Braun Tibor, Glänzel Wolfgang, Schubert András: Országok, szakterületek, folyóiratok tudománymetriai mutatószámai 1981-1985 (A MTAK Informatikai És Tudományelemzési Sorozata 6., 1992)

Methodology

A. SCHUBERT, W. GLÄNZEL, T. BRAUN : SCIENTOMETRIC DATAFILES Methodology There are some fundamental methodological questions to be answered before any actual work in building scientometric indicators could be done. 1. What kind of publication channels are to be taken into account in an evaluative study? (All titles included in the database, journals only, a selected set of journals etc.) 2. What publication types are to be counted? (Research articles only, or reviews, letters, notes etc. are also to be included) 3. How to assign publications to countries? (On all authors basis, on first author basis, fractionating between coauthors etc.) 4. How to classify publications into science fields or subfields? (Based on journal or title of publications or author's affiliation etc.) 5. What to choose as the filing year of a publication? (The year indicated on the cover of the journal or the year it was included into the database) 6. How to choose the citation period? (One, two or three years after publication or two source years and citations in the subsequent year etc.) 7. How to match citations to cited papers? (Requiring exact match of all bibliographic data, using some search key etc.) It would be, of course, most instructive to present here the results of all possible alternatives, and to let the reader select among them. It is, however, clear that, since the number of possible combinations of the above loosely sketched alternatives is about 1000, it is practically impossible to carry out such a parallel experiment. What we can offer here is just one set of choices, not lacking a certain amount of arbitrariness, but incorporating all the experience we happened to have accumulated. Ad 1. The SCI database covers annually cca. 3500 journals and a few hundred non-journal titles (multiauthored books, monographs, etc.). Since an average journal publishes about 1000 papers per year, whereas a non-journal title contains only a few dozens, no significant loss is caused by completely omitting non-journal items. (Whereby the potential scientific merits of such publications are not in the slightest way doubted.) A further motivation of this omission lies in the difficulty of finding citations to such publications. As source items, nonjournal titles are recorded in the SCI database by a special code beginning with the characters BK#. When recording cited items, the SCI follows the same form the citing author used in his/her paper. Therefore, it is hardly possible to find citations to non-journal source items by a computer search of the SCI database. Sricnlontrtrics M (11.10) 5