The Institutionalization of Scientific Information : A

A SCIENTOMETRIC MODEL (ISI-S model) is introduced for describing the institutionalization process of scientific information. The central concept of ISI-S is that the scientific information published may develop with time through permanent evaluation and modification processes toward a cognitive consensus of distinguished authors of the respective scientific field or discipline. ISI-S describes the information and knowledge systems of science as a global network of interdependent information and knowledge clusters that are dynamically changing by their content and size. ISI-S assumes sets of information with shortor long-term impact and information integrated into the basic scientific knowledge or common knowledge. The type of the information sources (e.g., lecture, journal paper, review, monograph, book, textbook, lexicon) and the length of the impact are related to the grade of institutionalization. References are considered as proofs of manifested impact. The relative and absolute development of scientific knowledge seems to be slower than the increase of the number of publications. MODELSOF THE GROWTHOF SCIENCE According to the information model of science suggested by Nalimov & Mulchenko (1969)one can assume that scientific research is an organized information generating system and that science is a system of organized knowledge. Scientific research is fed with information as input for generating information as output that is new (original) or restructured knowledge compared to the input. The growth of science is preferably described in the literature by models based on the cumulative growth ofpublications. In each model the cumulaPeter Vinkler, Chemical Research Center, Hungarian Academy of Sciences, Pusztaszeri lit 5967,1025 Budapest, Hungary LIBRARYTRENDS, Vol. 50, No. 3, Winter 2002, pp. 553-569 02002 The Board of Trustees, University of Illinois 554 LIBRARY TRENDS/WINTER 2002 tive number of publications in a given year depends on the number of publications in the starting year, the rate of growth, and the length of the time period elapsed (Gilbert, 1978; Wolfram, Chu, & Lu, 1990). The linear model calculates with constant increases during equal time periods. Rescher (1978) suggested, for example, a linear growth function for the first-rate publications. The exponential model predicts an exponential increase of publications without limits to growth (e.g., Price, 1963; Egghe, 2000; Gupta & Karisiddappa, 2000).The logistic growth takes into account that scientific research is not a closed system and physical, economic, intellectual, etc. limitations occur that may bring about an upper limit to the growth (e.g., Price, 1963; Egghe & Rao, 1992; Gupta, Praveen, & Karisiddappa, 1997). The application of cumulative numbers of publications for describing the development of science is, however, inappropriate, since the method does not take into account the aging of information. The concept, “cumulative number of papers,” would indicate that all information previously p u b lished was relevant (regarding currency or recency) in the year of the study. This cannot be valid, considering, for example, the decreasing percentage shares of references with years referenced in Science Citation Index or Journal Citation Reports (SCI or JCR) for anyjournal. Several authors (e.g., Egghe & Rao, 1992; Egghe, 2000) try to describe the development of science with the assumption of exponential increase of publications and exponential decrease of the relevant information. Theoretically, the model may be correct but practically, the synchrony between the opposing trends cannot be justified for any period. Rescher (1978) tackled the “Rousseau law,” suggesting “that the historical situation has been one of a constant progress of science as a cognitive disciplinenotwithstanding its exponentialgrowth as a productive enterprise” (p. 111). The calculation of the annual increase and subsequent aging of publications may give only an approximation to the growth of scientific knowledge in different fields of the natural sciences. Science works with great redundancy; there are numerous parallel papers, and several results already published are republished as original works (Price, 1963; Merton, 1968). Menard (1971) investigated the publication development of chemistry, geology, and physics. The number of papers in physics increased linearly up to 1914 and then showed an exponential growth. The number of publications on chemistry was found to increase exponentially from the beginning of this century. Menard found very fast development in some hot fields, such as particle physics, where the annual rate was 15 percent in the 1950s and 60s. Menard distinguished three types of subfields: Stable fields, which increase linearly or exponentially at very slow rates; fast, exponentially growing fields; and cyclic fields, with stable and fast growth periods alternating. In support of Menard’s results, Vinkler (2000) found that the VINKLER/SCIENTIFIC INFORMATION 555 mean publication growth (i.e., mean annual number of publications) of different scientific fields strongly depends on the time period selected. For example, for Chemical Abstracts, a 6 percent mean annual increase was calculated between 1962-1979, and only one percent from 1980-1992, whereas 4 percent was observed between 1993-1999. Consequently, one may conclude that there is nogeneral law “governing” the publication growth of disciplines for longer periods. The (cumulative) increase (or decrease) in the annual number of publications depends on several factors within and without science. The time/number of publications functions may be valid only for the period studied and have no predictive power. Several attempts have been made to describe the development of science with nonscientometric models (Kuhn, 1962; Goffman &Warren, 1980; Crane, 1972; Mulkay, Gilbert, & Woolgar, 1975; Mullins, 1973). Gupta & Karisiddappa (2000) distinguished four developmental phases where cognitive content, methodology, type of publications, social structure, and institutionalization of the scientific research is characteristically different. According to this model, the information in the first phase is published primarily in “innovative” documents and reprints, in the second phase in papers, in the third phase in specific journals and textbooks, and in the fourth phase in journal bibliographies. The main institutional frameworks of emerging disciplines are as follows: Informal (nonorganized) stage, small symposia, congresses and formal meetings, university departments. GROWTHOF THE LITERATURE BY THE CHARACTERIZED RELATIVE PUBLICATION INDEX GROWTH For describing the publication growth of science, one may borrow an analogue from physics: The velocity of moving bodies is equal to the length of distance covered during a time unit. In scientometrics we may select one year as the time unit and the number ofjournal papers as the distance. Consequently, the annual number ofjournal papers published in a specific field of science may be accepted as Publication Velocity (PV) of the respective field (Vinkler, 2000). For characterizing the relative growth of the scientific literature during a time period, the mean Relative Publication Growth, RPG(t) index has been introduced (Vinkler, 2000). The RPG(t) index relates the number ofpublications issued in a gzven year to that published during a preceding time pm’od selected (t) . The length of the preceding period (termed as relevance period) may preferably refer to two, five, ten, or twenty years. The length of period t may be assumed as the maximum age of recent, relevant (RR) papers. RR papers are the publications that may contain all the information required for generating new information. It may be assumed that papers referenced in scientific papers at a given time may contain such information. The number of publications referenced during a period of seventeen to thirty years were followed in Chemical Abstracts (CA),Inspec Section A 556 LIBRARY TRENDS/WINTER 2002 (I),Psychological Abstracts (PA), Biological Abstracts (BA), Science Citation Index (SCI) ,and Mathematical Abstracts (MA).A relevance period of two years was applied. The RPG(2) indices were found as follows: CA (19621993),0.53;I (1980-1998), 0.52; BA (1964-1993), 0.53;SCI (1980-1998), 0.52; PA (1960-1979), 0.56; MA (1952-1990), 0.55 (the time periods studied are given in brackets). It may be easily concluded that the RPG(2) values refer to an average yearly percentage increase of about 4, 3, 4, 3, 8, and 7 percent, respectively (Vinkler, 2002). The Pearson's correlation coefficients characterizing the annual increase of papers in time were found significant, positive, and relatively high (> 0.92) for all cases. In contrast to this, the trends of the yearly RF'G(t) values gave controversial patterns. In some cases, they were significant but negative; in other cases, they were not significant. From the RPG(t) values calculated for the different disciplines the following conclusions may be drawn: The RPG(t) values depend on the length of the relevance period (t) selected; greater tvalues result in lower RPG(t) data; The greater the annual percentage increase of publications, the smaller the ratios between RPG(2) /RPG(5) /RPG( 10); RPG(t) values calculated with similar t-data are similar for the different disciplines; The mean RPG(2,5,10) values are higher than the theoretically calculated ones (0.50, 0.20, 0.10, respectively), meaning that there is an increase in the relevant information production within the time periods studied; The very low standard deviation values may indicate relatively constant RPG(t) values for the time periods studied. Latter findings indicate that the increase ofthe recent, releuant body of scientijic information is slower than that of the total information. For lower aggregation levels, the data referring to RPG(2) and (yearly percentage increase) between 1970-1998 were found as follows: Applied chemistry and technology, 0.533 (4.22 percent); biochemistry, 0.529 (4.05 percent); physical and analytical chemistry, 0.520 (2.94 percent); macromolecular chemistry, 0.525 (2.89 percent); organic chemistry, 0.505 (0.46 percent). For comparison, RPG(t) values were calculated for some fast developing topics, such as AIDS research, fullerenes, nanostructures, composites, antisense nucleotides, etc. The respective RPG(t) values were found to be szpniJicantlyhigherthan those for whole disciplines (Vinkler, 2002). The findings mentioned are in accordance with the concept recently suggested by van Raan (2000) :Science can be regarded as a dynamic integrative system where the development results from the growth of several subsystems with very different publication velocities. The models based on the concept of the cumulative or relative publiVINKLER/SCIENTIFIC INFORMATION 557 cation growth of science, which calculate with the number of papers published yearly, can give a simplified picture only. The aim of the present paper, however, is to describe the development of science by a scientometric model that integrates the production, evaluation, modification, and aging processes of scientific information. MAINCATEGORIES AND GENERALFEATURES OF THE INSTITUTIONALIZATION INFORMATION, OF SCIENTIFIC A SCIENTOMETRIC MODEL(ISI-S MODEL) According to the central concept of the 1%-S model the scientific information disclosed may develop with time through various evaluation and modification processes toward a cognitive consensus of distinguished authors of a scientific field or discipline. The ISI-S model assumes permanent production, evaluation, and modification of scientific information. It describes the information and knowledge systems of science as a global network of interdependent information and knowledge clusters that are dynamically changing by their content and size. The content and size of the individual clusters are regulated by different assessment processes. The definitions (below) and the categories (Table 1) of ISI-S suggested here should be regarded as approximations. The term “information” refers always to natural science information. Information in scient$cpublications (e.g., papers, book chapters, conference lectures) is: Addressed to the respective scientific community; Reviewed by peers before publishing and revised by the authors, if necessary; Disclosed by generally accepted norms of scientific publication of the respective discipline. Scientific publication is a means of announcingpriority (Price, 1963; Garvey, 1979) and contains (or at least should contain) all the information required for understanding and repeating the results published (Vinkler, 1998). The ISI-S model postulates five main information sets, which can partly overlap: Information in publications; information of short-term impact; information of long-term impact; basic scientific knowledge; and common scientific knowledge. The rank of the information clusters as mentioned represents the hierarchical grade of institutionalization (see below) of scientific information (Table 2 and Figure 1). 1%-S postulates three main and several additional evaluation processes. The first process refers to public access of the information to be published, the second to the relevancy and use of the information published, and the third to its general acceptance as part of the basic scientiJic knowledge of a discipline (Figure 1). Ta bl e 1. S ur ve y of t he M ai n C at eg or ie s of t h e IS IS M O D E L . Fo rm s of S ci en tif ic C la ss es o f Ty pe s o f Im pa ct In fo rm at io n by In fo rm at io n D is cl os ed of I nf or m at io n T er m s o f Im pa ct D is cl os in g by R el ev an cy Pu bl is he d of In fo rm at io n TV pe s o f th e Pr oc es se s C la ss es o f E vd lu dt or S M ai n E va lu at io n Fo rm s N on -p ub lis he d, R el ev an t, no nre le va nt M an if es te d, Sh or t, lo ng , Sc ie nt if ic r es ea rc h; Pe er s, A cc ep ta nc e or re fu se su bm it te d, (f au lty o r r ed un da nt ), la te nt , ve ry lo ng pu bl ic at io n; re le va nt a ut ho rs , of pu bl ic at io ns pu bl is he d ag ed la ck o f im pa ct ab so rp ti on , di st in gu is he d au th or s su bm it te d, ev al ua tio n, u se , re ce ns io n w ri tte n an d m od if ic at io n on p ub lic at io ns , of in fo rm at io n ci tin g or n eg le ct in g