Radosvet Todorov

Representing a scientific field: A bibliometric approach

A new bibliometric method is proposed for representing links between subfields as defined by a classification scheme. The frequency of co-occurrence of articles from different subfields in selected journals is used for measuring the degree of relatedness between these subfields. The results of such quantitative analysis could be compared to the tree topology of the classification network established in a qualitative analysis. The method is applied to describe the internal links within the field of condensed matter physics using the 1984 Physics Abstracts database. A distinction is made between experimental and theoretical links on the basis of treatment codes assigned to journal articles. The links described by cluster analysis are matched against the cross-reference network of the International Classification for Physics.

Distribution of physics literature

Percentage distributions of physics papers in 36 countries over ten subfields are compared. Factor and cluster analyses are applied to data from 1978Physics Abstracts. Countries load highly on seven factors. Their meaning is estimated by factor scores. Clusters of countries with similar publishing patterns are presented in a tree diagram.

MAPPING AUSTRALIAN GEOPHYSICS: A CO-HEADING ANALYSIS

Descriptive capacities of a new bibliometric method, namely co-heading analysis, are investigated. The method uses the appearance and co-appearance of classification subdivisions (headings) in the document records of 1988 INSPEC database to display correspondingly the main topics of Australian geophysics and their links. The findings, in the form of inclusion maps (resulting from multidimensional scaling and cluster analysis) provide new insights into geophysics national activity and into its structure.

Condensed matter physics journals

On the basis of a citation/reference criterion, 20 core journals are selected in the field of condensed matter physics. Citation data and indicators from 1980Journal Citation Reports reveal their different characteristic features such as applied orientation, communication function and longevity. The manually obtained data for the core journals are written into a matrix in order to determine an appropriate ranking parameter. The method ofPrice is used first to reduce the diagonal elements and then, following the method ofGeller, influence weights are calculated for the core journals. Influence weights are determined also for non-core journals using only the references received from the core.

Determination of influence weights for scientific journals. An example from elementary particle physics

A new citation matrix is proposed for the computation of journal influence weights applying Gellers methodology. Instead of self-citations, references to fringe journals are introduced and the proportions of all references (without self-citations) are used as matrix elements. On the basis of the determined weights, relative weights for fringe journals are calculated, i.e. the initial set (core) of journals is extended and simultaneously a rank order is obtained (by total influence and influence per article). The procedure is examplified on elementary particle physics (EPP) journals. Using first a reference/citation criterion, eight core journals were determined by iteration and then influence weights were calculated. A ranking by total influence and influence per article is presented for core and fringe journals in EPP.

Displaying content of scientific journals: A co-heading analysis

A co-heading analysis is proposed for representing the subject content of scientific journals. It is based on the subject headings assigned to documents before their input in bibliographic databases. The method utilizes specifically the co-appearance of headings in the document records to display the subject scope of a given journal not only by the topics covered but by their relationships as well. This analysis is applied in superconductivity for displaying the content of some physics journals using data from 1984 Physics Abstracts. The results (in the form of inclusion maps) could help information scientists and library staff in evaluating and selecting appropriate journals.

An overview of Mike Moravcsik's publication activity in physics

A bibliometric online technique is applied on data from the INSPEC bibliographic file to describe some aspects of Moravcsiks publication activity (co-authorship, source journals, etc.). Separately, a co-occurrence method is used to represent the subject structure (the main topics and their links) of his papers in physics. The principle underlying this method is to develop a network based on common appearances of classification subdivisions (headings) as well as of controlled terms in Moravcsiks document records. The results, in the form of line and point graphs, give a global picture of Mike Moravcsiks research profile in physics.

Facts or imposed names on facts

ConclusionsAccording to the title, there is no doubt that the objective of the authors is to attract attention on the state of health of quantitative studies of science (scientometrics) and encourage research vigor. Every effort in this direction is praised in any field. At the same time, it is strange that (instead of concentrating on the patient) they have extended the set of quantitative methods without considering the purpose of their application and nature of treated data. Thus, some of the statements (e.g. growing number of publications) “seem” to be more names imposed on facts than the facts themselves.

Konstruktion von Beziehungen zwischen Klassifikationsklassen: ein quantitativer Ansatz

Es wird eine alternative Methode zur Herstellung von Querverweisen in hierarchisch strukturierten Klassifikationssystemen vorgestellt, die ohne das subjektive Wissen von Experten des jeweiligen wissenschaftlichen Spezialgebiets auskommt. Das Verfahren geht von der Grundannahme aus, das die Haufigkeit gemeinsamen Auftretens von Basiselementen eines Klassifikationssystems in den konkreten, in der Datenbasis enthaltenen Dokumenten (oder anderen Objekten) eines ausgewahlten wissenschaftlichen Fachgebiets als ein Mas fur die inhaltliche Verwandtschaft dieser Elemente angesehen werden kann. Die Methode basiert auf einer (automatischen) Auszahlung der Haufigkeiten des einfachen und gemeinsamen Auftretens der Klassifikationselemente in einer gegebenen Menge von Dokumenten und fuhrt ein angemessenes Ahnlichkeitsmas zur Bestimmung der inhaltlichen Verwandtschaft der Elemente ein. Das Verfahren wird in einem Teilbereich der internationalen Klassifikation fur Physik sowie anhand der Zeitschriftenzuordnung einer multidisziplinaren Datenbank demonstriert.

FORMAL COMMUNICATION IN SCIENCE: A MODEL BASED ON GENERALIZED NETS

A schematic overview of the formal communication process is first presented. Emphasis is placed only on the specific functions and participants required to transfer article manuscripts from the author to the reader of scientific journals. For the description of this process a mathematical model based on generalized nets (GN) is then proposed. Some advantages of the GN as compared to other models are shown. Model application is not included since the programme package which realizes the GN is in a stage of preparation.