A Software Tool for lnformetric Analysis of Citation Linkage

Can a computer system create raw data for writing the history of science? More than 40 years ago, a report about ,,The use of citation data in writing the history of science" was published (Garfield, Sher & Torpie, 1964). It concludes (p. 33), ,,it is felt that citation analysfs has been demonstrated to be a valid and valuable means of creating accurate historical descriptions of scientific fields, especially beyond the first quarter of the twentieth century when bibliographic citation had become well established as part of scientific publication". This report, produced by the Institute of Scientific Information (ISI) makes use of the Science Citation Index to study the history of science. Subsequently, in the early 1970s Henry Small developed his co-citation method for clustering documents (Small, 1973; Small & Griffith, 1974). These methods permitted disciplinary and national mappings a kind of ,,geography of science" (Small & Garfield, 1986), that is, ,,scientography" (Garfield, 1986; Garfield, 1994) and the product ,,IS1 Atlas of Science". ,,Our maps represent where researchers have been. As historical records, then, these maps are surveys of the geography of scientific ideas and discoveries intellectual gazetteers, if you will for a given year" (Garfield, 1986, p. 3). Algorithmic historiography makes use of the ,,axiom", that bibliographic records (including cited references) contain concise information about scientific (or other historically relevant) content. ,,In our conception of facilitating historiography -that is, writing the history of modern science -we make the basic assumption that the bibliographic information contained in a collection of published scientific articles is sufficient for the purpose of recapturing the historiographic structure of the field (Garfield, Pudovkin & Istomin, 2003, p. 400). Analyzing science in this way leads to three types of citation analysis, (1st) to directed graphs of information flows from an information sender to an information receiver or vice versa of reputation flow from the receiver to the sender and (2nd) to undirected graphs derived from co-citations or bibliographic coupling. Both methods allow one to monitor (3rd) scientific developments by (a) comparing the science maps year by year or by (b) analyzing the citation history of a knowledge domain. So Henry Small could describe a scientific revolution in the sense of Thomas S. K u h n (1962) by the means of co-citation analysis (Small, ign). HistCite is a software tool for analyzing and visualizing direct citation linkages between scientific papers. Its inputs are bibliographic records (including cited references) from Thomson Scientific's Web ofKnowledge (WOK) or other sources. Its outputs are various tables and graphs with informetric data about the knowledge domain under study. HistCite covers types 1 and 3(b) of the above mentioned methods of citation analysis. It does not utilize co-citation mapping (kind 2 and 3(a)). Here another informetric software tool, e.g. Citespace, is applicable (Chen, 2004; Chen, 2006). What is the ,,philosophy" behind algorithmic historiography of science? What role does HistCite play in wTiting history of science or -more precisely in analyzing a specific knowledge domain? HistCite was designed originally to enable scholars to create genealogical microhistories of authors or topics "automatically" that is, algorithmically. In its further development it became apparent that it could be used for many purposes too numerous to mention here. In the library, given a key word or subject heading one can search WoKfor a group of papers on a topic and then use Histcite to identify the most-cited core papersusually from 25 to 50. From this HistCite generates tables and historiographs showing the evolution of the field.