Various methods for the mapping of science

The dynamic mapping of science using the data in theScience Citation Index was put on the research agenda of science studies byDe Solla Price in the mid 1960s. Recently, proponents of ‘co-citation cluster analysis’ have claimed that in principle their methodology makes such mapping possible. The study examines this claim, both methodologically and theoretically, in relation to other means of mapping science. A detailed study of a co-citation map, its core documents' citation patterns and the related journal structures, is presented. At these three levels of possible study of aggregates of citations, an analysis is pursued for the years 1978 to 1984. The many different statistical methods which are in use for the analysis of the respective datamatrices—such as cluster analysis, factor analysis and multidimensional scalling—are assessed with a view to their potential to contribute to a better undérstanding of the dynamics at the different levels in relation to each other. This will lead to some recommendations about methods to use and to avoid when we aim at a comprehensive mapping of science. Although the study is pursued at a formal and analytical level, in the conclusions an attempt is made to reflect on the results in terms of further substantial questions for the study of the dynamics of science.     

Mapping the backbone of science

Summary This paper presents a new map representing the structure of all of science, based on journal articles, including both the natural and social sciences. Similar to cartographic maps of our world, the map of science provides a bird’s eye view of today’s scientific landscape. It can be used to visually identify major areas of science, their size, similarity, and interconnectedness. In order to be useful, the map needs to be accurate on a local and on a global scale. While our recent work has focused on the former aspect,1 this paper summarizes results on how to achieve structural accuracy. Eight alternative measures of journal similarity were applied to a data set of 7,121 journals covering over 1 million documents in the combined Science Citation and Social Science Citation Indexes. For each journal similarity measure we generated two-dimensional spatial layouts using the force-directed graph layout tool, VxOrd. Next, mutual information values were calculated for each graph at different clustering levels to give a measure of structural accuracy for each map. The best co-citation and inter-citation maps according to local and structural accuracy were selected and are presented and characterized. These two maps are compared to establish robustness. The inter-citation map is then used to examine linkages between disciplines. Biochemistry appears as the most interdisciplinary discipline in science.     

Comparative indicators of interdisciplinarity in modern science

A set of scientometric indicators of interdisciplinary links between advancing fields of biomedicine is suggested. Twenty jounals listed in theJCR of theSCI for 1988 are analyzed. An index of interdisciplinarity for a given journal is calculated as the sum of ratios between the numbers of journals from all other disciplines (except for general-scientific and miscellaneous journals) and from the same discipline cited by that journal or citing it, and of ratios between the numbers of citations to and by these journals. Some interdisciplinary patterns of 20 andrology journal articles are scientometrically assessed, too. The combined usage of this method with co-classification and co-citation methodology can optimize interdisciplinarity evaluation and promotion.     

Journal co-citation analysis of semiconductor literature

The purpose of this study is to map semiconductor literature using journal co-citation analysis. The journal sample was gathered from the INSPEC database from 1978 to 1997. In the co-citation analysis, the data compiled were counts of the number of times two journal titles were jointly cited in later publications. It is assumed that the more two journals are cited together, the closer the relationship between them. The journal set used was the 30 most productive journals in the field of semiconductors. Counts of co-citations to the set of semiconductor journals were retrieved from SciSearch database, accessed through Dialog. Cluster analysis and multi-dimensional scaling were employed to create two-dimensional maps of journal relationships in the cross-citation networks. The following results were obtained through this co-citation study: The 30 journals fall fairly clearly into three clusters. The major cluster of journals, containing 17 titles, is in the subject of physics. The second cluster, consisting of 9 journals, includes journals primarily on material science. The remaining cluster represents research areas in the discipline of electrical and electronic engineering. All co-cited journals share similar co-citation profiles, reflected in high positive Pearson correlation. Two hundred and ninety-six pairs (68%) correlate at greater than 0.70. This shows that there is strong relationship between semiconductor journals. Five individual journals in five paired sets with co-citation frequency over 100,000 times include Physical Review B, Condensed Matter; Physical Review Letters; Applied Physics Letters; Journal of Applied Physics; and Solid State Communications. This revised version was published online in August 2006 with corrections to the Cover Date.     

Specialties and disciplines in science and social science: An examination of their structure using citation indexes

The technique of co-citation cluster analysis is applied to a special three-year (1972–1974) file of theSocial Sciences Citation Index. An algorithm is devised for identifying clusters which belong to a discipline based on the percentage of source documents which appear in a disciplinary journal set. Clusters in three disciplines (economics, sociology and psychology) are identified using this algorthm. Clusters in a specialty of natural science (particle physics) obtained from the 1973Science Citation Index are compared and contrasted with the three groups of social sciences clusters. Certain common structural characteristics of the social science and natural science groups suggest that knowledge is developing in parts of the social science disciplines in a manner similar to the natural sciences.Prepared for presentation at the joint meeting of The Society for Social Studies of Science and the Research Committee on the Sociology of Science of the International Sociological Association, Cornell University, Ithaca, New York, November 4–6, 1976.     

Citation networks in information science

The method of co-citation analysis is used to build citation networks in information science. As data base the first 13 volumes (1961–1973) of the leading Soviet journal in the field (Nauchno-tekhnicheskaya Informatsiya) were used. The results reveal the topical structure of information science, the communities of authors and the names of single leading scientists. The evaluation of scientists' work is based on two measures: productivity (with or without co-authorship) and popularity (popularity of authors and popularity of papers).     

A journal co-citation analysis of library and information science in China

This paper aims to reveal the relationship and structure of library and information science (LIS) journals in China. 24 core LIS journals in China are selected and the relevant data of journal co-citation are retrieved from Chinese Journal Full-Text Database constructed by China National Knowledge Infrastructure during the period of 1999–2009. By calculating mean co-citation frequencies and correlation coefficients, we find that there is a strong relationship among LIS journals in China. Utilizing the methods of cluster analysis, multidimensional scaling analysis and factor analysis, we analyze the data of journal co-citation. LIS journals in China are divided into four clusters. The relatedness among journals is shown manifestly through their locations in the two-dimensional map. A three-factor solution is obtained with the factor loading of each journal. Finally, we interpret and discuss the results to get some conclusions and also expect to describe the network characters of journal co-citation in future research.     

Reassessment of co-citation methods for science indicators: Effect of methods improving recall rates

Although co-citation techniques are very powerful structuring tools, the use of science policy indicators based on co-citation has often been criticized, especially on ISI research fronts. A major issue is the small fraction of literature retrieved, i.e. the recall rate problem. Our investigations indicate that at the level of micro/meso studies high recall rates can be achieved by (a) the use of appropriate clustering techniques limiting singletons and (b) the enrichment of cocited cores by medium-cited items. This combination of appropriate clustering and extension of recall proves to be efficient, provided that careful trade-offs are sought between the extension and relevance of recall. It leads to a reassessment of the performance of the co-citation approach for structuring scientific fields and providing related indicators not limited to the leading edge. It also opens new opportunities for comparison/combination with other relational methods such as co-word analysis.     

A co-citation analysis of representative authors in philosophy: Examining the relationship between epistemologists and philosophers of science

The relation between philosophy of science and epistemology is studied using the author co-citation technique. Co-citation links among 62 authors – a representative list of various styles and approaches to rationality – were established using the Arts and Humanities Citation Index. Multidimensional scaling results in a two-dimensional map of authors, where the axes represent the subject (philosophy of science to epistemology) and the method (qualitative to quantitative), respectively. The authors on the map can be clustered into more or less coherent groups at different levels of resolution.     

Intellectual structure of Antarctic science: A 25-years analysis

To delineate the intellectual structure of Antarctic science, the research outputs on Antarctic science have been analyzed for a period of 25 years (1980–2004) through a set of scientometrics and network analysis techniques. The study is based on 10,942 records (research articles, letters, reviews, etc.), published in 961 journals/documents, and retrieved from the Science Citation Index (SCI) database. Over the years interest in Antarctic science has increased, as is evident from the growing number of ratified countries and research stations. During the period under study, the productivity has increased 3-times and there is a 13-fold increase in collaborative articles. Attempt has been made to identify important players like scientists, organizations and countries working in the field and to identify frontier areas of research that is being conducted in this continent. The highest 41% scientific output is contributed by the USA and the UK, followed by Australia and Germany. British Antarctic Survey (BAS), UK and Alfred Wegener Institute of Polar & Marine Research, Germany are the most productive institutes in Antarctic science. Maximum number of research articles on Antarctic science, have been published in the journal Polar Biology, indicating substantial work being done on the biology of this continent. The journals — Nature and Science — are the highly-cited journals in Antarctic science. The paper written by J. C. Farman et al., published in Nature in 1985, reporting depletion of ozone layer, is the most-cited article. Semantic relationships between cited documents were measured through co-citation analysis. J. C. Farman and S. Solomon are co-cited most frequently.     

Citation context analysis of a co-citation cluster: Recombinant-DNA

The techniques of co-citation clustering and citation context analysis are combined to concretely define the shared knowledge within a research specialty. The cluster for a large and fast moving biomedical specialty, recombinant-DNA, is presented in terms of the highly cited documents comprising it and their co-citation links. By examining citation contexts in the papers citing the highly cited documents, it is possible to label each of the documents in the cluster with its specific cognitive meaning for the citing authors. Co-citation contexts are used to reveal the relationships among the concepts symbolized by the highly cited documents, providing a cognitive equivalent of the co-citation links. This may open a new way to the investigation of the logic of conceptual change at the specialty level.     

Hybrid documents co-citation analysis: making sense of the interaction between science and technology in technology diffusion

The paper presents a methodology called hybrid documents co-citation analysis, for studying the interaction between science and technology in technology diffusion. Our approach rests mostly on patent citation, cluster analysis and network analysis. More specifically, with the patents citing Smalley RE in Derwent innovations index as the data sets, the paper implemented hybrid documents co-citation network through two procedures. Then spectrum cluster algorithm was used to reveal the knowledge structure in technology diffusion. After that, with the concordance between network properties and technology diffusion mechanisms, three indicators containing degree, betweenness and citation half-life, were calculated to discuss the basic documents in the pivotal position during the technology diffusion. At last, the paper summarized the hybrid documents co-citation analysis in practise, thus concluded that science and technology undertook different functions and acted dominatingly in the different period of technology diffusion, though they were co-activity all the time.     

Maps of science as interdisciplinary discourse: co-citation contexts and the role of analogy

Interdisciplinarity can be manifest in many forms: through collaboration or communication between scientists working in different fields or through the work of individual scientists who employ concepts or methods across disciplines. This latter form of interdisciplinarity is addressed here with the goal of understanding how ideas in different fields come together to create new opportunities for discovery. Maps of science are used to suggest possible interdisciplinary links which are then analyzed by co-citation context analysis. Interdisciplinary links are identified by juxtaposing a clustering and mapping of documents against a journal-based categorization of the same document clusters. Links between clusters are characterized as interdisciplinary based on the dissonance of their category assignments. To verify and probe more deeply into the meaning of interdisciplinary links, co-citation contexts for selected links from five separate cases are analyzed in terms of prominent cue words. This analysis reveals that interdisciplinary connections are often based on authors’ perceptions of analogous problems across scientific domains. Cue words drawn from the citation contexts also suggest that these connections are viewed as important and ripe with both opportunity and risk.     

Tracking and predicting growth areas in science

Summary We explore the possibility of using co-citation clusters over three time periods to track the emergence and growth of research areas, and predict their near term change. Data sets are from three overlapping six-year periods: 1996-2001, 1997-2002 and 1998-2003. The methodologies of co-citation clustering, mapping, and string formation are reviewed, and a measure of cluster currency is defined as the average age of highly cited papers relative to the year span of the data set. An association is found between the currency variable in a prior period and the percentage change in cluster size and citation frequency in the following period. The conflating factor of “single-issue clusters” is discussed and dealt with using a new metric called in-group citation.     

Scientometric definition of science: In what respect is the humanities more scientific than mathematical and social sciences?

“What is science” is not only intellectually interesting but also politically crucial in the proper allocation of budget. As science does not define itself and only philosophy defines everything including science, this paper first sketches the philosophical view of science. Then, hypotheses are presented as to what definition is actually given for science by scientific circles themselves. The hypotheses are tested in a scientometric way by observing the trend in the magazine Science. Unexpected results are obtained. The actual trend in Science does not reflect what has long been considered about science. Specifically, chemistry is at the top in the number of papers, far above physics. More papers are in historical sciences (part of the humanities) than in mathematics, computer science and social science. It is discussed in what respect chemistry is the most scientific, and the humanities is more scientific than the abovementioned three scientific fields. It is interpreted that, out of the two aspects in Galilei’s view of science (metodo compositivo and metodo risolutivo.), the latter (empirical solution of problems by using technical instruments) dominates the former (systematic theory using mathematics) in Science.     

A dense network sub-grouping algorithm for co-citation analysis and its implementation in the software tool Sitkis

Summary Clustering algorithms are used prominently in co-citation analysis by analysts aiming to reveal research streams within a field. However, clustering of widely cited articles is not robust to small variations in citation patterns. We propose an alternative algorithm, dense network sub-grouping, which identifies dense groups of co-cited references. We demonstrate the algorithm using a data set from the field of family business research and compare it to two alternative methods, multidimensional scaling and clustering. We also introduce a free software tool, Sitkis, that implements the algorithm and other common bibliometric methods. The software identifies journal-, country- and university-specific citation patterns and co-citation groups, enabling the identification of “invisible colleges.”     

Combining full-text analysis and bibliometric indicators. A pilot study

Summary In the present study full-text analysis and traditional bibliometric methods are combined to improve the efficiency of the individual methods in the mapping of science. The methodology is applied to map research papers from a special issue of Scientometrics. The outcomes substantiate that such hybrid methodology can be applied to both research evaluation and information retrieval. The subject classification given by the guest-editors of the special issue is used for validation purposes. Because of the limited number of papers underlying the study the paper is considered a pilot study that will be extended in a later study on the basis of a larger corpus.     

Assessing assessments of British science. Some facts and figures to accept or decline

Conclusions After a rather comprehensive analysis of severalSCI based publication productivity indicators, it can be concluded that neither the view of a continuing decline⁵ nor that of a remarkable increase⁶ of British science in the first half of the eighties can be supported by valid bibliometric/scientometric arguments. The annual changes of any of the indicators considered had no statistical significance, and no trend distinguishable from the effect of random fluctuations could be observed. Some of the indicators showed a local minimum in 1982, the significance of which can be clarified only in the frame of a longer range study.All the abovementioned refers to British science in general. As a previous paper⁵ reported also a decline of British publication performance in some science fields and subfields these would have to be investigated one by one with the above described methodology. The present authors did investigate the research performance of British analytical chemistry which has been mentioned as a declining subfield.⁵ Our results however showed¹² that analytical chemistry is a flourishing subfield in Britain.Finaly we mention thatNederhof in a recent study on the performance of six industrial nations in the field of biotechnology insisted that his findings contradict earlier reports⁴ showing UK losses in article productivity in basic research.¹³     

Clustering thescience citation index ® using co-citations

Earlier experiments in the use of co-citations to cluster theScience citation Indey (SCI) database are reviewed. Two proposed improvements in the methodology are introduced: fractional citation counting and variable level clustering with a maximum cluster size limit. Results of an experiment using the 1979SCI are described comparing the new methods with those previously employed. It is found that fractional citation counting helps reduce the bias toward high referencing fields such as biomedicine and biochemistry inherent in the use of an integer citation count threshold, and increases the range of subject matters covered by clusters. Variable level clustering, on the other hand, increases recall as measured by the percentage of highly cited items included in clusters. It is concluded that the two new methods used in combination will improve our ability to generate comprehensive maps of science as envisioned byDerek Price. This topic will be discussed in a forthcoming paper.     

Is chemistry 'The Central Science'? How are different sciences related? Co-citations, reductionism, emergence, and posets

Summary According to a widely used introductory chemistry text by T. E. Brown et al.,¹ chemistry is 'The Central Science'. But scientometric co-citation analyses indicate that biochemistry seems presently to be more interconnected to other sciences. On the other hand, mathematics is considered by many to permeate all sciences and hence might compete as the choice for centrality. A critical commentary and argument leads to a proposal for an alternative partially ordered hierarchical “framework” map of sciences. This argument is supplemented by a scientometric approach based on university course requirements for different curricula, so as to support our partially ordered map. This alternative “framework” mapping then is seen to indicate a special position for chemistry, as where significant branching begins.