This essay was originally published in the Current Contents print editions November 7, 1994, when Clarivate Analytics was known as Institute for Scientific Information.

While the term scientography has not been widely adopted, it is a very apt label for the mapping of science 1. Science maps serve as tools for navigating the research literature by depicting the spatial relations between research fronts, which are areas of significant activity 2. Such maps can also simply be used as a convenient means of depicting the way research areas are distributed and conveying added meaning to their relationships.

Trend Analysis

Even with a database that is completely up-to-date, we are still only able to create maps that show where research fronts have been. Because of the publication cycle, research has already moved on by the time the corresponding journal articles are published. Except in one’s own area of expertise, these maps give an otherwise unrealized view of where the action is and give a hint where it may be going. However, as we expand the size of the database from one year to a decade or more, the map created through citation analysis provides a historical, indeed historiographical, window on the field that we are investigating 3.

From a global viewpoint, these maps show relationships among fields or disciplines. The labels attached or embedded in the graphics reveal their semantic connections and may hint at why they are linked to one another. Furthermore, the maps reveal which realms of science or scholarship are being investigated today and the individuals, publications, institutions, regions, or nations currently preeminent in these areas.

Longitudinal Mapping

By using a series of chronologically sequential maps, one can see how knowledge advances. While maps of current data alone cannot predict where research will go, they can be useful indicators in the hands of informed analysts. By observing changes from year to year, trends can be detected. Thus, the maps become forecasting tools. And since some co-citation maps include core works, even a novice can instantly identify those articles and books used most often by members of the “invisible college.”

The creation of maps by co-citation clustering is a largely algorithmic process. This stands in contrast to the relatively simple but arduous manual method we used over 30 years ago to create a historical map of DNA research from the time of Mendel up to the work of Nierenberg and others 4.  Shortly afterward, Derek Price published his classic work on networks using our data as input 5.

As far back as 1948, Samuel Bradford referred to “a picture of the universe of discourse as a globe, on which are scattered, in promiscuous confusion, the mutually related, separate things we see or think about. 6” J.D. Bernal 7,  among others, created by laborious manual methods what we would today describe as historiographs. However, dynamic longitudinal mapping was made uniquely possible by the development of the ISI database. Indeed, it gave birth to scientometrics and new life to bibliometrics.

Customized Mapping

An important distinction must be made between ad hoc or customized mapping and large scale mapping exemplified by the ISI research front database. Searching this database was discussed last month 8.  On the other hand, ad hoc mapping is possible by using ISI’s SCI-Map PC software together with the SCI-Map database–a derivative of the research front database. The SCI-Map system creates maps of individual research areas specified by the user. Given an author, paper, or key word as a starting point, one can “seed” a map and then “grow” the map by specifying various desired connections at different thresholds of co-citation strength or distance 9.

Building a Structure. The network of connected nodes that constitutes a map is formed by a series of iterations of clustering, including additional core papers with each successive node. The nodes are selected according to the strength of their links, and positioning is determined by the geometric triangulation method (see  ). The nodes can be “collapsed” to allow the building of larger scale structures.

Viewing the Structure. As the network is being built, it can be viewed on SCI-Map. The magnification of the view can be increased to focus on a smaller portion of the cluster or decreased to see a broader view.

Other Systems. Other available systems include BIBMAP, which is software that performs a variable level clustering process designed for small sets of information 10,  and Bibliometrics Toolbox, which is a computer software program that ranks entries by productivity, calculates minimum Bradford zones, and graphs the bibliography. Bibliometrics Toolbox allows division of datasets into groups for analysis 11.

 

Conclusions

Scientography produces a veritable map of the invisible college “campus.” Areas of interest can be compared on many levels, and progress from year to year can be surmised. Truly one of the most interesting aspects of scientometrics, mapping can reveal a wealth of information about highly specific areas of research as well as large subspecialties or disciplines.

Dr. Eugene Garfield 
Founder and Chairman Emeritus, ISI

 

References

  1. Garfield E.Towards scientography. Essays of an Information Scientist. Philadelphia: ISI Press®, 1986, Vol. 9. p. 324.
  2. Small H, Garfield E.The geography of science: disciplinary and national mappings. J. Inform. Sci. 11:147-59, 1985.
  3. Garfield E.Co-citation analysis of the scientific literature: Henry Small on mapping the collective mind of science. Essays of an Information Scientist. Philadelphia: ISI Press, 1993, Vol. 15. p. 293-303.
  4. Garfield E, Sher I H, Torpie R J.The Use of Citation Data in Writing the History of Science, ISI® Monograph, Institute for Scientific Information®, Philadelphia, 1964.
  5. Price D J D.Networks of scientific papers. Science 149:510-5, 1965.
  6. Bradford S C.Documentation. 1948, London, Crosby Lockwood & Sons, p. 137.
  7. Bernal J D.Science and Industry in the Nineteenth Century. 1955, London, Routeledge Kegan Paul.
  8. Garfield E.Research fronts. Current Contents® (41):3-7, 10 October 1994.
  9. Small H.A SCI-Map case study: Building a map of AIDS research. Scientometrics 30(1):229-41, 1994.
  10. Persson O, Stern P, Holmberg K-G.BIBMAP: a toolbox for mapping the structure of scientific literature. In: Representations of Science and Technology. 1992, Leiden, the Netherlands, DSWO Press, p. 189-99.
  11. Brooks T A.The Bibliometrics Toolbox (Computer file). Seattle, North City Bibliometrics.

Figure 1. Cluster of papers with “Abadie V 89” (Abadie V. CPG dinucleotides are mutation hot spots in phenylketonuria. Genomics 5:936, 1989) as the seed paper.

Abadie V, et al. CPG dinucleotides are mutation hot spots in phenylketonuria. Genomics 4:936, 1989. Frequency*: 28

Avigad S, Cohen B E, Bauer S, Schwartz G, Frydman M, Woo S L C, Niny Y, Shiloh Y. A single origin of phenylketonuria in Yemenite Jews. Nature 344(6262):168-70, 1990. Frequency: 10

Daiger S P, Chakraborty R, Reed L, Fetete G, Schuler D, Berenssi G, Nasz I, Brdicka R, Kamaryt J, Pijackova A, Moore S, Sullivan S, Woo S L C. Polymorphic DNA haplotypes at the phenylalanine-hydroxylase (PAH) locus in European families with phenylketonuria (PKU). Amer. J. Hum. Genet. 45(2):310-8, 1989. Frequency: 20

Daiger S P, Reed L, Huang S S, Zeng Y T, Wang T, Lo W H Y, Okano Y, Hase Y, Fukuda U, Oura T, Tada K, Woo S L C. Polymorphic DNA haplotypes at the phenylalanine-hydroxylase (PAH) locus in Asian families with phenylketonuria (PKU). Amer. J. Hum. Genet. 45(2):319-24, 1989. Frequency: 10

Dworniczak B, Aulehlascholz C, Horst J. Phenylketonuria: Detection of a frequent haplotype-4 allele mutation. Hum. Genet. 84(1):95-6, 1989. Frequency: 16

John S W M, Rozen R, Laframboise R, Laberge C, Schriver C R. Novel PKU mutation on haplotype-2 in French-Canadians. Amer. J. Hum. Genet. 45(6):905-9, 1989. Frequency: 17

John S W M, Rozen R, Schriver C R, Laframboise R, Laberge C. Recurrent mutation, gene conversion, or recombination at the human phenylalanine-hydroxylase locus: Evidence in French-Canadians and a catalog of mutations. Am. J. Hum. Genet. 46(5):970-4, 1990. Frequency: 19

Lichterkonecki U, Schlotter M, Yaylak C, Ozguc M, Coskun T, Ozalp I, Wendel U, Batzler U, Trefz F K, Konecki D. DNA haplotype analysis at the phenylalanine-hydroxylase locus in the Turkish population. Hum. Genet. 81(4):373-6, 1989. Frequency: 10

Lyonnet S, Caillaud C, Rey F, Berthelon M, Frezal J, Rey J, Munnich A. Molecular genetics of phenylketonuria in Mediterranean countries: a mutation associated with partial phenylalanine-hydroxylase deficiency. Am. J. Hum. Genet. 44(4):511-7, 1989. Frequency: 19

Okano Y, Wang T, Eisensmith R C, Steinmann B, Gitzelmann R, Woo S L C. Missense mutations associated with RFLP haplotype-1 and haplotype-4 of the human phenylalanine-hydroxylase gene. Am. J. Hum. Genet. 46(1):18-25, 1990. Frequency: 23

Okano Y, Wang T, Eisensmith R C, Guttler F, Woo S L C. Recurrent mutation in the human phenylalanine-hydroxylase gene. Am. J. Hum. Genet. 46(5):919-24, 1990. Frequency: 12

Spiegelberg R, Aulehlascholz C, Erlich H, Horst J. A b-thalassemia gene caused by a 290-base pair deletion: Analysis by direct sequencing of enzymatically amplified DNA. Blood 73(6):1695-8, 1989. Frequency: 10

Wang T, Okano Y, Eisensmith R, Huang S Z, Zeng Y T, Lo W H Y, Woo S L C. Molecular genetics of phenylketonuria in Orientals: Linkage disequilibrium between a termination mutation and haplotype-4 of the phenylalanine-hydroxylase gene. Am. J. Hum. Genet. 45(5):675-80, 1989. Frequency: 14

Wang T, Okano Y, Eisensmith R C, Fekete G, Schuler D, Berencsi G, Nasz I, Woo S L C. Molecular genetics of PKU in eastern Europe: a nonsense mutation associated with haplotype-4 of the phenylalanine-hydroxylase gene. Somat. Cell Molec. Genet. 16(1):85-90, 1990. Frequency: 10

Woo S L C. Molecular basis and population genetics of phenylketonuria. Biochem. 28(1):1-7, 1989. Frequency: 15

* Frequency = citation count for the period 1989-91

Figure 2. Cluster of the first 10 journals with Genomics as the seed journal title.