. This routine creates a classical data matrix,
where rows correspond to documents (records) and columns the chosen units
for the study.

Creating Pajek files in Bibexcel

In this section, we shall describe how to create Pajek files in Bibexcel.
However, we will not give a detailed account for the many functions in Pajek.
Pajek is freely available on the Web, and for those interested in this excellent
software should visit Pajek Wiki (http://pajek.imfm.si/doku.php). This
resource contains both the program and a manual.
In Bibexcel, we can create three types of Pajek files: network files (NET-file),
vector files (VEC-file), and files containing information about the cluster
membership (CLU-file). In this section, we shall describe how to produce these
files in Bibexcel and we will use the files to visualize Olle Persson's position in a
document space, or if you like Olle Persson’s intellectual contexts.
If we want to create files for Pajek in Bibexcel we must follow a specific
sequence. We always start by creating the network file (NET-file), and a NET-
file is created from a COC-file. We start by selecting the COC-file, and from
the Bibexcel menu we choose:
To Pajek -> Create net-file from coc-file
Since a network may be a directed or undirected graph, Bibexcel will ask us
“Do your data have directed arcs, like A cites B”. Since the COC-file used in
this example is the result of a co-citation analysis we do not have directed arcs

20
and we answer no to the question. Bibexcel completes the process and creates a
file with the extension *.net, which contains the co-citation network. After we
made the NET-file, we can continue to create VEC-files and CLU-files in any
order. The reason we need to do NET-file first is that the process that creates
NET-file also creates a file with extension *. Vel, which Bibexcel need to make
VEC and CLU files.
To make a VEC file, we use a CIT file. In our visualization of Olle Persson’s
intellectual contexts we also want to show which documents that are most
cited. We will therefore create a vector file that contains information about the
co-cited papers citation frequency. To do this VEC-file we start by selecting the
CIT-file that contains citation frequencies, and then we select:
To Pajek -> Create vec-file
It is possible to use other files than CIT files when we want to make a VEC
file. What is important is that the file has the same structure as CIT-file, i.e. a
tab-delimited text file where the first column consists of some values for the
vector and the second column the name of the nodes included in the NET file.
A third type of Pajekfilen we can create in Bibexcel is the CLU-file, which
contains information about the partitions. You can create partitions in Bibexcel
with an algorithm developed by Olle Persson (Persson’s party cluster).
However, the CLU-file can be based on any type of partition principle, as long
as the file we tell Bibexcel to use has the right structure. In our example, we
shall use Olle Persson own algorithm and for more details of the algorithm see
Persson (1994).
To partition the co-citations matrix with Olle Persson algorithm we select
COC-file and from the menu we choose:
Analyze -> Co-occurances -> Cluster pairs
Bibexcel will create three files, each containing information about the clusters
created from the COC-file. The files has the extension *. pe2, *. pe3, and *. per.
The file we need to make a CLU-file is the PE2-file. We select the PE2-file and
from the menu we choose:
To Pajek -> Create clu-file
Importing NET, VEC and CLU files in Pajek is simple. The NET file we
open as "Networks", the VEC-file we open as "Vectors" and the CLU file we
open as "Partitions". After we have opened the files in Pajek, we choose the
following option from the Pajek menu:
Draw -> Draw-Partion-Vector

21
Figure 1 Olle Persson intellectual context
The map displayed in Figure 1 has been created with Pajek. The co-citation
map shows the context in which Olle Persson scholarly works have been used.
The documents are represented by the first author and publication year. It
should be noted that many of Olle Person’s publications are represented by
other authors, and the most cited of these are Luukkonen T, 1992, Luukkonen
T, 1993 and Melin G, 1996, which is Luukkonen, Persson & Sivertsen (1992),
Luukkonen, Persson & Tijssen (1993), and Melin & Persson (1996). The cluster
algorithm produced six clusters, and we can aggregate them into three main
intellectual themes. In the upper part of the map, we find publications primarily
used in an information science context. Most cited of these is Persson (1994),
which is Olle’s analysis of the intellectual base of Journal of the American Society for
Information Science. It is in this article, which is an author co-citation analysis, that
Olle presents his algorithm for “party clustering”. The article is highly co-cited
with Henry Small’s classic article and several articles by White and McCain. The
main theme of this cluster is obviously co-citations analysis. In the same cluster,
but further down the map, we find Olle Persson article on all author co-citation
analysis, a proposed solution to the first author problem in traditional author
co-citation analysis. Another article by Olle Persson in this cluster is Olle’s
article on Online bibliometrics (Persson 1986). In the middle and the lower left
part of the map we find publications that address issues of research

22
collaboration and science internationalization. A close examination of the co-
cited papers in the smaller cluster in the map's center reveals that the
orientation of these studies is on formation of social networks, while the larger
cluster deals with issues relating to the increasing share of co-authored articles
and internationalization of science. On the map's right side, we find articles
dealing with research evaluation and technology transfer. The map gives us an
insight into the breadth and importance of Olle Persson scientific achievement.

References
Danell, R. and O. Persson (2003). "Regional R&D activities and interactions in
the Swedish Triple Helix." Scientometrics 58(2): 205-218.
Glanzel, W., R. Danell and O. Persson (2003). “The decline of Swedish
neuroscience: Decomposing a bibliometric national science indicator.”
Scientometrics 57(2): 197-213
Luukkonen, T., O. Persson and G. Sivertsen (1992). "Understanding Patterns
of International Scientific Collaboration." Science Technology & Human
Values 17(1): 101-126.
Luukkonen, T., R. J. W. Tijssen and O. Persson (1993). "The Measurement of
International Scientific Collaboration." Scientometrics 28(1): 15-36.
Mahlck, P. and O. Persson (2000). "Socio-bibliometric mapping of intra-
departmental networks." Scientometrics 49(1): 81-91.
Melin, G. and O. Persson (1996). "Studying research collaboration using co-
authorships." Scientometrics 36(3): 363-377
Melin G. and O. Persson (1998) “Hotel cosmopolitan: A bibliometric study of
collaboration at some European universities”, Journal of the American
Society for American Society for Information Science, 49(1): 43-48
Melin, G., R. Danell and O. Persson (2000). "A bibliometric mapping of the
scientific landscape on Taiwan." Issues & Studies 36(5): 61-82.
Meyer, M. and O. Persson (1998). "Nanotechnology - Interdisciplinarity,
patterns of collaboration and differences in application." Scientometrics
42(2): 195-205.
Persson, O. (1986). "Online Bibliometrics – A Research Tool for every Man."
Scientometrics 10(1-2): 69-75.
Persson, O. (1994). "The Intellectual Base and Research Front of JASIS 1986-
1990." Journal of the American Society for Information Science 45(1): 31-
38.
Persson, O. and M. Beckmann (1995). "Locating the Network of Interacting
Authors in Scientific Specialties." Scientometrics 33(3): 351-366.
Persson, O. (2001). "All author citations versus first author citations."
Scientometrics 50(2): 339-344.

23
Persson, O., W. Glanzel and R. Danell (2004). “Inflationary bibliometric values:
The role of scientific collaboration and the need for relative indicators in
evaluative studies.” Scientometrics 60(3): 421-432
Schneider, J.W. (2004). Verification of bibliometric methods' applicability for
thesaurus construction.
PhD dissertation. Aalborg. Department of Information Studies, Royal School
of Library and Information Science, 2004. xiii, 356 p. plus Appendix
volume, 125 p.
___________________________

Addresses of congratulating authors:

RICKARD DANELL
Department of Sociology, Umeå University,
SE-90187 Umeå, Sweden
E-mail: rickard.danell@soc.umu.se
JESPER WIBORG SCHNEIDER
Royal School of Library and Information Science,
DK-9220, Aalborg
E-mail: jws@db.dk

24
The Use
Use of Bibliometric
Bibliometric Techniques
Techniques in Evaluating
Evaluating
Social Sciences
Sciences and Humanities
Humanities
Isabel Iribarren-Maestro, María Luisa Lascurain-Sánchez & Elias Sanz-Casado
Carlos III University of Madrid, C/Madrid, 126. Getafe 28903 Madrid (Spain)

Research in the social sciences and humanities: Evaluation Problems

Research is one of the most important tasks being performed in the world
today. It is difficult, if not impossible, to understand the future of mankind if
divorced from scientific activity. The challenges that human beings face can
only be dealt with through knowledge, technology and innovation. To date,
human activity on the planet has been closely linked to the development and
use of knowledge obtained through the hard sciences. This means that
research, and especially technology, have been defined along fairly narrow lines.
However, new trends more in line with the times, in which sustainability plays
an important role, are causing scientific activities to be undertaken from a more
cross-disciplinary perspective, with the participation of knowledge acquired
from different areas. In this new paradigm, the social sciences and humanities
are contributing a type of knowledge that is more immediate to ever-changing,
more flexible needs and realities.
The active participation of all fields in knowledge-generating processes must
be evaluated, and social science and the humanities are no exception. Such
evaluation may be performed using bibliometric indicators, as is done in other
sciences, given that the only requirement is that researchers must produce
scientific knowledge that can be quantitatively analyzed. However, certain
characteristics of researchers in these fields are obstacles to measurement and
assessment using the same criteria as for scientists working in the pure,
experimental and technical sciences, also know as “hard sciences”. Indeed,
since their publication habits differ considerably from those of other groups ,
most of their research cannot be retrieved from the databases normally used to
obtain information that can be bibliometrically analyzed.
The habits and characteristics of scientists working in the social sciences and
humanities have been analyzed and described in both bibliometric and user
studies. These studies began to shed light on certain characteristics in both
publications and in the use of information to generate new knowledge. As the
characteristics in question had been unknown until then, the features
characterizing other scientific groups were attributed to these scientists by
default. The results of bibliometric and user studies portrayed a previously

25
unimagined reality in which the profiles of the research habits of scientists
working in the social sciences and humanities began to be defined, and the
information centres most suitable for meeting their information needs began to
be designed (Brittain, 1979; Siatri, 1999).
In the case of humanists, the origin of these studies was the project launched
in 1976 by the Centre for Research in User Studies (CRUS) and funded by the
British Library, whose objective was to explore user information needs and behaviour,
identifying such aspects as their limited ability to work in teams or the types of
documents they use. Research on the habits of social scientists dates back
farther than the studies on the humanities, to the late 1960s. These studies were
triggered by both the libraries’ lack of knowledge about this group and by the
interest of professional social science associations in adapting the content of
their courses and programmes to the information needs of their member
researchers. Furthermore, the surge in development that took place in some
social sciences, such as economics and psychology, made it necessary to create
new information centres to meet the needs of scientists in these fields. To this
end, studies were conducted to ensure that these centres would be as well
suited to user needs as possible.
Now, more than thirty years after these beginnings, an intense need has
arisen to understand and evaluate the scientific production of social science and
humanities researchers, for which it is essential to define the features that
characterize them and set them apart from researchers in other areas of
knowledge. Indicators with which to efficiently and precisely determine what
resources have been invested in carrying out their scientific activity must also
be developed.
The publication of research results is fundamental for scientists, as it enables
them to disseminate their activities among the scientific community, which can
then compare and validate the findings. Despite the imperfections widely
acknowledged, this represents an important hurdle to be overcome in research
validation and concomitant inclusion in the scientific acquis. However, in the
social sciences and humanities, publications are often poorly reflected in
national and international databases, partly because of the characteristics of the
sources in which they are published, and partly because of the limited resources
that the producers and distributors of these databases allocate to developing
products in keeping with the characteristics of scientists in these fields. The
limited presence of social scientists’ and humanists’ publications in databases
masks much of their scientific output, and lowers the visibility and awareness of
their research among members of the scientific community working in other
fields.
The geographic scope of social sciences and humanities research is less
international than that of the pure, experimental and technical sciences, as a

26
large part of the former research deals with issues of local interest (Nederhof,
Luwel & Moed, 2001; Al, Sahiner & Tonta, 2006). This means that the journals
used by each group to disseminate their research results differ diametrically:
primarily local or national among the former, and much more frequently
international among the latter.
Therefore, when social scientists and humanists decide to use scientific
papers as a vehicle to convey their information, they usually select national
journals, due to the nature of their research and the fact that their chances of
being published in international journals are smaller than those of other
sciences (López Baena, 2001). Nonetheless, the influence of the local nature of
social sciences research on publication habits is waning, for a number of
reasons (Hicks, 1999). These include the increased internationalization of
national economies and certain technological factors such as the growing use of
electronic communication, which enables these scientists to expand their
research work to the international arena, and the rising percentage of
documents jointly written and published by institutions from different
countries (Katz, 1999).
Likewise related to the local character of the research performed by scientists
in many fields of the social sciences and humanities is the fact that, unlike the
publications of other groups, theirs tend to be in the scientists’ native languages
(Nederhof, Luwel & Moed, 2001). Garfield (1990) notes that documents
published in a language other than English are less visible to the international
community; therefore, these research results are disseminated more slowly than
findings that are more international in nature and reported in English, the
accepted scientific language in many areas of knowledge. This is intensified by
the fact that the results of the research conducted by these groups is not usually
communicated in scientific jargon, as normally occurs in the pure, experimental
and technical sciences; therefore, as they are disseminated in the vernacular,
they can reach a larger audience but not necessarily those who specialize in the
subject. All of the foregoing factors indicate that national publications are the
most suitable channel for the dissemination of this research. This issue has also
been studied at Norwegian universities by researchers in close contact with
academic circles (Kyvik, 1991). The results were similar to the findings of
bibliometric studies conducted in Spain on disciplines such as economics
(García Zorita, 2000), psychology (Lascurain Sánchez, 2001) and the humanities
(Sanz-Casado et al., 2002).
Each scientific discipline is “expressed” through the channels most
appropriate for disseminating the knowledge generated by its researchers. The
suitability of the source is related to several factors. One, probably the most
important, has to do with the obsolescence of the information being reported.
In the case of the humanities and many of the social sciences, the half-life of

27
information is very long, which means that its rate of obsolescence or loss of
usefulness is very low. Therefore, monographs are one of the types of
documents most widely used by researchers to disseminate the knowledge they
generate (Sanz-Casado et al., 2002). Evidently, the low obsolescence rate in
these fields of knowledge means that the content of books, whose publication
can take several years from the time the research process ends, is current for a
long period of time.
However, this situation is very different in the pure, experimental and
technical sciences; as the obsolescence rate is much higher, journal articles are
the means most widely used by scientists in these fields to disseminate their
research results, while monographs are rejected (Hicks, 2004).
All of the foregoing is also very closely related to another characteristic
associated with the social sciences and humanities: the pace of the work
involved in the research processes is much slower than in other areas.
Nevertheless, this has been changing recently in some disciplines, such as
archaeology and anthropology, in which the use of methods borrowed from the
experimental and technical sciences is stepping up the pace, and thereby
providing for speedier scientific results. Therefore, publications in these
disciplines have multiplied. As a result, since information becomes obsolete
faster, these researchers are using scientific journals to publish their research
results.
Furthermore, in certain social sciences, such as economics and psychology,
different types of documents (e.g., journals and monographs) may coexist
because the intermediate rate of obsolescence of the information they publish
makes this information suitable for both types (García Zorita, 2000; Lascurain
Sánchez, 2001). This fact is also reflected in Suárez Balseiro’s bibliographic
review on this subject (2004).
Another factor that influences publication habits in the social sciences and
humanities is the pressure on these researchers to have their scientific activity
evaluated. For example, in the case of Spain, the scientific system that defines
the evaluation criteria on both a national and regional level is beginning to
establish a distinction between the criteria applicable to humanists and social
scientists as opposed to the rest of the scientific community, taking into
consideration how the habits of each group influence its behaviour in acquiring
and conveying knowledge.
At this time, scientists are systematically evaluated for different reasons,
which may be related to selection or promotion processes in their research
careers or to financial incentives linked to their scientific productivity and the
quality of their research. Although the situation in the university environment
cannot be generally applied to all scientists working in the social sciences and
humanities, the most significant research activity in these disciplines takes place

28
in this setting. A study on the scientific production of researchers working in
the Autonomous Community of Madrid showed that over 60% of the activity
in these areas took place in universities (CINDOC-CSIC, 2004).
In Spain, since the Constitutional Act on Universities (Ley Orgánica de
Universidades – LOU, 2001) came into force the scientific activity of these
groups is evaluated on a national level by the National Quality Evaluation and
Accreditation Agency (Agencia Nacional de Evaluación de la Calidad y
Acreditación - ANECA). On a regional level, this evaluation is performed by
assessment agencies in each autonomous community. In the Community of
Madrid, responsibility for such evaluation is incumbent upon the Quality,
Accreditation and Planning Agency (Agencia de Calidad, Acreditación y
Prospectiva – ACAP). A review of the evaluation criteria used by these agencies
shows that, regardless of a researcher’s area of study, one of the common
criteria for all is publication in renowned international journals, usually
periodicals listed in multidisciplinary databases: Science Citation Index (SCI),
Social Sciences Citation Index (SSCI) and Art & Humanities Citation Index
(A&HCI), all part of ISI Web of Knowledge. Nonetheless, the possibility that
the research submitted by a scientist may not be reflected in such sources is
now being taken into consideration. In these cases, other sources to be
consulted are specified, although the publications cited are evaluated by the
evaluation committee.
The scientific activity of all researchers working out of universities or large
public research institutions is evaluated using the same procedures and similar
criteria. All must prove that they are in possession of sexennials, i.e., merits
earned after each six years of service based on outside peer reviews of papers
published. This recognition is accompanied by a boost in prestige as well as
financial rewards.
However, as noted, due to factors inherent in many social science and
humanities disciplines, only a small portion of the output of scientists in these
fields, particularly humanities, is reflected in the aforementioned international
databases. This situation is the result of both the local nature of the research
discussed earlier and the fact that the Web of Science databases provide very
limited coverage of publications that are not in the English language. Only
these databases and Scopus, however, include the bibliographic references
needed to quantify the author or paper citations, through the computerized
processing and analysis of these references.
In the case of Spain, the Web of Science’s A&HCI database for the
humanities offers greater coverage. However, this database does not even
provide a citation index, a tool that is very widely used in the evaluation of
other disciplines. The lack of an index may be attributed to this group’s citation
habits, in turn a result of the obsolescence of the literature in this area of

29
knowledge. Indeed, the index that measures the impact of publications (the
impact factor) is calculated the same way for the social sciences as for the
experimental and technical sciences, i.e., based on the citations a journal
receives in the two years following its publication. Consequently, it ceases to be
meaningful for the analysis of literature with a much longer, irregular citation
period, as in the case of the humanities.
The alternative to such databases, i.e., national resources, generally do not
include the bibliographic references to documents, thereby ruling out any
citation-related bibliometric analysis of scientific papers. To solve this problem,
indices and databases are now being created based on Spanish publications in
certain disciplines of the humanities (RESH, IN-RECS, MODERNITAS
CITAS, etc [explained below]).
Another of the most unique characteristics of social scientists and humanists,
especially in certain fields of the humanities, is researchers’ tendency to work
alone. The many studies that have stressed this issue (Stone, 1982) have found
significant differences between such researchers and experimental and technical
scientists, for whom cooperation between authors and institutions is a
widespread, well-established practice, which has risen steadily for many years.
The tendency of these scientists to publish individually may be detrimental to
their productivity, as several authors found a direct relationship between this
productivity and cooperation among several researchers (Endersby, 1996;
Durden & Perri, 1995). Nevertheless, even though scientific cooperation is not
an easy task, as it requires researchers to communicate and pool their
knowledge, there is a crucial factor that motivates them to make the effort: the
need to publish (Crase & Rosato, 1992). Researchers in this group may be
starting to change their ways, due to the pressure being put on them to
strengthen their curricula vitae. However, this trend has not been observed in
recent papers published on Spanish researchers in modern history (Fernández
Izquierdo et al, 2007).
Nevertheless, the new needs arising in the increasingly competitive scientific
world, where research ever more urgently requires the contribution of different
and complementary expertise from a variety of specialities and disciplines can
be expected to influence this pattern of scientific communication and
cooperation to bring about a gradual change.
In summary, the characteristics that differentiate scientists depending on
their areas of knowledge hinder the creation of bibliometric indicators
appropriate to these characteristics. Moreover these indicators are directly
related to the content and availability of databases that cover these disciplinary
areas. The social sciences and humanities are obviously at disadvantage in this
regard, especially in terms of international databases that are clearly biased
toward journals in the English-speaking milieu, papers published in English and

30
journal articles, to the detriment of other types of documents and other
languages. In this regard, Hicks (1999) also drew attention to the Web of
Science’s databases, to the effect that "the bibliometric community has adopted
the SCI as its de facto standard source [...] However, the more fragmented and
polyglot literature of the social sciences is more difficult to cover in a single
database".

What can be done to improve scientific evaluation studies in the social sciences and
humanities?
Few bibliometric studies have been carried out on the scientific activity of
researchers in the areas of the social sciences and humanities, and in many
cases, when such studies have been conducted, the research results of these
groups have been interpreted in the light of the patterns found in other,
previously studied scientific groups. As a result, some very distinctive realities
in their scientific work have long been concealed, and these need to be
appropriately treated and interpreted in each particular case when the scientific
results of researchers in these disciplines are evaluated.
A number of conceptual and methodological considerations that must be
taken into account are proposed here to improve the interpretation of the
research results of scientists engaging in the social sciences and humanities.
With regard to the conceptual proposals, such scientists have their own
deeply ingrained research habits, as already mentioned. These habits have
gradually been revealed as different information metrics studies have been
undertaken, especially those relating to user and bibliometric studies; (Line,
1971; Brittain, 1979; García-Zorita, 2000; Lascurain-Sánchez, 2001, etc). The
difficulty of learning about the scientific activity of these groups in no way
speaks of the quality of their research; rather, it is indicative of the strategy
followed by the creators and distributors of the major international databases
when selecting the information sources they include, and of the fact that these
groups have aroused little interest among bibliometric specialists, who have not
developed an appropriate conceptual framework to study their production. In
this regard, for a little more than a decade, the Information Metric Studies
Laboratory (Laboratorio de Estudios Métricos de Información - LEMI), a
research group whose aim is to evaluate research and conduct bibliometric and
scientometric studies, has undertaken to pursue studies on these disciplines
(Sanz Casado et al., 2002; Sanz Casado, Conforti & collaborators, 2005,
Fernández Izquierdo et al, 2007). They have conducted research in Spain on
some to obtain theoretical knowledge about the characteristics of these groups
in their scientific activity, as well as methodological knowledge intended to
develop bibliometric techniques, and particularly indicators, that more closely
match their research habits.

31
Some of the findings of this research have made it possible to identify the need
to create benchmarks for comparing the scientific activity of researchers in the
same disciplines working out of different institutions or from different
countries, to ascertain their progress in the aforementioned activity. The
globalization taking place in different segments of society also has a singular
effect on the scientific system; thus, the quality of the research conducted by a
community, group or individual must be compared to the quality of the
research of their national and international peers. To this end, it is essential to
have standards that serve as a basis for such comparisons.
As far as the methodological proposals are concerned, the significant boom
that is currently taking place in the development of bibliometric techniques that
serve as a basis for studies on any scientific group, and especially those groups
working in the social sciences and humanities, means that it is essential to
address methodological considerations that will provide solutions for many of
the issues that have been addressed throughout this paper.
The first of these methodological considerations refers to the sources from
which data are obtained. Because of the limited coverage provided by national
and international databases, the search for sources where researchers in the
social sciences and humanities publish their work poses serious difficulties, and
a significant effort must be made to implement strategies to minimize this
problem. Some of these strategies depend on finding and developing specific
data sources, such as national databases created especially for these groups.
The fact that governments at different levels in Spain have no policy actively
aimed at creating their own databases that would include the vast majority of
Spanish journals in the areas of the social sciences and humanities, along with
the bibliographic references of all the articles published, makes it difficult to
systematically undertake bibliometric studies on these groups to ascertain their
actual scientific output, as well as the impact and visibility of the research they
conduct (Giménez-Toledo, Román-Román & Alcaín-Partearroyo, 2007). The
authorities responsible for developing a coherent policy in this area have always
used the high cost of producing and maintaining these databases as an excuse;
however, with the fast-paced technological advances in today’s world, this
should not be a problem that would justify the failure to take the action needed
to create such data sources.
Research projects are being conducted by scientific groups, aimed at
designing and creating specific databases for certain fields in the social sciences
and humanities to surmount the problems caused by the lack of a centralized
public policy. Even once created, such databases are very difficult to maintain
and update, with notable exceptions. These include the IN-RECS project
developed at the University of Granada’s School of Library and Information
Science, which embraces a variety of disciplines in the social sciences. To date,

32
databases have been developed for the following fields: library and information
science, economics, education, geography, sociology and psychology
(http://ec3.ugr.es/in-recs/). Another initiative is the research project carried
out by research groups affiliated with the Institute for Documentary Studies on
Science and Technology - Centre for the Humanities and Social Sciences
(Instituto de Estudios Documentales sobre Ciencia y Tecnología - Centro de
Ciencias Humanas y Sociales - IEDCYT-CCHS), known as RESH (Spanish
Social Sciences and Humanities Journals) (http://resh.cindoc.csic.es).
Finally, another project with similar characteristics merits mention. This
project one is being conducted by the Information Metric Studies Laboratory
(LEMI) of Carlos III University of Madrid, the History Institute and IEDCYT-
CCHS, both attached to the CSIC (Spanish Council for Scientific Research -
Consejo Superior de Investigaciones Científicas). This project has spawned the
citation index known as Modernitas Citas (www.moderna1.ih.csic.es/emc),
which is being developed with data taken from modern history journals. As for
RESH, the first stage of this project consisted of selecting the journals in this
field with the highest quality. To this end, a variety of evaluation criteria were
used, ranging from quantitative to qualitative. The bibliographic references for
each article have been included in this index, to create a tool with which to
both establish queries by source publication and retrieve the works cited.
The three projects discussed above take account of each journal’s citation
index, whereby the use and influence of each in its scientific field can be
assessed. As they analyze the citations received by journals, these projects can
be used to ascertain the visibility of Spanish publications in these disciplines.
Similarly to the Web of Science’s databases, they can also be classified
according to this visibility, and their evolution over time can be monitored.
The reports of universities and other research centres are another important
source of data to be taken into consideration in studies evaluating these
scientific groups. Often, as a result of the national database related problems
discussed earlier, and of the fact that universities and other research centres
need to periodically and exhaustively analyze the scientific production of their
teaching staffs to evaluate progress in these areas, programmes designed to
reflect all of the scientific activities conducted by the teaching staff on a yearly
basis are underway in academic domains.
These reports are an extraordinarily precise source of information for
evaluating the research conducted by teaching staff, to implement a policy for
the control and distribution of available resources that is fair to the members of
the institution and appropriate to each individual’s efforts in his or her scientific
work. They would also be one of the best data sources for the study of the
research output of social scientists and humanists, as such reports contain very

33
detailed information about these groups’ scientific activity, information that is
very difficult to find for the reasons discussed earlier.
The researchers’ curricula vitae are another source that should be taken into
consideration; they are particularly useful for bibliometric studies that involve
the individualized analysis of research activity. The fact that many regional and
national agencies are evaluating the scientific production of researchers makes it
necessary, indeed, essential, to systematically ensure that all researchers’
curricula vitae are complete, and in particular those of scientists who work in
the areas of the social sciences and humanities, for their activity is more
difficult to ascertain. In order to standardize this task, several Spanish
universities have begun to develop a new tool to manage research activity,
known as Universitas XXI. This tool represents a significant advance, as its
structure includes a large number of fields that have been adapted to the
specific case of scientific activity in a university environment, in order to gather
information about most of the contributions made by researchers and
especially those in the social sciences and humanities.
Other important methodological aspects to be taken into consideration when
studying the scientific output of researchers in the social sciences and
humanities are data acquisition, processing and analysis. Specific methodologies
should be developed to adapt these processes to their research habits and
characteristics.
In the acquisition of scientific production data, longer time frames must be
taken into consideration, and subject searches must be more exhaustive. The
reason is that the research times of groups working in these disciplines are
usually longer than those of the experimental and technical sciences, and the
types of documents they publish tend to be monographs. This is why it is
difficult to obtain accurate insight into the scientific activity of these researchers
when short time periods, such as normally taken into consideration for other
disciplines, are used.
Subject searches, in turn, must be exhaustive because of the difficulty of
obtaining the scientific production of these researchers when querying
specialized databases, for their research usually covers a wide range of subjects
that are not as strictly delimited as the research of scientists in experimental and
technical sciences.
Another factor to take into consideration with regard to the acquisition of
data is the type of publication in which these researchers disseminate their
work; as these types can vary widely and are usually less visible, data may be
more difficult to find, as discussed earlier.
As far as data analysis is concerned, the fact that researchers in each group
present results that are typical of the discipline in which they work must be
taken into consideration; therefore, they must be analyzed and conclusions

34
must be reached in the context of this discipline and not extrapolated to other
fields (Kyvik, 1991).
Finally, special emphasis should be placed on obtaining bibliometric
indicators adapted to the characteristics of these researchers. This is another
essential aspect of studies evaluating the scientific activity of researchers in the
social sciences and humanities. With the use of these tools, the evaluation
studies that can be conducted on these groups are more complete, yielding a
more accurate reflection of their scientific situation. Specific indicators are
being obtained for this purpose, which in some cases differ from and in others
supplement the indicators used in bibliometric studies conducted in the
experimental and technical sciences. They are essential for acquiring data from
different perspectives relating to the particular characteristics of their activity.
Bibliometric studies of these groups must also obtain a significant number of
indicators to reflect the variability of their research, and seek the convergence
of these indicators, to reveal the peculiarities of their work from different
perspectives.
A great effort is being made at this time to develop multidimensional or
relational bibliometric indicators, which are the most appropriate indexes for
analyzing scientific activities as heterogeneous as those of researchers in the
social sciences and humanities. These indicators are yielding a holistic view of
scientific activity through the simultaneous comparison of different
characteristics involved in this activity.

References
Al, U.; Sahiner, M. & Tonta, Y. (2006) Arts and Humanities Literature:
Bibliometric Characteristics of Contributions by Turkish Authors. Journal of
the American Society for Information Science and Technology. Vol. 57, nº 8, p. 1011-
22.
Brittain, J. M. (1979) Information and its uses. A Review with special reference to the
Social Science. Bath: Bath University Library.
CINDOC (CSIC). (2004) Indicadores de Producción Científica y Tecnológica de la
Comunidad de Madrid (PIPCYT): 1997-2001. Madrid: Comunidad de Madrid.
Crase, D. & Rosato, F.D. (1992) Single versus multiple Authorship in
professional journals. JOPERD: 28-31.
Durden, G. C. & Perri, T. J. (1995) Coauthorship and Publication Efficiency.
Atlantic Economic Journal. Vol, 23, nº 1, p. 69-76.
Endersby, J. W. (1996) Collaborative Research in the Social Sciences: Multiple
Authorship and Publication Credit. Social Science Quarterly. Vol. 77, nº 2, p.
375-92.
España. Ley Orgánica, de 21 de diciembre de 2001, de Universidades, Boletín Oficial
del Estado, 24 de diciembre de 2001. 309: 49400-49425

35
Fernandez Izquierdo. F.; Román Román, A.; Rubio Liniers, C.; Moreno-Díaz
del Campo, F.J.; Martín Moreno, C.; García Zorita, C.; Lascurain Sánchez,
M.L.; Efraín-García, P.; Povedano, E. & Sanz Casado, E. (2007)
Bibliometric study of Early Modern History in Spain based on bibliographic
references in national scientific journals and conference proceedings. In
Torres Salinas, D. & Moed, H.F. Proceedings of ISSI 2007. 11th International
Conference of International Society for Scientometrics and Informetrics. Madrid: CSIC,
p. 266-71.
García Zorita, J. C. (2000) La actividad científica de los economistas españoles, en función
del ámbito nacional o internacional de sus publicaciones: estudio comparativo basado en
un análisis bibliométrico durante el período 1986-1995 [Doctoral thesis]. Getafe:
Departamento de Biblioteconomía y Documentación, Universidad Carlos
III de Madrid.
Garfield, E. (1980) Is information retrieval in the arts and humanities inherently
different from that in science? The effect that ISI's citation index for the
arts and humanities is expected to have on future scholarship. Library
Quarterly. 50: 40-57.
Garfield,, E. (1990) The languages of science revisited: English (only) spoken
here? Current Contents, nº 31, p. 3-18.
Giménez-Toledo, E., Román-Román, A. & Alcaín-Partearroyo, M.D. (2007).
From experimentation to coordination in the evaluation of Spanish
scientific journals in the humanities and social sciences. Research Evaluation.
Vol. 16, nº2, p. 137-48.
Hicks, D. (1999) The difficulty of achieving full coverage on international social
science literature and bibliometric consequences. Scientometrics. Vol. 44, nº 2,
p. 193-215.
Hicks, D. (2004) The four literatures of Social Sciences. En: Moed, H. F.;
Glänzel, W. y Schmoch, U. (Eds). Handbook of Quantitative Science and
Technology Research. Dordrecht: Kluwer Academic Publishers; 473-496.
Katz, J. S. (1999) Bibliometric indicators and the social sciences. Brighton: SPRU,
University or Sussex.
Kyvik, S. (1991) Productivity in Academia. Oslo: Norwegian University Press.
Lascurain Sánchez, M. L. (2001) Análisis de la actividad científica y del consumo de
información de los psicólogos españoles del ámbito universitario durante el período 1986-
1995 [Doctoral thesis]. Getafe: Departamento de Biblioteconomía y
Documentación, Universidad Carlos III de Madrid.
Line, M. B. (1971) The information uses and needs of social scientists: An
overview of INFROSS. ASLIB Proceedings. Vol. 23, nº 8 p.: 412-34.
López Baena, A. J. (2001) Innovaciones en la evaluación y mejora de la investigación
científica: una perspectiva institucional [Doctoral thesis]. Córdoba: Unidad para la
Calidad de las Universidades Andaluzas.

36
Nederhof, A. J.; Luwel, M. & Moed, H. F. (2001) Assessing the quality of
scholarly journals in Linguistics: An alternative to citation-based journal
impact factors. Scientometrics. Vol. 51, nº 1, p. 241-65.
Sanz Casado, E.; Conforti, N., & collaborators. (2005) Análisis de la actividad
científica de la Facultad de Humanidades de la Universidad de Mar de Plata,
durante el período 1998-2001. Revista Española de Documentación Científica. Vol.
28, nº 2, p. 196-205.
Sanz, E.; Castro, F.; Povedano, E.; Hernández, A.; Martín, C.; Morillo-Velarde,
J.; García-Zorita, C.; Nuez, J. L. de la & Fuentes, M. J. (2002) Creación de
un índice de citas de revistas españolas de Humanidades para el estudio de
la actividad investigadora de los científicos de estas disciplinas. Revista
Española de Documentación Científica. Vol. 25, nº 4, p. 443-54.
Siatri, R. (1999) The evolution of user studies. Libri, Vol.49, nº 3, p. 132-41
Stone, S. (1982) Humanities scholars: information needs and uses. Journal of
Documentation. Vol. 38, nº 4, p. 292-313.
Suárez Balseiro, C. A. (2004) Perfiles de actividad científica de los departamentos de la
Universidad Carlos III de Madrid: un estudio con variables de recursos y resultados del
proceso científico durante el período de 1998 a 2001 [Doctoral thesis]. Getafe:
Departamento de Biblioteconomía y Documentación, Universidad Carlos
III de Madrid.
___________________________

Addresses of congratulating authors:

ISABEL IRIBARREN-MAESTRO
Carlos III University of Madrid, C/Madrid, 126. Getafe 28903 Madrid (Spain)
Email: iiribarr@bib.uc3m.es
MARÍA LUISA LASCURAIN-SÁNCHEZ
Carlos III University of Madrid, C/Madrid, 126. Getafe 28903 Madrid (Spain)
Email: mlascura@bib.uc3m.es
ELIAS SANZ-CASADO
Carlos III University of Madrid, C/Madrid, 126. Getafe 28903 Madrid (Spain)
Email: elias@bib.uc3m.es

37
Persson’s universe of bibliometrics – Has his mapping
changed the discipline?

Martin Meyer1 and Wolfgang Glänzel2

1 SPRU Science and Technology Policy Research, Freeman Centre, Brighton, (UK)
2 Centre for R&D Monitoring (ECOOM), Katholieke Universiteit Leuven, Dekenstraat 2, (Belgium)

Introduction
Nowadays we are faced with a plethora of impressive maps reflecting the
structure of the research landscape and the universe of documented scholarly
communication, expressing important positional and relational aspects by
measuring the distance among and similarity of individuals objects and clusters.
Beyond doubt, this cartography of science and technology permits insight into
important aspects of the cognitive structure of scientific research, helps
monitor the evolution, that is, the emergence, convergence and decline of
research topics and disciplines and thus the changing universe of science.
Studying these maps and their frequently appearing updates, the question
arises of how scientists, librarians and information practitioners can translate
their observation into the practical needs of their daily work. In other words,
how can scientific information provided by these maps broaden the
consciousness of what is relevant for the own and the colleagues’ research and
thus possibly improve the efficiency of communication in science as well? The
co-citation based Atlas of Science developed and issued by the Institute for
Scientific Information (ISI) was actually one of the first endeavours in mapping
the cognitive structure of the research landscape and this atlas was considered a
new kind of ‘review literature’ which might also be suited to help students in the
choice of career in science (Garfield, 1975, 1988). This also implies that (future)
scientists might learn from these visualisations, in particular, what is ‘useful’ and
what might be ‘hot’ in their discipline, and might thus be able to better find and
position their own research tasks. This function of science maps reaches far
beyond the scope of information science, in general, and bibliometrics, in
particular.
A second question arises from this perspective, namely: could this effect be
strengthened if scientists can prepare their own maps to better understand their
own role and position in the network of scientific communication? The answer
is given by Olle Persson’s work; it is not a suddenly formulated clear and unique
answer but – as we will see in the following – the solution was found in
presenting a toolbox and in continuous interaction with its users. This solution
can be considered an extension of Persson’s notion of online bibliometrics as a

39
research tool for everybody (Persson, 1986). In order to “measure” this effect
we will compare utilisation and impact of this toolbox in the scientific
community with the impact of Persson’s “regular” scientific work.

Bibmap and Bibexcel – a Swedish look at science

Most results of the cartography of science are presented as results of scientific
research in information science or as commercial products as they are provided.
Although some of these products are “tailor-made”, a real interaction between
producers and users of the maps remains rather the exception to the rule. Of
course, there is public-domain software for plotting the results of some analysis
of data downloaded from bibliographic databases or for conducting network
analysis based on such data (e.g., Pajek, see Batagelj and Mrvar, 2002).
However, the application of such tools requires careful data processing prior to
the use. Their application was therefore rather restricted to experts or at least to
researchers with basic skills in both bibliometric techniques and computer
programming. The gap between the immediate download from a citation index
and a user-friendly and completely customisable tool to be used for analysis and
visualisation of bibliometric data was bridged by Olle Persson’s ingenious
design of Bibexcel as early as the beginning 1990s. The idea is intimately
connected with ‘BIBMAP’ (Persson et al., 1992). The basic idea behind
Bibexcel is to prepare and process the downloaded records so that those can be
used as the direct input of any advanced spreadsheet application. The author’s
description of this product sounds rather modest and sober (Persson, 2009).
“Bibexcel is designed to assist a user in analysing bibliographic data, or any data of a
textual nature formatted in a similar manner. The idea is to generate data files that
can be imported to Excel, or any program that takes tabbed data records, for further
processing.”
Nevertheless, this tool has revolutionised the work of many users word-wide.
Its use had been documented through citations or references to its author
and/or website. Traditional abstract and citation databases are therefore not
suitable for retrieving the application of Bibexcel. Hence,we decided to use the
web and, above all, Google Scholar (Google Inc., Mountain View, CA, USA). In
order to measure the impact of Bibexcel on the community, we compare the
outcomes with citations received by other Persson’s work (excluding Bibmap
and Bibexcel). The reference data have been retrieved in and extracted from the
Web of science (WoS, Scientific – Thomson Reuters, Philadelphia, PA, USA).
Citing papers have been retrieved for the same period as for the documents in
Google Scholar (from 1992 on). Besides Persson’s papers indexed in the WoS
since 1992, his two contributions in the handbooks of Quantitative Science and
Technology Research (Ed: AFJ van Raan, Elsevier, 1988; Eds: HF Moed, W

40
Glänzel, U Schmoch, Kluwer, 2004) have been taken into account. The results
are analysed and discussed in the following section.

The impact of Bibexcel on the community

The direct comparison between the utilization of Bibexcel and the reception of
Persson’s research work by the community immediately reveals interesting
divergences in the use of information. In this context, we have to stress that the
validity of the interpretation of the results taken from different sources will be
somewhat limited. This has to do with different quality criteria of indexing
documents in the two databases and the “language barrier”. While almost all
citing papers in the WoS are in English, a great deal of documents found in
Google Scholar are in other languages (Swedish, Spanish, Chinese etc.).
Nevertheless, we are confident that the outcomes of this comparison are
sufficiently significant. Above all, the question of who are the Bibexcel users
and what they are dealing with will be in the foreground. Before we tackle this
issue, will have a view at Bibexcel utilization in the mirror of Google Scholar
(see Table 1).
Table 1 The use of Bibmap/Bibexcel in the mirror of Google Scholar
(*Citations counts estimated for the complete year)
Citations received
PY
Increments Cumulated
1992 2 2
1994 1 3
1996 1 4
1998 1 5
1999 2 7
2000 4 11
2001 6 17
2002 9 26
2003 8 34
2004 14 48
2005 20 68
2006 15 83
2007 23 106
*2008 18 124

Since data were retrieved in 2008, we had to extrapolate on the basis of the
trend of the previous ten years. For the period January–July 2008 we have
counted 10 citations. Based on the exponential model suggested by Figure 1,

41
we have applied an exponential extrapolation to estimate the complete citation
impact for 2008.

Figure 1 Growth of Bibexcel application as measured by citations recorded in Google Scholar

The geographical distribution of Bibexcel users immediately shows a
considerable deviation from that of citers of Olle Persson’s research work (cf.
Table 2). The fact that the software is mostly used in Sweden is obvious since
Persson as the author and inventor of this tool is assumed to used it frequently
in his daily work. The rest of the distribution substantiates that Bibexcel is “a
research tool for every man” indeed (cf. Persson, 1986). Scientists from less
developed and newly industrialised countries, above all, in Latin America,
Africa and Asia are well-represented. By contrast, we find authors from those
countries on top of the ranking list of Persson citers, which are also otherwise
among the most active in bibliometric research.
Table 2 The geographical distribution of users of Bibexcel in comparison with that of citers of
Persson’s research work (share ≥2.5%)

Bibexcel Persson
Rank
Country Cites Share(%) Country Cites Share(%)
1 Sweden 21 18.1 USA 65 16.0
2 England 14 12.1 Netherlands 44 10.9
3 Denmark 13 11.2 Sweden 43 10.6
4 USA 10 8.6 England 40 9.9
5 Spain 9 7.8 Spain 33 8.1
6 China 8 6.9 Canada 30 7.4

42
 Bibexcel Persson
Rank
Country Cites Share(%) Country Cites Share(%)
7 Cuba 8 6.9 Germany 28 6.9
8 Finland 6 5.2 Belgium 24 5.9
9 India 6 5.2 France 21 5.2
10 Mexico 6 5.2 Denmark 19 4.7
11 South Africa 6 5.2 Japan 18 4.4
12 Belgium 4 3.4 China 18 4.4
13 Germany 4 3.4 India 15 3.7
14 Argentina 3 2.6 Finland 14 3.5
15 Australia 13 3.2
16 Hungary 11 2.7

The second question concerns the topic or context in which Bibexcel is used.
The large share of users in information science in both populations is, of
course, not surprising. Nevertheless, there is an essential deviation outside this
user group (see Tables 3 and 4). While more than 50% of the papers referring
to or reporting utilisation of Bibexcel can be assigned to subjects outside
information science and a significant user group could be found even in the life
sciences, citers of Persson’s research work are according to the ISI database
rather restricted to “main discipline” of information science. The disciplines in
Computer science are “greyed out” because those are a by-effect of the
multiple assignment of most information-sciences journals (JASIST,
Scientometrics, JOI, Journal of Information Science, IP&M, etc.). Only the
weight field of Operations and Management science and related fields roughly
coincides in both populations. In this context we have to mention that we
avoided multiple assignments in the Bibexcel user group while we just used the
not uniquely defined ISI Subject categories for the papers citing Persson’s
research work.
Table 3 Domains in which Bibexcel is used (based on the share of citing documents indexed in
Google Scholar; ≥5%)

Field Share (%)

Information Science & Scientometrics 46.7
Health & Medicine 17.1
Operations & Management science 9.5
Social sciences & Education 5.7
Others 21. 0

43
Table 4 ISI Subject Categories in which Persson is cited (based on the share of citing documents indexed in the Web of
Science; ≥5%)
Field Share (%)
Information Science & Library Science 68.1
Computer Science, Interdisciplinary applications 34.8
Computer Science, Information systems 20.2
Management 9.1
Planning & Development 6.2

Now it is time to have a look at the impact of Bibexcel from the dynamic
perspective. The annual change of documented Bibexcel utilisation was already
presented in Table 1. Figure 1 presents both annual increments and cumulated
number of citations in Google Scholar. The trend is estimated on the basis of
an exponential regression. This model provided the best fit with a strong
correlation with a correlation coefficient of r = 0.995. This picture is contrasted
by the evolution of citations to Persson’s research work which, in turn, can be
characterised as a sub-exponential but supra-linear growth (see Figure 2). The
power model proved to provide the best estimate. The correlation coefficient
amounts to r = 0.998 and the regression equation is an almost perfect quadratic
function (y = x2). Both cases reflect a supra-linear growth of the impact of
Persson’s work, where the popularity of Bibexcel outruns the effect of the
research impact by approaching a quasi-exponential growth although this trend
seems to somewhat drop in the last two year (cf. Figure 1).

Figure 2 Evolution of citations received by Persson’s research papers as reported by the Web of Science

44
Another way of looking at the citation of Bibexcel is to think of its adoption by
the community as the diffusion of an innovation. There are several approaches
modelling these diffusion processes (e.g. Rogers, 1995; Bass, 1969), most of
which suggest an s-shaped curve to describe the (accumulated) uptake of an
invention. Rogers (41995) made a distinguished different categories of adopters
– innovators (accounting for about the first 2.5% of new users), early adopters
(13.5%), early majority (34%), late majority (34%), and laggards (16%).
Roughly, the growth in the application of Bibexcel compares to the early
adopter stage in Rogers’ model. Assuming that the model applies in this
instance, this would suggest that there is still plenty of room for potential new
users to adopt Bibexcel. A closer inspection of the citation data in GoogleScholar
suggest that since the early years of our decade (especially from 2003 onwards),
Bibexcel has been increasingly used outside the Library and Information
Science community, winning over new users in fields, such as health, education,
sociology as well as technology and engineering management.

Conclusion
From measuring and comparing the reception and utilisation of Olle Persson’s
research work and his software tool Bibexcel we may conclude that Bibexcel
proved indeed a tool for everybody in research and application not only in
information-science related disciplines. This tool is widely used within as well as
outside the main areas (both geographically and in a cognitive sense). The
increasing popularity of Bibexcel can be characterised as even being
exponential. Persson’s tool attracts more and more new users in other
communities and is becoming more and more what DeSolla Price (1984) once
called an ‘instrumentality’ – an instrument, technique, or procedure that serves
as a driver of research across fields and disciplines in science and technology.

References
Batagelj V, Mrvar A (2002). Pajek – analysis and visualization of large
networks. Graph Drawing, 2265, 477–478.
Bass FM (1969), A new product growth model for consumer durables.
Management Science, 15,
215-227.
Garfield E (1988), The encyclopedic ISI Atlas of Science launches three new
sections: Biochemistry, Immunology, and Animal & Plant Science, Current
Contents, (7), 3-8.
Garfield E (1975), ISI’s Atlas of Science may help students in choice of career
in science, Current contents, (29), 5-8.
Persson O (1986), Online Bibliometrics. A research tool for every man.
Scientometrics, 10, 1-2, 69-75.

45
Persson O, Stern P, Holmberg K-G (1992), BIBMAP: a toolbox for mapping
the structure of scientific literature. In: Representations of Science and
Technology. Leiden, the Netherlands, DSWO Press, p. 189-199.
Persson O (0000), BIBEXCEL. Accessible via:
http://www8.umu.se/inforsk/Bibexcel/index.html.
Price, DJdS (1984). The science/technology relationship, the craft of
experimental science, and policy for the improvement of high technology
innovation. Research Policy 13(1), 3-20.
Rogers EM (1995), Diffusion of innovations. 4th edition, Free Press: New
York.
___________________________

Addresses of congratulating authors:

MARTIN MEYER
SPRU Science and Technology Policy Research, Freeman Centre, Brighton, UK)
E-mail: M.S.Meyer@sussex.ac.uk
WOLFGANG GLÄNZEL
Centre for R&D Monitoring (ECOOM), Katholieke Universiteit Leuven,
Dekenstraat 2 (Belgium)
E-mail: Wolfgang.Glanzel@econ.kuleuven.be

46
The most influential editorials

Ronald Rousseau
KHBO, Industrial Sciences and Technology, Zeedijk 101, 8400 Oostende (Belgium)
K.U.Leuven, Department of Mathematics, Celestijnenlaan 200B, 2001 Leuven (Heverlee), (Belgium)

Abstract
This article studies the citation influence of editorials. It is found that only a
few journals publish highly cited editorials (New England Journal of Medicine,
Nature and Science) and that their authors come overwhelmingly from the USA,
with Harvard as the leading university. It turned out that this article is much
more about methodology than about finding the most influential editorials.

Introduction
An editorial is like a short essay or a written speech. Most speeches make no or
little impression and are forgotten the moment they are spoken. Yet, some have
a lasting influence: take for instance Dr. Martin Luther King Jr’s “I have a
dream” speech, Mahatma Gandhi’s “Quit India” speech and J.F. Kennedy’s
“Ich bin ein Berliner” speech.
Editorials are often just a presentation of the contents of a journal’s issue
including maybe some comments on salient points. Other editorials contain
reflections related to some special event. Yet, every now and then an editorial
contains a forceful message, a call for action, perhaps a little known scientific
fact with far-reaching consequences is brought into the limelight. These are the
editorials that are remembered by fellow scientists. Whatever the concrete
contents of such editorials, whatever the concrete journal and field in which
they are published, one may rightly say that the editorial itself is not a scientific
contribution in that particular field. It is a literary piece, to be classified as an
article in the humanities, even if written by a professional physicist, cell
biologist or economist. But, are these the editorials that are most cited? In this
article, dedicated to Olle Persson on the occasion of his 60th birthday, we
investigate which editorials are highly cited and in which journals they are
published.

47
Definition

Definition
How to define an editorial? An editorial in a regular newspaper is usually a
short article expressing an opinion or point of view, written by the main editor
or another member of the publication staff. Yet, as mentioned above, in
scientific articles an editorial is somewhat different. As it is, moreover,
impossible to read all articles that might be considered to be an editorial, we
used the Web of Science’s concept of ‘editorial material’ as a starting point.
After some try-outs we opted for the following definition: an editorial is a
publication classified as ‘editorial material’ (in the WoS), of length at most 3
pages, with a reference list of at most 10 items and written by one person (or
published anonymously). We recall that Garfield (1987) has published an
internal (ISI) grading algorithm for determining so-called substantial articles
that are not published in the same way as regular articles or reviews. These
substantial articles are published as letters, editorial materials, comments but are
actually research articles and should be treated as such. The editorials we are
looking for are not substantial research articles. If Thomson Reuters still uses
an algorithm similar to the one published by Garfield, then these substantial
articles are automatically eliminated from the category of editorial materials. We
think, however, that this is not the case, as we found many ‘substantive’ articles
among the so-called editorial material. So, for this reason we had to delineate
the set of editorials further.

Data Collection
For the period 1975 – 2008 we collected each year’s five most-cited editorials
(according to the definition above). This leads to a total of 172 articles (two ties
on the fifth place). Data collection took place during the first week of February
2009. When checking some results we found out that some articles registered in
the WoS as a single-authored paper were actually written by a committee, using
a group name; or were just misrepresented in the WoS. The following
publication is a case in point:

The article
A working formulation for the standardization of nomenclature in the
diagnosis of heart and lung rejection: Heart Rejection Study Group.
By The International Society for Heart Transplantation: Billingham, ME, Cary
NR, Hammond ME, Kemnitz J., Marboe C., McAllister HA, Snovar DC,
Winters GL, Zerbe A.
Journal of Heart Transplantation (1990), vol.9, pp. 587-93

48
is registered in the Web of Science (with zero citations), but all citations are
assigned to:

ME. Billingham
International Society for Heart-Transplantation
Journal of Heart Transplantation (1990), vol.9, p. 587-587

This article received 1,241 citations (February 2009), hence was first included in
our list of most-cited editorials. We checked this case and found that on page
587 Margaret Billingham, the first author of the actual article wrote a short
introductory note (one that rightly could be called an editorial). However, ISI,
assigned all citations to the standardization of nomenclature article to this short
introductory note. Actually, the journal itself is not very clear, but its table of
contents is. Indeed: this journal issue begins with three sections: President’s
Message (this is Margaret E. Billingham’s short editorial), International Society for
Heart Transplantation and Original Articles. Clearly, ISI has used the name of the
section as the title of Billingham’s untitled editorial. The section International
Society for Heart Transplantation contains two proposals for standardization: one
by the Heart Rejection Study Group and one by the Lung Rejection Study
Group (beginning on page 593 and cited 343 times). This example highlights
one of the methodological problems related to the study of scientific editorials.

Results

The Top 10
We begin by showing the top 10 most-cited editorials published over the period
1975 – 2008. Three of these articles are from Harvard University, two of which
written by Nobel laureate Walter Gilbert. The first one has one author, who,
however, acts for a committee.
Table 1 The 10 most-cited editorials (according to our definition)

Rank Document No. citation

Manfred Zimmermann
Ethical guidelines for investigations of experimental pain in conscious animals
1 2452
Pain (1983), vol.16, pp. 109-110; 9 references
Published as: Guest editorial

Walter Gilbert
Why genes in pieces?
2 1520
Nature (1978), vol. 271, p.501; 10 references
Published as: News and Views

49
 Rank Document No. citation

Lewis C. Cantley
The phosphoinositide 3-kinase pathway
3 1257
Science (2002), vol.296, pp. 1655-1657; 8 references
Published as: Viewpoint

Craig L. Hill
Introduction: Polyoxometalates - Multicomponent molecular vehicles to probe
4 fundamental issues and practical problems. 761
Chemical Reviews, (1998), vol.98, pp. 1-2; 0 references
Published as: Introduction

Walter Gilbert
Origin of life – the RNA world.
5 711
Nature (1986), vol. 319, p.618; 9 references
Published as: News and Views

Cyrus Chothia
Proteins – 1000 families for the molecular biologist
6 543
Nature (1992), vol.357, p.543-544; 10 references
Published as: New and Views

David Wynford-Thomas
P53 in tumor pathology – can we trust immunocytochemistry?
7 446
Journal of Pathology (1992), vol.166, 329-330; 6 references
Published as: Editorial

Robert J. Bodnar
Revised equation and table for determining the freezing-point depression of H2O-
8 NaCl solutions 413
Geochimica et Cosmochimica Acta, (1993), vol.57, pp.683-684; 2 references
Published as: Scientific comment

Arnold S. Relman
Assessment and accountability – the third revolution in medical care
9 375
New England Journal of Medicine (1988), vol.319, pp.1220-1222; 4 references
Published as: Editorial

Christopher S. Foote
Definition of type I and type II photosensitized oxidation
10 370
Photochemistry and Photobiology (1991), vol.54, p. 659; 7 references
Published as: Guest Editorial

Journals- Fields
In which journals are these most-cited editorials published? Essentially there
are three journals in which highly-cited editorials are published: the New England
Journal of Medicine occurs 39 times in the list, Nature 30 times and Science 24
times. The next journal in the list is Chemical Reviews with 5 editorials. Yet, it is
no surprise that the size-frequency list of journals follows a Lotka (power law)

50
 0.73
distribution: f ( y ) = , where f(y) denotes the relative number of journals
y 2.43
with y contributions to this list. Fitting has been performed using the LOTKA
program (Rousseau & Rousseau, 2000).
Next we determined to which fields these journals belong. Writing successful
or at least highly cited editorials is a two field business: Multidisciplinary Sciences
(54) and Medicine, General & Internal (51). Also here the complete list of fields
0.665
follows a Lotka distribution: f ( y ) = , where now f(y) denotes the
y 2.18
relative number of fields with y contributions to this list.

Addresses
In which countries do the writers of highly cited editorials work? Several of
these editorials do not have an address, but among those who have the USA
leads by a wide margin. Table 2 shows this elite group of countries. The UK
figure consists of 18 contributions from England, 2 from Wales and 1 from
0.43
Scotland. Even this short list has Lotka characteristics ( f ( y ) = ).
y1.58
Table 2 Countries
Number of Number of
Country Country
contributions contributions
USA 93 Switzerland 2
UK 21 Australia 1
Germany 9 Canada 1
Japan 4 Italy 1
The Netherlands 3 Sweden 1
Denmark 2

As to universities or institutes: Harvard has 10 contributing articles, the

National Institute of Health (NIH) 5 as does the University of California,
Berkeley; the University of Texas has 4 contributions and so does Oxford
University. One could say that this is a mini-list of top universities and
institutes.

Some further remarks

The most-cited editorials have, by the definition we used, between zero and ten
references. One would expect that a ‘real’ editorial has no references. This
turned out to be true, in the sense that zero references is indeed the mode of

51
the number of references. Yet, many so-called editorials have a large number of
references (7 to 10). Details are shown in Table 3.
Table 3 Number of references
Number of Number of
Number of articles Number of articles
references references
0 37 6 11
1 9 7 21
2 6 8 12
3 6 9 18
4 9 10 31
5 12

We also performed a search to find the most-cited editorial in the information

sciences. Several articles by Eugene Garfield almost made it, but their
references lists were too long. The same is true for Jean Tague-Sutcliffe’s
introduction to a special issue on informetrics in Information Processing and
Management (Tague-Sutcliffe, 1990). The one we found is the late Michael
Moravcsik’s announcement in Scientometrics of the first Derek de Solla Price
awardee:
M.J. Moravcsik – 8 citations
Address at the presentation of the 1st Price, Derek D. award to
Garfield, Eugene on December 20, 1984
Scientometrics (1985), vol.7, pp. 143-144; no references

Conclusions
Newspaper editorials have been studied in Journalism and Communication
Science, and also Garfield has paid attention to the interesting ideas that are
published in editorials (Garfield, 2000), but to the best of our knowledge this is
the first article trying to highlight influential editorials in academic journals.
This investigation turned out to be a pilot study for those who want to use
editorials as part of a larger research evaluation exercise, or as a part of an
investigation on the structure of science. Two serious problems have been
detected. The first is related to the definition of an editorial. Probably, the
definition we used is still too broad. Several of the ‘editorials’ we found can
better be described as short scientific communications. We, most certainly, did
not find an equivalent to the famous speeches mentioned in the introduction.
Taking into account that the first characteristic we applied to define an editorial
was to be considered ‘editorial material’ in the Web of Science, this shows how
this category is actually a very mixed bag.

52
The second problem is related to the indexing of the Web of Science. We
cannot be sure if the single-authored articles (based on the records in the WoS),
we retained as editorials are actually single-authored.
Clearly, these problems can only be solved by using a more refined definition
of ‘an editorial’ and by having access to the original articles, namely the
supposed editorials themselves, in order to visually check their exact content.
Maybe interesting proposals, or calls to arms (similar to the famous speeches)
can also, or even more often, be found as Letters to the editor or Correspondences.
One final idea: can some types of editorials be used to track new or even future
developments in science? If the answer is yes, this would make them a data
mining tool, among many other ones, to predict new and emerging trends.

Acknowledgments
I thank Nadya Verschelden and Kurt Noppe (Ghent University, Belgium) for
their help in finding and describing Margaret Billingham’s article, and Mikael
Graffner (Lund Univ.) for checking Zimmermann’s contribution. Raf Guns
(Antwerp University) is acknowledged for useful comments on an earlier
version.

References
Garfield, E. (1987). Why are the impacts of the leading medical journals so
similar and yet so different? Item-by-item audits reveal a diversity of
editorial material. Current Comments, #2, p.3 January 12.
Garfield, E. (2000). Foreign language editorials should be translated for the
Web. The Scientist, 14(9), p. 6.
Rousseau, B. & Rousseau, R. (2000). LOTKA: a program to fit a power law
distribution to observed frequency. Cybermetrics, 4(1), paper 4. Data
Retrieved May 8, 2009 from:
http://www.cindoc.csic.es/cybermetrics/articles/v4i1p4.html
Tague-Sutcliffe, J. (1992). An introduction to informetrics. Information
Processing and Management, 28, 1-3.
___________________________

Address of congratulating author:

Ronald Rousseau
KHBO, Industrial Sciences and Technology, Zeedijk 101, 8400 Oostende (Belgium)
K.U.Leuven, Department of Mathematics, Celestijnenlaan 200B, 2001 Leuven (Heverlee), (Belgium)
Email: ronald.rousseau@khbo.be

53
Publication patterns in all fields

Gunnar Sivertsen
NIFU STEP, Wergelandsveien 7, N-0167 Oslo (Norway)

My first encouter with bibliometrics was in April 1988 in Elsinore, Denmark at

a conference about scholarly publishing that was organized by the Nordic
Council of Ministers. I remember well how bibliometrics suddenly personalized
as the floor was taken by a bearded Swede in tweed and jeans. He
demonstrated empirically the need for further internationalization of the social
sciences in Scandinavia by using articles in scientific journals as a data source
(Persson 1988). It was amazing.
I attended the conference because I had been invited to speak about
international marketing of scholarly publications (Sivertsen 1988). At that time,
I was the editorial director of the journals department at Scandinavian
University Press. I knew nothing about bibliometrics, although I do recall that
my office – for a reason I could not understand – subscribed to an obscure
journal from Eastern Europe with an orange cover and a futuristic title.
Anyways, I never read nor opened the journal before Olle demonstrated what
bibliometrics could do in April 1988.
Only a few weeks later, I shifted over to my present position as a researcher
at NIFU STEP and started with bibliometrics from day one. It did not take
long before Nordic collaboration became both a subject matter and a research
practice resulting in co-publications in Norwegian (Luukkonen, Persson &
Sivertsen 1990), in German (Luukkonen, Persson & Sivertsen 1991a), and in
English (Luukkonen, Persson & Sivertsen 1991b). We would also publish
together in the obscure journal with an orange cover (Luukkonen, Tijssen,
Persson & Sivertsen 1994).
As Olle noted in his presentation at the conference in 1988, there was an
increasing focus at that time on publication output and citation impact as
indicators for research evaluation and policy. But bibliometric studies were
mainly performed within the natural sciences. Olle was ahead of time with his
paper about the social sciences. Now, a little more than twenty years later, and
after the digital revolution, bibliometrics is expected to cover all areas of
research.
Some of the pressure in this direction comes from the new models for
performance based funding of research institutions. In these models, the
performance of institutions as such is measured, and the experience so far (e.g.
in Flanders, United Kingdom and Sweden) shows that it is politically
impossible to leave out the social sciences and the humanities. Two or three

55
alternative – or combined – solutions are therefore discussed (Hicks & Wang
2009): One of them is to rely on “the recent aggressive expansion by WoS and
Scopus”, which points in the direction of full coverage of “all sound journals”.
The second “is hinted at in two current metrics-based systems, the Norwegian
and Australian. Both rely on national research documentation systems.” The
third is “creating an electronic, full text infrastructure for European SSH
literature”.
For the discussion of the alternatives, more information seems to be needed
about the current publication practices in different fields. The Norwegian
system (Sivertsen 2008) so far provides complete data for 30.000 scientific
publications (fractionalized counts) from Norway’s higher education sector in
four years, 2005-2008. I have analyzed the data in a simple manner in table 1. It
shows field variation in publication practices in three dimensions: Coverage by
Web of Science (WoS), use of foreign language (versus Norwegian), and
publication type (articles in journals and series, articles in books or proceedings,
and books). Articles in series with an ISSN are counted as journal articles. All
publication counts have been fractionalized between the authors and their
institutions. The counts are limited to publications that have appeared with a
scientific or scholarly content and format in a publication channel with peer
review, but publications in local publication channels (with more than two
thirds of the publishing authors are from the same institution) are not counted.
The variations in publication patterns are shown as percentages of the total
of publications within each subfield and major field. The analysis shows that
WoS currently covers two thirds of the scientific journal articles from the higher
education institutions in Norway. If books and articles in books are also
considered, WoS covers about half of the total output. But these shares show
large variations, not only between the major fields, but also within them. The
large variations within the humanities and social sciences indicate that
publication patterns differ with the aims and the subject matter of research, and
that it is difficult to point at a certain publishing practice as a quality standard.
Although new results from research generally need to be exposed to criticism
and further use among the widest possible audience of experts, it is not
necessarily a sign of higher quality that the publication is a journal article in an
international language, and that this article is indexed for a certain database.
A journal for sociologists publishing in Swedish, Sociologisk Forskning, has for
a long time been covered by the WoS. The parallel journal for political
scientists in Sweden, Statsvetenskaplig Tidsskrift, is not covered by WoS. This may
not seem to make much difference, since there are so many other journals in
political science that are covered by WoS, and since it is generally agreed that
the social sciences should strive for internationalization. But from our
Norwegian data, we know that while international publishing – from one

56
country’s point of view – is widely dispersed among many publication channels,
publishing on the national level is concentrated in a few channels that have very
few publications from other countries. In Norwegian sociology, 54 per cent of
the journal articles are concentrated in only four national journals. The other 46
per cent are dispersed among 57 other journals (including Acta Sociologica). In
political science, 27 per cent of the articles are concentrated in only two
national journals. The rest of the articles are published in 121 other journals.
Such skewed distributions are even more apparent when we look at book
publishing.
Since publication patterns vary more across disciplines than across countries,
I expect that we will see similar skewed distributions in other countries as well.
This means that the omission or addition in a central international database of
one national publication channel, e.g. Sociologisk Forskning or Statsvetenskapelig
Tidsskrift, will have great effect on the measurement of the publication output
from a national point of view. But from an international point of view, the
effect on the overall coverage and indicators will be marginal or almost
invisible. Bradford’s law and the notion of “core journals” can continue to be
the cornerstones of coverage policy. It can even be argued, as I once did, that
the addition of internationally insignificant journals may distort international
comparisons (Sivertsen 1992). But for other purposes, publication channels
that are significant on national level do indeed represent a coverage problem.
As noted above, bibliometric databases are now expected to fulfill such other
purposes. In response, Scopus (Elsevier) and Web of Science (Thomson Reuters)
are competing and expanding by covering more journals, more proceedings,
and even books. The picture I show in two of the columns in table 1 is
becoming historic. WoS coverage is increasing and Scopus is expanding maybe
even beyond WoS. It will be interesting to see how WoS and Scopus will meet
the challenges that can be seen in the four other columns of table1. It will also
be interesting to see if there will be alternative or supplementary ways of
meeting these challenges. Anyhow, the coverage of the major international
bibliographic databases is probably no longer only the question of how to
combine Bradford’s law with market opportunities and profit margins. Table 1.
Braun Score values (in per cent) for some scientometricians

References
Hicks, Diana and Jian Wang. 2009. Towards a Bibliometric Database for the
Social Sciences and Humanities – A European Scoping Project.
Unpublished report, School of Public Policy, Georgia Institute of
Technology, April 2009.

57
Luukkonen, Terttu, Olle Persson and Gunnar Sivertsen. Et "verdenskart" over
internasjonalt forskningssamarbeid. Forskningspolitikk (Oslo), no. 3-
4/1990, p. 12-14.
Luukkonen, Terttu, Olle Persson and Gunnar Sivertsen. 1991a. Internationale
wissenschaftliche Kooperationsnetze. In: P. Weingart et al. (Hg.):
Indikatoren der Wissenschaft und Teknik. Theorie, Methoden,
Anwendungen, Campus Verlag, Frankfurt am Main, S. 11-33.
Luukkonen, Terttu, Olle Persson and Gunnar Sivertsen.1991b. Nordic
Collaboration in Science - a bibliometric study.Nord 1991:28. Nordic
Council of Ministers, Copenhagen.
Luukkonen, Terttu, Robert J. W. Tijssen, Olle Persson and Gunnar Sivertsen.
The Measurement of International Scientific Collaboration. Scientometrics
1993, 28, 1, p. 15-36.
Persson, Olle. 1988. Nordisk forskning på den internationella
tidskriftsmarknaden innom samhällsvetenskap. In: Elizabeth Lundberg (ed.):
Internationell vetenskaplig publicering i Norden, Nordisk Ministerråd,
Copenhagen, p. 40-57.
Sivertsen, Gunnar. 1988. Internasjonal markedsføring av nordiske
forskningspublikasjoner. In: Elizabeth Lundberg (ed.): Internationell
vetenskaplig publicering i Norden, Nordisk Ministerråd, Copenhagen, p. 58-
67.
Sivertsen, Gunnar. 1992. Should a new bibliometric database for international
comparisons be more restricted in journal coverage? In: A.J.F. van Raan et
al. (ed.): Science and Technology in a Policy Context. Select Proceedings of
the joint EC - Leiden Conference on Science & Technology Indicators,
Leiden 1991, DSWO Press, Leiden 1992, p. 35-50.
Sivertsen, Gunnar. 2008. Experiences with a bibliometric model for
performance based funding of research institutions. In: J. Gorraiz J. and E.
Schiebel (eds): Excellence and Emergence. A New Challenge for the
Combination of Quantitative and Qualitative Approaches. Book of
Abstracts. 10th International Conference on Science & Technology
Indicators. Vienna, Austria. 17-20 September 2008. Vienna (AUT): Austrian
Research Centers GmbH, pp. 126-131.
___________________________

Address of congratulating author:

GUNNAR SIVERTSEN
NIFU STEP, Wergelandsveien 7, N-0167 Oslo (Norway).
E-mail: gunnar.sivertsen@nifustep.no

58
Table 1 Distribution of scientific publications in Norway’s Higher Education Sector 2005-2008
according to coverage by Web of Science, use of foreign language and publication type.

WoS coverage of all

Foreign language

Articles in books
WoS coverage of
journal articles
publications

(only ISBN)
Articles
(ISSN)

Books
Major field

Subfield
Engineering Engineering 63 % 86 % 97 % 74 % 26 % 0%
Health Biomedicine 97 % 98 % 100 % 98 % 2% 0%
Sciences
Clinical Medicine 94 % 95 % 83 % 99 % 1% 0%
Dentistry 57 % 57 % 64 % 99 % 1% 0%
Neurology 95 % 99 % 99 % 97 % 3% 0%
Nursing Sciences 40 % 47 % 54 % 86 % 14 % 1%
Pharmacology and Toxicology 88 % 91 % 93 % 98 % 2% 0%
Psychiatry 79 % 84 % 92 % 94 % 5% 1%
Psychology 49 % 65 % 68 % 76 % 22 % 2%
Social Medicine 63 % 72 % 79 % 87 % 12 % 1%
Social Work and Health Care 9% 20 % 33 % 43 % 51 % 6%
Sports Sciences 62 % 79 % 91 % 79 % 20 % 0%
Surgery 93 % 96 % 100 % 97 % 3% 0%
Veterinary Sciences 87 % 88 % 89 % 98 % 2% 0%
Health All subfields 75 % 84 % 81 % 90 % 9% 1%
Sciences
Humanities Archaeology 11 % 22 % 50 % 49 % 47 % 4%
Architecture and Design 5% 8% 44 % 59 % 36 % 6%
Art History 9% 18 % 44 % 51 % 39 % 10 %
Asian and African Studies 9% 21 % 89 % 45 % 48 % 7%
Classical Studies 7% 11 % 50 % 65 % 31 % 4%
English Studies 18 % 51 % 86 % 35 % 59 % 6%
Ethnology 4% 9% 34 % 47 % 46 % 7%
Gender Studies 6% 14 % 31 % 43 % 56 % 1%
Germanic Studies 10 % 27 % 96 % 38 % 54 % 8%
History 16 % 33 % 36 % 48 % 46 % 7%
Linguistics 21 % 36 % 75 % 59 % 38 % 3%
Literature 10 % 17 % 28 % 58 % 39 % 3%
Media and Communication 3% 8% 52 % 38 % 55 % 8%
Music 8% 16 % 34 % 51 % 45 % 5%
Philosophy 7% 12 % 38 % 58 % 34 % 9%
Religion and Theology 7% 14 % 39 % 48 % 45 % 7%
Romance Studies 18 % 45 % 82 % 40 % 51 % 9%
Scandinavian Studies 0% 1% 12 % 30 % 64 % 6%
Slavic Studies 6% 12 % 86 % 50 % 44 % 7%
Theatre Studies 9% 14 % 50 % 60 % 39 % 2%
Humanities All subfields 9% 18 % 44 % 47 % 47 % 6%

59
 WoS coverage of all

Foreign language

Articles in books
WoS coverage of
journal articles
publications

(only ISBN)
Articles
(ISSN)

Books
Major field

Subfield
Natural Biology 85 % 89 % 97 % 96 % 4% 0%
Sciences
Chemistry 96 % 99 % 100 % 97 % 3% 0%
Geosciences 92 % 96 % 99 % 95 % 4% 0%
Informatics 22 % 55 % 93 % 40 % 59 % 1%
Mathematics 75 % 85 % 96 % 88 % 11 % 1%
Physics 94 % 96 % 99 % 97 % 3% 0%
Natural All subfields 81 % 90 % 97 % 90 % 10 % 0%
Sciences
Social Anthropology 12 % 22 % 65 % 56 % 37 % 7%
Sciences
Business and Administration 18 % 32 % 61 % 58 % 38 % 4%
Economics 55 % 69 % 78 % 80 % 18 % 1%
Educational Research 7% 14 % 33 % 49 % 45 % 5%
Geography 35 % 44 % 76 % 78 % 19 % 2%
Law 2% 3% 27 % 64 % 29 % 8%
Library and Information Science 33 % 39 % 93 % 85 % 14 % 1%
Political Science 27 % 60 % 64 % 45 % 51 % 4%
Sociology 12 % 26 % 39 % 45 % 50 % 6%
Social All subfields 18 % 30 % 49 % 60 % 36 % 5%
Sciences
All fields All subfields 48 % 67 % 71 % 72 % 25 % 3%

60
A Webometric Analysis of Olle Persson

Mike Thelwall
Statistical Cybermetrics Research Group, School of Computing and Information Technology,
University of Wolverhampton, Wulfruna Street, Wolverhampton WV1 1LY (UK)

Abstract
This chapter is a webometric analysis of leading scientometrician Olle Persson:
a systematic compilation of evidence of his research impact drawing only upon
web sources. Whilst a purely bibliometric analysis of his refereed journal articles
would demonstrate his mainstream intellectual impact, web-based measures can
potentially reveal a wider impact, including within education. Although
limitations of the data collection method restricted the analysis to just the
impact of Olle’s department Inforsk and his bibliometric software Bibexcel, the
results showed a wide international impact for both. According to the web data,
Bibexcel is the product for which Olle’s department is best known and is
widely used and recommended in education and research. The production and
support of Bibexcel is an unusual research activity that would not be fully
appreciated in a bibliometric analysis but is a major achievement that marks
Olle as a highly influential information scientist.

Introduction
Olle Persson, a widely published scientometrician, has contributed much more
to science than just his articles. As head of the Department of Sociology, Umea
University, as author of the bibliometric software Bibexcel (Persson, 2009), and
as the creator of many maps of disciplines and fields (e.g., Bibliometric maps of
research fields, http://www8.umu.se/inforsk/) using Bibexcel and other
software (e.g., Persson, 1994), his impact spreads far wider than just the
scientific literature. As a consequence, he forms an interesting case study to
explore the extent to which the web can provide impact evidence
supplementing that available from traditional bibliometric sources, such as the
Thomson-Reuters Web of Science.
There have been many previous webometric impact studies, but none
focusing on an individual. As far back as 1998, an investigation was published
to find out how often five highly-cited information scientists in the U.S. were
mentioned on the web, and why. This study found mentions (or “invocations”)
using name queries in commercial search engines. The results revealed a wide
range of different reasons for mentioning the academics’ names online,
including conference information and resource guides (Cronin, Snyder,

61
Rosenbaum, Martinson, & Callahan, 1998). Although there have apparently
been no person-centred webometric studies since 1998, other than co-
authorship maps (Kretschmer, 2004; Kretschmer & Aguillo, 2004), web impact
methods have been developed since then and applied to web sites, web pages
and ideas (Ingwersen, 1998; Thelwall, Vann, & Fairclough, 2006). These are
partially encapsulated in a “web impact report” or a “link impact report” which
contain a variety of statistics derived from mentions of one or more phrases or
identifying links to one or more web sites, respectively. These have been
prepared for organisations including the BBC World Service Trust, the UK’s
National Endowment for Science, Technology and Innovation Research and
the United Nations Millennium Development Programme. The reports are also
used to evaluate online resources like digital libraries and digital archives
(Zuccala, Thelwall, Oppenheim, & Dhiensa, 2007).
This chapter draws upon web impact report and link impact report methods
in an attempt to identify as much evidence as possible about the online impact
of Olle Persson as a way of quantitatively revealing aspect of his impact. Whilst
link and web impact reports typically are focused on a predefined set of web
sites or search phrases, this ego-centred report has a wider focus: anything
online connected with Olle Persson. In contrast, the report does not use the set
of comparators employed in the former types of report because the objective is
not to evaluate Olle’s impact but to describe it.

Data and Methods

Scientometric analyses are often constrained by the ease with which data can be
accessed. An analysis is often possible if there is a pre-existing database such as
the Science Citation Index used by the Web of Science. For a webometric
analysis the web itself is the de-facto database but analyses of the whole web
are normally constrained to the kind of information that can be extracted from
queries submitted to commercial search engines. In this study, two kinds of
queries are considered: link searches and text searches.
Link searches are available in the sibling search engines Yahoo! and AltaVista
and also, in a truncated form, in Google. Live Search allows a kind of link
search not useful here (outlinks). The linkdomain: command allows a search for
all links to any page with a given domain name. For example, the command
linkdomain:X matches all pages containing a link to any page in Olle’s
Department of Sociology. Adding –site:X excludes all pages within Olle’s
university and so linkdomain:X –site:X returns external inlinks or site
inlinks. Like citations, these results can be used as an indicator of how much
impact the department has and where its impact occurs. It is also possible to
search for all pages that link to a given single URL via the link command. For

62
example, link:X –site:X in Yahoo! returns a list of pages linking to X.
This type of search is useful for a more fine-grained web impact analysis.
Text searches are simple search engine text-based queries. Whilst they are, in
principle, a good way to find out how and where an individual is mentioned
online, in practice they only work well for unambiguous queries. For Olle
Persson, the problem is that both Olle and Persson are common Swedish
names and so many or most results for an "Olle Persson" search would
be about different Olles. Similarly, a search for "Persson, O" returns many
irrelevant matches. Although in principle it would be possible to manually filter
out the irrelevant matches, this is impractical due to the large numbers
involved. An alternative strategy is to identify words closely associated by Olle
and with search results dominated by him. Two such words are Inforsk, the
research group of which Olle is a founder member, and Bibexcel, Olle’s free
scientometric analysis software. Whilst an analysis of these gives only a partial
picture of Olle’s work, it is likely to be a useful complement to a standard
scientometric analysis and so cast light on an aspect of Olle’s work that would
not normally be investigated with metrics.
The methods used and reported below are Web Impact Analysis and Link
Impact Analysis (Thelwall, 2009). The essential points of these are described
below alongside the results.

Results

Digital library mentions of Bibexcel and Inforsk

The traditional way of assessing impact is to count citations using the Science
Citation Index or a similar source. The Web of Knowledge (WoK) was used
instead for Inforsk to assess its impact (2 April 2009). A search for Inforsk in
the address field yielded 27 articles having a very impressive total of 368
citations (Citation Report Address=(inforsk) Timespan=All Years.
Databases=SCI-EXPANDED, SSCI, A&HCI, CPCI-S) and peaking in 2007.
The top-cited article had 100 citations: Understanding Patterns of International
Scientific Collaboration (1992) by Luukkonen T, Persson O, Sivertsen G, Science
Technology & Human Values Volume: 17 Issue: 1 Pages: 101-126. This
information suggests a narrow, deep impact for Inforsk: producing a moderate
number of articles, and these articles being excellent.
This information was complemented with a Google Scholar keyword search
for Inforsk, returning a total of 263 research mentions. (2 April 2009). This
suggests a much wider impact than the WoK results. Some was due to
BibExcel and other results were due to the inclusion of a wider class of
documents, such as research reports.

63
Bibexcel was also explicitly searched for in WoK, but it only yielded 6 results –
bibliometric articles written by various authors using Bibexcel for their analysis.
These articles were cited only 6 times in total. Google Scholar produced many
more mentions, probably because of its full-text search facility. It found 185, of
which 9 were citations of Bibexcel in various forms. Eliminating these, Google
Scholar reported the existence of at least 176 research documents mentioning
Bibexcel. (2 April 2009). For this purpose it seems best to disregard the WoK
results and use the Google Scholar results as evidence of significant academic
use for Bibexcel.

Quantitative overview of web mentions of Bibexcel and Inforsk

In order to gather evidence of web mentions of Bibexcel and Inforsk, keyword
searches for both terms were submitted separately to Google, Yahoo! and Live
Search. The professional version of LexiURL Searcher was used with the
standard query splitting option switched on. This program submits queries to
search engines via their Applications Programming Interfaces, which allow
automatic searches. The query splitting option allows extra results to be
obtained beyond the normal maximum of 1,000 URLs per search (Thelwall,
2008). LexiURL Searcher was then used to combine the results of all three
search engines, eliminating duplicate URLs and counting the number of unique
URLs, domains, web sites (identified by domain name ending, e.g.,
Microsoft.com), Top-Level Domains (TLDs) (e.g., .com, .org, .uk) and second
or top-level domains (STLDs) (e.g., .ac.uk, .com.pk, .fr). The statistics
generated are summarised in Table 1. Note that the most useful figure is
probably the number of domains, because content can sometimes be replicated
across different pages of a domain, generating multiple matching URLs from a
single cause. Full quantitative results are available at
http://cybermetrics.wlv.ac.uk/audit/persson/.
Table 1 shows that Inforsk is more widely mentioned online than Bibexcel
but both are mentioned hundreds of times, indicating significant impact.
Table 1 Summary of web matches for Bibexcel and Inforsk.
Base query URLs Domains Sites STLDs TLDs
bibexcel 419 247 207 51 39
inforsk 1303 560 465 80 63

Figure 1 summarises the main locations of interest in Bibexcel and Inforsk, as

judged by web mentions. Ignoring the generic TLDs, it is clear – and somewhat
surprising – that both are extremely international in nature and not just known
in Sweden, Scandinavia or even Europe. U.S. universities (.edu), Germany,
Spain, Finland, the UK and Denmark also feature prominently. Given the

64
relative size of these countries (and Norway) it still seems that Scandinavia is
the region where Bibexcel and Inforsk are best known, however.

Figure 1 The Top Level Domains containing the most websites mentioning Bibexcel or Inforsk.

Content analysis of invocations of Bibexcel and Inforsk

In order to discover why Inforsk and Bibexcel were mentioned online, an
informal content analysis was conducted on 100 random URLs from the
combined search results list, using the lists generated by LexiURL Searcher
reports. These lists contain a maximum of one URL per domain name to avoid
the results being dominated by individual web domains. The content analysis
was inductive: starting with a predefined list of categories and adding extra
categories when common new types of web mention were found.
Figure 2 reports the results of the content analysis of 100 random web sites
mentioning Inforsk. Just over a third of these sites list Inforsk in a list of

65
academic departments. Many of these lists are copies of the Dmoz Open
Directory project category: Science: Science in Society: Academic Departments. Just
under a third of online mentions of Inforsk occur in the context of Bibexcel –
mentioning or crediting the department as the host for this software. Just under
a fifth of the mentions derive from an Inforsk-authored article in a digital
library or hosted on a web site elsewhere on the web. The results overall
suggest that Bibexcel is the major product of Inforsk, perhaps even more
significant than its research papers. This relative importance is only a tentative
suggestion, however, because the content analysis is based upon a maximum of
one page per web site and some digital library web sites contain or list many
Inforsk-authored articles (e.g., Google Scholar, Ingenta, Wiley Interscience).

Figure 2 Reasons for mentioning Inforsk in 100 random web sites.

Figure 3 reports the content analysis of 100 random web sites mentioning
Bibexcel. The majority of the contexts, almost two thirds, are academic
publications: journal articles, conference papers or research reports. Almost all
of these papers use Bibexcel to process their data. This shows genuinely
extensive use of Bibexcel, especially given that some of the digital library web
sites mentioned contain multiple articles mentioning Bibexcel. In addition to
this core use of Bibexcel many web site list it as a useful resource, either in a
link list or as part of a discussion of how to conduct bibliometric analyses.

66
Figure 3 Reasons for mentioning Bibexcel in 100 random web sites.

Conclusion
The web mention data combined from Google, Yahoo! and Live Search clearly
shows that Bibexcel and Inforsk are widely known and mentioned online, with
hundreds of web sites invoking both. The spread of impact is also international,
extending significantly beyond Scandinavia to the rest of Europe and the U.S.
The content analysis of web mentions of Inforsk reveals that it is most
frequently mentioned as part of a list of similar departments originating in the
open directory dmoz. The second most common cause is Bibexcel, with
Inforsk credited as originating the software. This highlights the importance of
Bibexcel for Inforsk. The content analysis of Bibexcel reveals it to be
commonly credited for use in academic and other research and also often listed
as a useful free bibliometric analysis tool. Hence it is clear that Bibexcel is
widely seen as a valuable and practical tool for bibliometrics.
In terms of the contribution of Olle Persson, although for technical reasons
it was not possible to directly measure the full impact of his ideas, the results
clearly show a man with an extraordinary impact on research in terms of his
department and software. Bibexcel was an inspired idea that has proven useful
to hundreds of researchers around the world and has therefore been an
important contribution to scientometrics.

67
Appendix
Table 2 Top-level domains mentioning Bibexcel or Inforsk.
domain bibexcel inforsk total domain bibexcel inforsk total
com 44 135 179 il 0 4 4
se 26 71 97 mx 3 1 4
org 26 50 76 nu 1 3 4
edu 15 26 41 is 1 3 4
net 6 32 38 ru 1 3 4
de 10 21 31 us 0 3 3
es 15 16 31 eu 0 3 3
fi 14 16 30 th 1 2 3
uk 10 14 24 jp 2 1 3
dk 9 12 21 kr 0 3 3
br 4 9 13 hu 0 2 2
cu 7 6 13 ir 0 2 2
pl 2 10 12 my 0 2 2
info 2 9 11 co 1 1 2
fr 4 7 11 ch 0 1 1
in 3 8 11 ro 0 1 1
cn 6 3 9 gr 0 1 1
si 2 7 9 hr 1 0 1
au 2 6 8 bg 0 1 1
no 1 7 8 ua 0 1 1
za 5 3 8 at 0 1 1
ca 4 3 7 zm 0 1 1
nl 2 5 7 pe 0 1 1
tw 2 5 7 tv 0 1 1
it 1 6 7 id 0 1 1
be 1 5 6 yu 0 1 1
tr 1 5 6 ag 0 1 1
gov 4 1 5 np 0 1 1
I.P. 2 3 5 lv 0 1 1
sg 2 3 5 mt 0 1 1
cz 0 5 5 cg 0 1 1
ar 4 1 5

Reference
Cronin, B., Snyder, H. W., Rosenbaum, H., Martinson, A., & Callahan, E.
(1998). Invoked on the web. Journal of the American Society for
Information Science, 49(14), 1319-1328.
Ingwersen, P. (1998). The calculation of Web Impact Factors. Journal of
Documentation, 54(2), 236-243.
Kretschmer, H. (2004). Author productivity and geodesic distance in
bibliographic co-authorship networks, and visibility on the Web.
Scientometrics, 60(3), 409-420.
Kretschmer, H., & Aguillo, I. F. (2004). Visibility of collaboration on the Web.
Scientometrics, 61(3), 405-426.

68
Persson, O. (1994). The intellectual base and research fronts of JASIS 1986-
1990. Journal of the American Society for Information Science, 41(5), 31-
38.
Persson, O. (2009). Bibexcel. A tool-box developed by Olle Persson, Inforsk,
Umeå univ, Sweden. Retrieved April 15, 2009 from:
http://www.umu.se/inforsk/Bibexcel/.
Thelwall, M. (2008). Extracting accurate and complete results from search
engines: Case study Windows Live. Journal of the American Society for
Information Science and Technology, 59(1), 38-50.
Thelwall, M. (2009). Webometrics.New York: Morgan & Claypool.
Thelwall, M., Vann, K., & Fairclough, R. (2006). Web issue analysis: An
Integrated Water Resource Management case study. Journal of the
American Society for Information Science & Technology, 57(10), 1303-
1314.
Zuccala, A., Thelwall, M., Oppenheim, C., & Dhiensa, R. (2007). Web
intelligence analyses of digital libraries: A case study of the National
electronic Library for Health (NeLH). Journal of Documentation, 64(3),
558-589.
___________________________

Address of congratulating author:

MIKE THELWALL
Statistical Cybermetrics Research Group, School of Computing and
Information Technology, University of Wolverhampton,
Wulfruna Street, Wolverhampton WV1 1LY (UK)
E-mail: m.thelwall@wlv.ac.uk

69
Pennants for Strindberg and Persson

Howard D. White
College of Information Science and Technology, Drexel University, Philadelphia, PA 19104 (USA)

Abstract
My contribution to the Persson festschrift is another installment, with new
data, in my current research program linking (1) relevance theory from
linguistic pragmatics with ideas from (2) bibliometrics and (3) information
retrieval. Pennant diagrams are a visualization technique I created to bring all
three together. I interpret two interesting pennants substantively and conclude
with some technical details on creating them.

Why These Two

The title indeed links the renowned Swedish playwright and the honoree of this
festschrift, and, yes, it has a personal significance. While I was in Stockholm for
the 2005 meeting of the International Society for Scientometrics and
Informetrics, I visited the Strindberg Museum (Strindbergs-Museet) and talked
about it with Olle at the conference dinner. After I had returned to the States,
he sent me a memento—the following quotation from the beginning of
Strindberg’s novel, The Red Room, which is subtitled Scenes of Artistic and Literary
Life:

Now the bells of Santa Katrina chimed seven and were echoed by Santa
Maria’s reedy treble, the Abbey and the German church joined in with
their basses, and soon the whole air vibrated with the city's seven bells.
And as, one after another, they fell silent, the last one could still be
heard in the distance, singing its peaceful evensong. This had a higher
note, a purer ring and a swifter tempo than the others....there in the
Santa Klara churchyard, whence the bell could still be heard...

TIME: 1879
PLACE: Stockholm
CIRCUMSTANCE: A May evening at seven o'clock.

—August Strindberg, The Red Room, trans. Elizabeth Sprigge, J. M.

Dent & Son, London, 1967, (Everyman Edition), p. 3.

71
Olle added, “Santa Klara is the church just above Scandic Hotel Continental
where we stayed.” He also attached a sepia illustration of the Red Room itself
as it looked in the 1870s—a hall packed with diners under large chandeliers in
Berns Restaurant in central Stockholm. So he and Strindberg remain associated
in my memory—perhaps a connection not many others would make!

Introduction
Nevertheless, in bibliometrics even the most dissimilar authors can yield similar
patterns of data at the statistical level. As I mulled over ideas for this
contribution, it occurred to me that, since both Strindberg and Persson have
long records as cited and co-cited authors, both could be mapped in pennant
diagrams, a new kind of visualization I introduced in White (2007a, b).
Pennants are intended as one demonstration of the explanatory power of Dan
Sperber’s and Deirdre Wilson’s (1995) relevance theory (RT) when it is applied
to information science. I thought it highly likely that pennants for Strindberg
and Persson as co-cited authors would once again exhibit the relations I had
found for other co-cited authors in earlier studies. The abstract of White
(2009)—a paper produced just before this one—states that “A central idea in
D. Sperber & D. Wilson’s relevance theory is that an individual’s sense of the
relevance of an input in a context varies directly with its cognitive effects and
inversely with its ease of processing in that context.” The paper goes on to
make the nonobvious claim that is explored at length in White (2007a, b): “[A]
formula used in information science for weighting search terms in relevance
rankings
Weight = term frequency * inverse document frequency
instantiates a central idea of Sperber & Wilson’s relevance theory from
linguistic pragmatics
Relevance = cognitive effects / processing effort.
In other words, cognitive effects and processing effort, which S&W discuss
almost exclusively as subjective experiences in individuals, have an objective
analogue in the tf*idf formula at the heart of classic information retrieval.”
The crisp definition of relevance as an effects/effort ratio is drawn from
Goatly (1997). But it is licensed by S&W in many places. For example, Wilson
(2007) uses this formulation in a course she gave on relevance theory at the
University of London (boldface hers):
- Revance to an individual
- Other things being equal, the greater the cognitive effects (of an input to
an individual who processes it), the greater the relevance (to that
individual at that time).

72
 - Other things being equal, the smaller the processing effort required to
derive these effects, the greater the relevance (of the input to that
individual at that time).
Pennants are a means of rendering these relations visually. I will defer until the
last section a discussion of some technical details that underlie them. My
present goal is simply to show how the pennants for my two authors make
qualitative sense, starting with Figure 1.

Figure 1 Pennant for authors co-cited at least 10 times with August Strindberg in Arts & Humanities Search on
Dialog, April 2009

Strindberg
The pennant is formed by using Strindberg’s name as a seed to set an overall
context, and the names of authors co-cited with him are, in the language of RT,
assumptions in that context. They are not the assumptions of a human mind;
rather, they are latent in bibliographic records and made manifest as predictions
by an algorithm. Yet all the pennants I have seen exhibit a kind of low-grade
artificial intelligence. For example, if one asks literary people to name who
immediately comes to mind when “Strindberg” is given as a stimulus, my guess
is that the great majority would answer “Ibsen,” the other giant of Scandinavian
drama. And, sure enough, the rightmost name on the horizontal cognitive effects
scale is Ibsen’s. On the basis of international scholarship, it is Ibsen whose
works are predicted to have the greatest cognitive impact when read with

73
Strindberg’s. Note that this is only a prediction, not a guarantee. But it accords
well with intuition.
More particularly, what “Ibsen H” and “Strindberg A” stand for here is any
of their works in any combination. Ibsen’s oeuvre is large, and Strindberg’s is
immense (it consists of much more than his plays). However, their oeuvres are
here being filtered through a mass of co-citing articles that discuss individual
works. If one drills down to the level of the actual articles, one might find, say,
a study comparing Hedda Gabler and Miss Julie. Therefore, a formidable reading
task involving entire oeuvres is not necessarily being implied.
Ibsen is the author most highly co-cited with Strindberg, and that is why he
has the highest score on the cognitive effects scale. But what does his middling
score on the vertical ease of processing scale mean? It means that he has been
cited in many contexts other than with Strindberg. Pennants always exhibit this
structure; the authors most highly co-cited with a seed author are also well cited
in other contexts. They tend to have large oeuvres with rich implications, both
for the seed and beyond. They are thus pulled to the middle of the ease of
processing scale. But overall, on grounds of the RT notions of relative
cognitive effects and relative processing effort, we can say that Ibsen is most
relevant to Strindberg in the pennant diagram.
Discussion of the ease of processing scale brings up the A, B, and C sectors
of the pennant. They are in general highly interpretable, both in Strindberg’s
case and, as will be seen, in Persson’s. While the sector lines were drawn by me
and not an algorithm, there are good qualitative reasons for putting them about
where they are. They reflect differences in the ease of associating authors in the
pennant with Strindberg, based on the specificity of what their works imply in that
context. It is not that any author is more specific than another as a name. Rather,
as noted, the names of authors designate oeuvres, and it is works in those oeuvres
that differ in the specificity of their relevance to Strindberg studies.
The ease of processing scale is actually based on the idf measure mentioned
above. Sparck Jones (1972) created idf as a measure for weighting the
“statistical specificity” of terms. The idf measure elevates terms of any sort
(here, author names) that occur relatively infrequently in a database, because that
is taken to indicate specificity. In information retrieval, from which idf comes,
more “statistically specific” terms are given higher weights so that documents
tagged with them are placed higher in a relevance ranking—the system’s
prediction that they are more relevant to a query. The opposite is true of terms
that occur relatively frequently, because such terms are taken to be more general,
more nebulous, less indicative of exact content. The idf measure pushes them
down in the rankings as probably less relevant to a query.
The idf measure does a similar thing here. Authors cited relatively
infrequently, and who are often not well known, are placed high on the ease of

74
processing scale. They turn out to be authors whose works refer to Strindberg
or his intellectual world in obvious ways—at the level of titles, subtitles, and
chapter headings. In contrast, the more citations authors have, the lower they
are placed on the scale and the more famous they are likely to be to domain
experts or even the general public. This fame is indicated by the large numbers
of citations they have received independently of those they share with the seed
author. The idf measure is penalizing the frequent occurrence of their names in
the database as if it indicated vagueness and generality. And in a sense it does;
their names as cited authors imply countless things. It is this very breadth of
implication that makes them relatively hard to relate to the seed.
To get down to cases in Figure 1, the authors in sector A are uniformly
associated with critical studies, biographies, and translations of Strindberg. (You
can look them up.) While I have not verified their nationalities, I think many
are Swedish. To readers not immersed in Strindberg studies, probably all are
unfamiliar. I recognized one: Elizabeth Sprigge, who appears as “Sprigge E” in
the top left corner. I know her because she is the translator in my Anchor
Books edition of Strindberg’s plays. The English version of The Red Room
mentioned at the outset is also hers. Although sector A authors have published
books and articles that are easy to relate to Strindberg, they are too specialized
to have high citation counts in other contexts, which is why they automatically
go to the top here.
Sector B includes many authors who, in contrast to sector A names, are well
known indeed. Those who are not may appear in sector B rather than A
because their citation counts have been increased by authors whose names are
homonyms of theirs. (I did not attempt to disambiguate homonyms.) Michael
Robinson, John Ward, and Walter Johnson, for example, are Strindberg
scholars whose names in last-name-and-initial style lend themselves to
conflation. Michael Meyer is perhaps the best known of all scholars associated
with Strindberg and Ibsen; he wrote biographies of both and is a leading
translator of Ibsen (by whose name he appears). But “Meyer M,” too, could
reflect an inflated count. Such doubts aside, it is evident that sector B includes
Strindberg’s world-class authorial peers. Other than Ibsen, the two pulled
nearest to him are, fittingly enough, Ingmar Bergman and Eugene O’Neill.
Chekhov, Shaw, and Brecht are close behind; the list of playwrights includes
even Sophocles and Euripides. Novelists co-cited with him include Joyce,
Mann, and Zola; poets, Yeats and Baudelaire; critics, Robert Brustein, Eric
Bentley, and György Lukács. Both Schoenberg, the composer, and
Schopenhauer, the philosopher, seem quite comprehensible in this context.
In Sector C some of the most famous authors in the world are mixed with
titans of literary fashion. In recent years scholars in the humanities have cited
Barthes, Foucault, Derrida, and Walter Benjamin as faithfully as they have

75
drawn breath; as a result, these four are co-cited with practically every artist
who ever lived. But looking for felicities, it makes sense that Nietzsche and
Freud are predicted to have more cognitive effects than Kant in the context of
Strindberg’s work, and Shakespeare more than Goethe.
When we get to names of this magnitude, we can draw explicit contrasts with
the names in sector A that make the differences in processing them very clear.
Which are easier to relate to Strindberg—works by sector A authors like
Strindberg’s Ghost Sonata (Egil Tornqvist), Strindbergs dramatik (Gunnar Ollen),
and Strindberg in Inferno (Gunnar Brandell); or works by Foucault, Goethe, and
Nietzsche? Shakespeare is the author of…well, nothing with “Strindberg” in
the title. The point is not that the sector C authors are irrelevant to Strindberg;
otherwise they would not be co-cited with him. It is that the relevance is much
less on the surface; it is literally harder to see.
This shows how the effects/effort ratio from Sperber & Wilson can explain
the tf*idf formula used to weight terms in classic document retrieval. The
function of idf in relevance rankings of documents is to push up documents
whose relevance is easy to see and to push down documents whose relevance is
harder to see. The tf*idf weighting of terms that I used to place authors in the
Strindberg pennant is doing much the same thing. It pushes to the top of the
pennant names like Tornqvist, Ollen, Brandell, and Sprigge, and to the bottom
names like Derrida, Freud, Benjamin, and Kant, even though in both cases the
algorithm is completely blind to the qualitative nature of the works each set of
names stands for. By the S&W criteria for relevance, works by the authors in
sector C are not irrelevant but less relevant than works by sector A authors,
because they require more effort to relate to the seed. Some inquirers will be
willing to make this extra effort, but many will be content, if they have
questions about Strindberg, to look no further than works like those by the
Strindberg experts mentioned above. Retrieval system designers are well aware
of this fact, and that is why they use tf*idf and other algorithms like it.
Retrievals of obviously relevant documents—You’re interested in Strindberg? Here’s
some stuff on Strindberg—make the designers’ systems look good to judges in
evaluation trials. Nevertheless, one still sees scholars putting considerable effort
into the pursuit of less obvious relations, such as the comparative studies
implied by the authors co-cited with Strindberg in sectors B and C.

Persson
Figure 2, Persson’s pennant, exhibits a similar structure to the one just
discussed. It is a bit less symmetrical than Strindberg’s because Persson’s
overall citation count is closer to that of his co-citees in sector A than in sector
B. His lower count also results in a shorter horizontal axis. But these minor
differences do not affect interpretation.

76
Figure 2 Pennant for authors co-cited at least 10 times with Olle Persson in Social Scisearch on Dialog, March 2009
There may be more European (and Scandinavian) researchers in Persson’s
pennant than there would be in, say, mine; if so, it seems only natural. Overall,
the authors cited with him reflect his identification with bibliometrics as
opposed to other specialties in information science. This is not to say he has no
ties to information retrieval; one sees, for example, Gerard Salton, F. W.
Lancaster, and Tefko Saracevic among his co-citees. But most of the names
connote areas of bibliometrics; there is scant evidence of links to research in,
say, information behavior or user studies.
The authors nearest Persson on the cognitive effects scale suggest he is most
identified with citation analysis. That is what the names of the four authors
most co-cited with him—in descending order, Henry Small, Eugene Garfield,
myself, and Derek J. de Solla Price—jointly connote. (In the raw data, Small
appears as both “Small HG” and “Small H”; I have combined the counts for
the two name-forms.) Persson’s most cited paper, “The intellectual base and
research fronts of JASIS 1986-1990” (1994), is an author co-citation analysis,
very much in the line of studies I am continuing here. It is also notable that
Katherine W. McCain and Loet Leydesdorff, citation analysts both, are among
those with high predicted cognitive effects in the context set by Persson. Most

77
of the other authors in sector B are mainstream information scientists (or
crossover figures) whose work reinforces the view of Persson as a
bibliometrician (not that anyone doubted it).
Three of Persson’s top co-citees in the pennant—Small, Price, and
Garfield—are in sector C, implying difficulty in relating their works to his.
Again, the idf part of my algorithm registers the fact that Small, Price, and
especially Garfield have thousands of citations beyond those they share with
Persson. (Garfield’s huge body of work is also less topically concentrated than
Small’s or Price’s.) However, if one goes to the level of the articles in which the
co-citations occur, this difficulty may be more apparent than real. Compare
these three with the other sector C authors at bottom left, Robert K. Merton,
Bruno Latour, and Thomas S. Kuhn. Their works differ much more in subject
matter from Persson’s than the works of Small, Price, and Garfield, and this
lessens their predicted cognitive effects. At the same time, these very famous
authors are highly cited in numerous disciplines besides information science,
and this tells the tf*idf algorithm that they are difficult authors to relate to
Persson.
The authors in sector A, in contrast, are easy to relate to him. They differ
little from him in research interests. I counted five co-citees who are also his
co-authors: Aksnes, Melin, Luukkonen, Tijssen, and Glänzel; all but the last are
in sector A. (My line separating sectors A and B is more arbitrary than in
Strindberg’s case.) Interestingly, Strindberg, too, has some co-authors in sector
A of his pennant—not contemporaries of his who share his bylines, but
present-day scholars who get title-page credit for editing, translating, or writing
introductions to his works. The point of bringing up co-authors is to show,
once again, how names placed in sector A imply much the same subjects as the
seed author—in the case of co-authors, identical subject matter. Sector A thus
represents narrowness of implication, and the other two sectors represent
increasing breadth.
A final illustration of high focus in sector A is Riitta Kärki at top left. She
has been co-cited with Persson 11 times (just above the threshold of 10 for
appearing in the pennant), and so is far from him on the cognitive effects scale.
However, she tops the ease of processing scale. Not only is her total citation
count quite low, but her co-citations with Persson involve just one article of
hers (Kärki 1996) and one article of his (Persson, 1994), both of them author
co-citation analyses. This echoes the claim in White (2007a) that “ease of
processing” may mean not simply obvious connections of subject matter but
also small oeuvres and relative brevity of content.

78
Background Notes
Schneider, Larsen, & Ingwersen (2007), a PowerPoint presentation available on
the Web, is a good guide to interpreting pennant diagrams of various kinds.
However, the reasoning behind them is complicated, since it must tack between
ideas from information science and Sperber & Wilson’s relevance theory. Also,
some unconventional techniques of analysis are involved. What follows is an
attempt to explain these matters in brief.
Pennants are scatterplots of points representing the cognitive effects and
processing effort of terms in the context of a seed term. They begin with two
sets of frequency counts: (1) the number of times each term in a distribution
co-occurs with a seed term, and (2) the number of documents in the database
in which each of those terms occurs. The first set of counts is labeled “term
frequencies” or tf. The second set is labeled “document frequencies” or df. The
tf count is used in a formula to operationalize “cognitive effects” from RT. The
df count is used in another formula to operationalize RT’s “processing effort.”
White (2007a) tells how to obtain both sets of counts from databases on
Dialog, which usually is easy to do. It involves forming the set of all documents
in which the seed term appears and ranking the terms in those documents with
Dialog’s Rank command. These are moves in the tradition of Persson’s 1986
article on “online bibliometrics,” and they can generate all the standard core-
and-scatter (i.e., bibliometric) distributions. (It is remarkable that, in many
databases, Dialog supplies the exact data needed to produce pennants but
makes no further use of them that I am aware of.)
The Dialog results can be copied into DeltaGraph, which is statistical
charting software. (Excel, unfortunately, is not usable.) Examples of raw and
derived values from a DeltaGraph spreadsheet for the Strindberg pennant
appear in Table 1. A judgment sample of authors in sectors A, B and C have
been sorted by their values in the “Sector %” column, which conveys the sharp
differences in the tf/df ratio over the three sectors. These differences may also
often be sensed in the increasing recognizability of author names from Eklund
to Foucault.
Table 1 Sample data for making and interpreting the Strindberg pennant
Count Count
Name Sector % Log tf Log idf Weight
with seed overall
Strindberg A 623 623 100.0 3.79 3.68 13.97
Eklund T 22 25 88.0 2.34 5.08 11.90
Lamm M 38 87 43.7 2.58 4.54 11.71
Brandell G 30 59 50.8 2.48 4.71 11.66
Sprinchorn E 34 80 42.5 2.53 4.57 11.58
Tornqvist E 28 88 31.8 2.45 4.53 11.09

79
 Count Count
Name Sector % Log tf Log idf Weight
with seed overall
Ibsen H 60 884 6.8 2.78 3.53 9.81
Meyer M 40 1027 3.9 2.60 3.47 9.02
Bergman I 26 578 4.5 2.41 3.72 8.97
Robinson M 23 554 4.2 2.36 3.73 8.82
ONeill E 22 533 4.1 2.34 3.75 8.79
Szondi P 22 836 2.6 2.34 3.55 8.33
Nietzsche F 40 9049 0.4 2.60 2.52 6.56
Shakespeare W 34 12727 0.3 2.53 2.37 6.01
Freud S 29 14547 0.2 2.46 2.31 5.70
Foucault M 22 19213 0.1 2.34 2.19 5.14
tf df (tf/df) * 100 1+Log(tf) Log(3mil/df) tf*idf

Manning & Schütze (1999) and Jurafsky & Martin (2000) suggest converting tf
and/or idf counts to logarithms to damp the original values. My untested (but
plausible) hypothesis is that logarithmic values are truer to our sense of
discriminable differences in both the cognitive effects of terms and the effort
needed to process them. I use the version of tf*idf weighting given in Manning
& Schütze (1999). For the ith term in document j:
weight(i,j) = (1 + log(tfi,j))*log(N/dfi)
where all term counts ≥ 1, logarithms are base 10, and N is the total number of
documents in the collection. In the present study I used 3 million as the value
for N in Arts & Humanities Search for Strindberg and Social Scisearch for
Persson. The scale values on the axes of both the Strindberg and Persson
pennants are base-10 logs.
Multiplying tf by an inverse measure, idf, corresponds to dividing cognitive
effects by processing effort. However, since idf values are inverse—high when
processing effort is low and low when it is high—it reduces mental gymnastics
to rename the idf scale “ease of processing”; then high idf means “easy” and
low idf means “difficult.”
Pennants can be used to show the effect of the tf*idf multiplication—indeed,
that was one of the main points of White (2007a)—but it should be noted that
they show tf and idf plotted separately on the two axes, as in Figures 1 and 2.
Pennants thus allow the predicted cognitive effects and processing effort of
each data point to be simultaneously read.
It should also be noted that tf is used differently here from its use in
information retrieval (IR). Here, it refers to terms in a bibliometric distribution
that are rank-ordered by the tf count. There, it is used to weight terms in
queries that searchers put to large collections of documents; it designates the
number of times each query term occurs in each document. In pennants, the

80
entire set of bibliographic records formed in Dialog is considered one big
document, and tf designates how frequently terms in that big document co-
occur with the seed term.
The meaning of document frequency or df also differs somewhat in my RT-
influenced line of research as against traditional information retrieval. When I
use “document frequency” with a bibliometric distribution generated by the
Rank command, it refers to how frequently each term in a large distribution of
terms occurs in a document in the database. In IR, “document frequency”
refers to the number of documents in the collection that contain a given search
term.
Even so, I do not see these differences as major, because my purpose in
adapting the tf*idf formula to bibliometric distributions is to show that
relevance theory can explain its function in information retrieval in a new way.
RT holds relevance-seeking to be a basic component of human cognition
(Sperber & Wilson, 1995). IR uses the tf*idf formula to rank documents
algorithmically by their relevance to a query. If relevance is defined as in the RT
ratio in the introduction—as varying directly with the cognitive effects and
inversely with the processing effort of a communicative input—then these two
variables should be discernible in relevance rankings of documents or terms
representing them, and so, in fact, they prove to be.
Since I lacked human relevance judgments of documents to work with, I
applied the tf*idf formula (as it appears in Manning & Schütze, 1999) to
bibliometric data on documents from Dialog, recalling that the bibliometric
distributions have long been considered, as Saracevic (1975) puts it, “relevance-
related.” What I found, and have repeatedly confirmed, is that the frequencies
of term co-occurrences with the seed term (tf) are a promising model of the
cognitive effects of those terms in that context. More interestingly, the inverse
document frequency (idf) measure is a promising model of the effort of
processing the same terms in that context.
Interpretations like mine may seem to read too much into the tf*idf formula,
a mechanical procedure. I would counter that the verbal parts of bibliometric
data (White, 2005) most need detailed, qualitative analysis when complicated
and somewhat novel concepts are being presented, as here. Despite my
somewhat poetic approach, I think the predictions sketched in my relevance-
theoretic work, starting with White (2007a, b), are empirically testable. The
testing will probably require someone more grounded in experimental research
than I. My goal at present is simply to interest researchers in using bibliometric
data psychologically. Relevance theory, which Sperber & Wilson have
consciously aligned with cognitive science, seems like a good place to begin
looking for theoretical foundations.

81
References
Goatly, A. (1997). The language of metaphors. London and New York:
Routledge.
Kärki, R. (1996). Searching for bridges between disciplines—An author
cocitation analysis on the research into scholarly communication. Journal of
Information Science, 22, 323-334.
Jurafsky, D., & Martin, J.H. (2000). Speech and language processing: An
introduction to natural language processing, computational linguistics, and
speech recognition. Upper Saddle River, NJ: Prentice Hall.
Manning, C.D., & Schütze, H. (1999). Foundations of statistical natural
language processing. Cambridge, MA: MIT Press.
Persson, O. (1986). Online bibliometrics—A research tool for every man.
Scientometrics, 10, 69-75.
Persson, O. (1994). The intellectual base and research fronts of JASIS 1986-
1990. Journal of the American Society for Information Science, 45, 31-38
Saracevic, T. (1975). Relevance: A review of and framework for the thinking
on the notion in information science. Journal of the American Society for
Information Science, 26, 321-343.
Schneider, J.W., Larsen, B., & Ingwersen, P. (2007). Pennant diagrams, what is
it [sic], what are the possibilities and are they useful? Presentation at the
12th Nordic Workshop in Bibliometrics and Research Policy, Copenhagen,
Denmark, September 13-14, 2007. Retrieved April 15, 2009 from
www.db.dk/nbw2007/files/2c_Peter_Ingwersen.pdf
Sparck Jones, K. (1972). A statistical interpretation of term specificity and its
application to retrieval. Journal of Documentation, 28, 11-21.
Sperber, D., & Wilson, D. (1995). Relevance: Communication and cognition.
(2d ed.) Oxford, UK, and Cambridge, MA: Blackwell.
White, H.D. (2005). On extending informetrics: An opinion paper.
Proceedings of ISSI 2005, the 10th International Conference of the
International Society for Scientometrics and Informetrics. Stockholm,
Sweden: Karolinska University Press. Vol. 2: 442-449.
White, H.D. (2007a). Combining bibliometrics, information retrieval, and
relevance theory. Part 1: First examples of a synthesis. Journal of the
American Society for Information Science and Technology, 58, 536-559.
White, H.D. (2007b). Combining bibliometrics, information retrieval, and
relevance theory. Part 2: Some implications for information science. Journal
of the American Society for Information Science and Technology, 58, 583-
605.
White, H.D. (2009). Some new tests of relevance theory in information science.
Paper accepted for the Proceedings of ISSI 2009, the 12th International

82
 Conference of the International Society for Scientometrics and
Informetrics. (In print.)
Wilson, D. (2007). Relevance: the cognitive principle. Lecture 3 of Pragmatic
Theory (PLIN2002) 2007-08. Retrieved April 15, 2009 from
http://www.phon.ucl.ac.uk/home/nick/content/pragtheory/PRAG3.doc
___________________________

Address of congratulating author:

HOWARD D. WHITE
College of Information Science and Technology, Drexel University,
Philadelphia, PA 19104 (USA)
E-mail: whitehd@drexel.edu

83
The Bibliography of Professor Olle Persson

Järvelin, Kalervo; Persson, Olle, The DCI Index: Discounted Cumulated

Impact-Based Research Evaluation. // Journal of the American Society for
Information Science and Technology. - 2008 (59) : 9, s. 1433-1440
Persson, Olle, Exploring the analytical potential of comparing citing and cited
source items. // Scientometrics. - 2006 (68) : 3, s. 561-572
Persson, Olle, Bibliometric survey of Swedish health economics research. //
Scandinavian Journal of Public Health : An Evaluation of Swedish Health
Economics Research. - 2006 : Supplement 68, s. 21-25
Persson, Olle, Bibliometri - aktuella trender. // InfoTrend : Nordisk Tidskrift
for Informationsspecialister. - 2006 (61) : 4, s. 107-111
Cornelius, Barbara; Landstrom, Hans; Persson, Olle, Entrepreneurial Studies:
The Dynamic Research Front of a Developing Social Science. //
Entrepreneurship Theory and Practice. - 2006 (30) : 3, s. 375-398
Cornelius, Barbara; Persson, Olle, Who's who in venture capital research. //
Technovation. - 2006 (26) : 2, s. 142-150
Persson, Olle, "Citation Indexes for Science" - A 50 year citation history. //
Current Science. - 2005 (89) : 9, s. 1503-1504
Dastidar, Prabir G; Persson, Olle, Mapping the global structure of Antarctic
research vis-à-vis Antarctic Treaty System. // Current Science. - 2005 (89) :
9, s. 1552-1554
Glenisson, Patrick; Glänzel, Wolfgang; Persson, Olle, Combining full-text
analysis and bibliometric indicators. A pilot study. // Scientometrics. - 2005
(63) : 1, s. 163-180
Persson, Olle; Glänzel, Wolfgang; Danell, Rickard, Inflationary bibliometric
values: The role of scientific collaboration and the need for relative
indicators in evaluative studies. // Scientometrics. - 2004 (60) : 3, s. 421-432
Persson, Olle; Danell, Rickard, Decomposing national trends in activity and
impact. : A study of Swedish neuroscience papers. // Handbook of
Quantitative Science and Technology Research. : The Use of Publication
and Patent Statistics in Studies of S&T Systems, Moed, Henk F,; Glänzel,
Wolfgang; Schmoch, Ulrich (ed.). - 2004. - S. 515-528
Meyer, Martin; Pereira, Tiago Santos; Persson, Olle; Granstrand, Ove, The
scientometric world of Keith Pavitt. A tribute to his contributions to
research policy and patent analysis. // Research Policy. - 2004 (33) : 9, s.
1405-1417
Glänzel, Wolfgang; Danell, Rickard; Persson, Olle, The decline of Swedish
neuroscience: Decomposing a bibliometric national science indicator. //
Scientometrics. - 2003 (57) : 2, s. 197-213

85
Danell, Rickard; Persson, Olle, Regional R&D activities and interactions in the
Swedish Triple Helix. // Scientometrics. - 2003 (58) : 2, s. 205-218
Persson, Olle: All author citations versus first author citations//
SCIENTOMETRICS – 2001 (50) 2, s. 339-344
Melin, Goran, Olle Persson & Rickard Danell: A bibliometric mapping of the
scientific landscape on Taiwan// ISSUES & STUDIES – 2000 (36) 5, s. 61-
82
Persson, Olle: A tribute to Eugene Garfield - Discovering the intellectual base
of his discipline// CURRENT SCIENCE – 2000 (79); 5, s. 590-591
Mahlck, Paula & Olle Persson: Socio-bibliometric mapping of intra-
departmental networks// SCIENTOMETRICS – 2000 (49); 1, s. 81-91
Persson, O: Luukkonen, T; Hälikkä, S: A Bibliometric Study of Finnish
Science – 2000, VTT Group for Technology Studies: Helsinki
Meyer, Martin & Olle Persson: Nanotechnology - Interdisciplinarity, patterns
of collaboration and differences in application//SCIENTOMETRICS –
1998 (42); 2, s. 195-205
Beckmann, Martin & Olle Persson: The thirteen most cited journals in
economics// SCIENTOMETRICS – 1998 (42); 2, s. 267-271
Danell, Rickard, Lars Engwall & Olle Persson: The first mover and the
challenger: The relationship between two journals in organization
research// SCIENTOMETRICS – 1997 (40); 3, s. 445-453
Melin, Goran; Persson, Olle: Hotel cosmopolitan: A bibliometric study of
collaboration at some European universities// JOURNAL OF THE
AMERICAN SOCIETY FOR INFORMATION SCIENCE - 1998
Persson Olle; Melin, Goran; Danell, Rickard; Kaloudis, Aris: Research
collaboration at Nordic universities// SCIENTOMETRICS – 1997 (39); 2,
s. 209-223
Persson, Olle; Melin, Goran: Equalization, growth and integration of
science//SCIENTOMETRICS – 1996 (37); 1, s. 153-157
Melin, Goran; Persson, Olle: Studying research collaboration using co-
authorships// SCIENTOMETRICS – 1996 (36); 3, s. 363-377
Persson, Olle: ISI miscount?//NATURE - 1996, (380); 6570, s. 100-100
Persson, Olle; Beckmann, Martin: Locating the network of interacting authors
in scientific specialties// SCIENTOMETRICS – 1995 (33); 3, s. 351-366
Persson, Olle: The intellectual base and research fronts of JASIS 1986-1990//
JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION
SCIENCE – 1994 (45); 1, s. 31-38
Luukkonen, T; Tijssen, RJW; Persson, O; Sivertsen, G: The measurement of
international scientific collaboration// SCIENTOMETRICS – 1993 (28); 1,
s. 15-36

86
Andersson, Ake E. & Olle Persson: Networking scientists// The Annals of
Regional Science – 1993 (27) 1, s. 11-21
Luukkonen, T; Persson, O; Sivertsen, G: Understanding patterns of
international scientific collaboration// SCIENCE TECHNOLOGY &
HUMAN VALUES – 1992 (17); 1, s. 101-126
Luukkonen, Tertuu, Olle Persson, Gunnar Sivertsen: Nordic Collaboration in
Science: A Bibliometric Study – 1991, Nordic Council of Ministers
Broady, Donald; Persson, Olle: Bourdieu in the USA - bibliometric notes//
SOCIOLOGISK FORSKNING – 1989 (26); 4, s. 54-73
Hoglund, Lars; Persson, Olle: Communication within a national r-and-d-system
- a study of iron and steel in Sweden// RESEARCH POLICY – 1987 (16);
1, s. 29-37
Hoglund, Lars; Persson, Olle: The use of knowledge: a long term research
program – 1987, Inforsk: Umeå
Persson, Olle: Online bibliometrics - a research tool for every man//
SCIENTOMETRICS – 1986 (10); 1-2, s. 69-75
Persson, Olle: Graphing online searches with Lotus 1-2-3// DATABASE –
1986 (9); 2, s. 57-59
Persson, Olle: Scandinavian social-science in international journals// SOCIAL
SCIENCE INFORMATION STUDIES – 1985 (5); 4, s. 185-190
Persson, Olle: Sociologists and their classics// SOCIOLOGISK
FORSKNING – 1985 (22); 1, s. 89-90
Ellis, D; Roberts, N; Hounsell, D; Saracevic, T; Persson, O: Information man
or information action as a heuristic for information studies - comments on
the 2 positions// SOCIAL SCIENCE INFORMATION STUDIES – 1985
(5); 1, s. 25-32
Persson, Olle: Informell kommunikation bland forskare och tekniker – 1980,
Sociologiska inst: Umeå universitet
Persson, Olle: Informell kommunikation bland forskare och tekniker – 1980,
Sociologiska inst: Umeå universitet
Hoglund, Lars; Persson, Olle: Kommunikation inom vetenskap och teknik –
1980, Reports from the Department of Sociology
Hoglund, Lars; Persson, Olle: Datorbaserad litteratursokning: två fallstudier –
1980, Sociologiska inst: Umeå universitet
Hoglund, Lars; Persson, Olle: Gatekeeperfunktioner och IoD-verksamhet –
1978, Sociologiska inst: Umeå universitet
Hoglund, Lars; Persson, Olle: A survey of studies on use and production of
scientific and technical information – 1978, Research reports from the
Department of sociology: University of Umeå

87
Hoglund, Lars; Persson, Olle: Information use within an applied technical field
– 1978, Research reports from the Department of sociology: University of
Umeå
Hoglund, Lars; Persson, Olle: Informationsutnyttjande inom ett tekniskt
specialområde – 1977, Sociologiska inst: Umeå universitet
Hoglund, Lars; Persson, Olle: Evaluation of a Computer-Based Current
Awareness Service for Swedish Social Scientists – 1975, Research reports
from the Department of sociology: University of Umeå
Hoglund, Lars; Persson, Olle: Evaluation of a Computer-based Current
Awareness Service for Swedish Social Scientists – 1975, Dept. of Sociology:
University of Umeå
Hoglund, Lars; Persson, Olle: Transportkonsumtionen i sociologisk belysning:
en explorativ analys av bilinnehav och bilanvändning – 1973, Sociologiska
inst: Umeå universitet

88
Addendum

Finally, it is our pleasant duty to announce the names of those colleagues and
friends of Olle who were not able to contribute to this volume, but who have
expressed their wish to congratulate Olle Persson on the occasion of his
birthday. Herewith, we kindly acknowledge best wishes expressed by

DAG AKSNES
NIFU-STEP (Norway)

LENNART BJÖRNEBORN
Royal School of Library and Information Science, Copenhagen
(Denmark)

BLAISE CRONIN
University of Indiana (U.S.A)

PETER INGWERSEN
Royal School of Library and Information Science, Copenhagen
(Denmark)

GÖRAN MELIN
Ministry of Education and Research (Sweden)

ED NOYONS
University of Leiden (Netherlands)

BALÁZS SCHLEMMER
ISSI administration (Hungary)

HENRY SMALL
Thomson Reuters (U.S.A)

ROBERT TIJSSEN
University of Leiden (Netherlands)

IRENE WORMELL
Informatiker Konsult AB (Sweden)

89