College and Research Libraries HANSJOERG KOLDER and IRWIN F. SIMPKINS Selective Dissemination of Information and the Academic Science Library The meaningful discharge of academic duties requires the effective use of up-to-date factual and conceptual knowledge. The wealth of the material available and its perishable character necessitate an expedi- tious procedure for information sampling, storing, and retrieving at the individual level. The feasibility of such a procedure, administered by the science library, is explored. The discussion consists of three parts: The inherent inadequacies of biomedical bibliographic services to fill the need of an individual for current awareness; the magnitude and cost of a modest personal information system; and the potential supportive participation of the science library in the teaching and re- search activities of members of the university community. REGULAR AND SIGNIFICANT CONTRIBU- TIONS to the pool of universal knowledge are expected from the members of the academic community, but this privilege seems something of a liability because the accelerated pace of scientific re- search has generated such a wealth of publications that it is more and more difficulty to keep abreast of develop- ments even in a narrow field. It can be · difficult sometimes to judge whether a scientific effort represents original work, and reliance is often placed upon the mere probability that research has been anticipated. In 1963 the number of jour- nals with scientific value was estimated to be approximately thirty-five thou- sand.1 A substantial portion of the in- 1 Charles M. Gottschalk and Winifred F. Desmond, "World-Wide Census of Scientific and Technical Serials," American Documentation, XIV (July 1963), 188-94. Dr. Kolder is Associate Professor of Phys- iology and Mr. Simpkins is Science Refer- ence Librarian in Emory University, Atlan- ta. This work was supported in part by the U. S. Public Health Service, National In- s_titute for Neurological Diseases and Blind- ness Grant NB 21,768. formation published in these journals has a limited useful lifetime. 2 Therefore, if a publication is to serve its purpose, it must find its audience quickly. The mul- titude of journals and the perishability of their contents make the need for an effective current awareness service most urgent. The purpose of this paper is to indicate mechanisms by which the sci- ence library could render a valuable service to teaching and research by establishing a current awareness service geared to the interests of individual members of the academic community. A brief discussion of a personal informa- tion system illustrates the direction of further developments. No bibliographical service currently functions satisfactorily as a means of keeping individuals up to date, although several comprehensive services are avail- able as retrospective searching devices. In biomedicine, for example, three ma- jor indexing and abstracting journals are available in English alone. Each of these journals, Biological Abstracts, Excerpta 2 R. E. Burton and R. W. Kehler, "The 'Half-Life' of Some Scientific and Technical Literatures," American Documentation, XI (January 1960), 18-22. I 53 54 I College & Research Libraries • January, 1967 Medica, and Index Medicus, now lists more than one hundred thousand pub- lications annually.3 None of these jour- nals, however, is prompt enough in re- porting the literature to constitute an ef- fective current awareness service. The listing in Index Medicus runs at least three months behind publication of the original report, the citations in Biological Abstracts are six months to one year late, and the abstracts in Excerpta Medica sometimes trail the original publication by two years. Current Contents might be considered an acceptable current aware- ness service but it lacks a subject ap- proach to the content pages. Another disquieting factor in the evaluation of bibliographic services is the unpredict- ability of editorial acceptance of · pub- lications for inclusion. In 1965, Biolog- ical Abstracts listed 6,735 serial titles,and contained 110,119 abstracts or titles. 41 As- suming an even distribution, this would mean that fewer than seventeen titles per journal were editorially selected for inclusion. Excerpta Medica listed about forty-four publications per year for each journal covered in 1964, while the com- parable figure for Index Medicus was sixty-three publications per journal. The inclusion ratios quoted constitute only a fraction of all papers published in the source journals used. This selectivity may be a desirable device for protection of the user from an inordinate · amount of noise; but Oehlerts5 reported that 1,635 of the nearly five thousand jour- nals on the "List of Serials Abstracted" in Biological Abstracts in 1960 were not represented by a single citation. Among the omitted titles were the Comptes 8 "List of Serials Abstracted," Biological Abstracts, XLVI (November 1965), 7373, and XLVI (Decem- ber 15, 1965), 8927; "A Review of Current Activities and Future Plans," Excerpta Medica, Sec. I, XIX (Jan- uary 1965), iii-xii; U.S. National Library of Medicine, Annual Report Fiscal Year 1964 (Washington: The Li- brary, 1965), p. 23. See also, Index Medicus, VI (December 1965), i. 4 Biological Abstracts, loc. cit. 5 Donald E. Oehlerts, The Most-Cited Serials in Bi- ological Abstracts in 1960 ( "Occasional Paper no. 65," [Urbana: University of Illinois Library School, 1962J). Rendus of both the Academie des Sci- . ~ences and the Societe de Biologie. Nor did Oehlerts find any citations to Gen- etica, Bibliographia Genetica, Archiv fiir Protistenkunde, and Planta. Other re- search journals of like stature were rep- resented by fewer than five citations. The difficulty of retrieval from biblio- graphic journals further diminishes the comfort derived from their services. Neither the permuted title index nor the conventional analytical index permit in- dexing in enough depth or spedficity to develop confidence in finding the mate- rial that the editors have seen fit for inclusion. Martyn and Slater6 found in tests on retrievability that, at best, 85 per cent of publications found through the author index of Biological Abstracts were retrievable through the permuted title index. The Index M edicus analytical subject approach yielded, in a less com- prehensive series of tests, a low of 5 per cent and a high of 70 per cent of the references known to have been in- cluded in that bibliographic journal.7 All this is not to say that bibliographic jour- nals are without merit, but one should be reminded that they have grave limita- tions. In order to compensate for the de- ficiencies of bibliographic journals, most academic teachers and scientists have to maintain a more or less elaborate per- sonal information storage and retrieval system. The input may consist of books, reprints, personal communications, pro- ceedings, minutes, graphs, etc. The in- formation is retained in subsystems of varying form, like shoeboxes, folders, notched cards, or 3 x 5 in. cards and may be found in alphabetic order or under broad subject headings without sufficient subdivision. The multiplicity of subsystems reflects the variety and mag- nitude of the input. One of the authors has maintained a personal information 6 John Martyn and Margaret Slater, "Tests of Ab- stract Joumals," Journal of Documentation, XX (De- cember 1964), 212-35. 7 Ibid. Selective Dissemination of Information I 55 system for the past fourteen years. Since tern is expected to reduce the retrieval 1959 this has been done systematically .. · time considerably, especially when the The main input to the system originates coordination of several key~ords is te- from the regular examination of some quired. Five simultaneous searches for eighty biomedical journals related to hu- coordination of up to seven keywords man physiology. The total file presently each through the whole tape memory consists of 7,400 precoded 5 x 7 in. cards require about fifteen minutes. Updating arranged in serial order. Each card rep- with new information and elimination of resents a document, which may be a obsolete information are standard pro- book, reprint, bulletin, specification cedures, as is the preparation of an al- sheet, manual, or a citation. Administra- phabetic list of authors and keywords. tive materia], corresp<;mdence, and au- On the other hand, the computer in- diovisual aids are not included in the volvement diminishes the accessibility system. The documents are kept in serial of the file since the computer is not al- order, but separate from the filing cards. ways instantly available. A published Retrieval of information is accomplished figure for a general, non-individualized, by an author index consisting of about computer-op~rated bibliographic. infor- seven thousand entries and a coordinate mation storage and retrievar··system is index with about four thousand key- $5 for each entry into the system, subject words. The system now demands ap- to variation with the number of entries proximately three hours weekly of the and the number of searches. 8 The infor- author's time for scanning and indexing mation system outlined above has been a of the source journals. The posting of valuable aid to both teaching and re- new serial numbers on author or key- search. The time and money invested are word cards and similar tasks ar&- per- a small price for the current awareness formed by a secretary. Out of approx- aspect alone. In addition, there is no imately ten thousand publications exam- problem in compiling references for re- ined in 1965, eleven hundred were add- search projects or reading lists for single ed to the file. On the average, 124 ar- students or whole classes. tides were found to be published per Certain objections can be raised against source journal scanned. This represents maintaining an extensive personal file. It a signal-to-noise ratio of 1:9; but it musl may appear redundant to index material be noted that those items retained are that will be included in one or more of highly relevant. The average estimated the major bibliographic services, and it cost per document retained in the whole seems doubtful whether a single indexer manual system is $1.25, based upon a can hope to cover more than a fragment two-year sample of time and expense ac- of the great mass of relevant literature. counting. This figure includes the ac- The charge of redundancy has already quisition of some 350 books, about four been disposed of in the discussion of the thousand reprints, filing facilities, and deficiencies of three major indexing and prorated secretarial help. The figure abstracting journals. These services are does not include compensation for time too slow, too general in their indexing, spent by the author. and too unpredictable in their selection Presently a computer program is be- to render superfluous some supplemen- ing written by J. C. Ziegler (Emory bio- tary personalized device. The justifica- medical data processing and analysis tion for indexing a small fraction of the center) for transfer of bibliographic in- biomedical journals derives from ev- formation from this personalized system to tape for machine retrieval on an IBM 1410 computer. The computer-based sys- 8 Charles P. Bourne and Donald F . Ford, "Cost Analysis and Simulation Procedures for the E valuation of Large Information System s," Ame rican D ocume nta- tion, XV (Aprill964), 142-49. 56 I College & Research Libraries • January, 1967 idence suggesting that the great bulk of ticipating member of the academic com- significant publications is concentrated munity. It might provide; as a second in comparatively few primary journals. step, copies of the content pages of jour- Brown9 counted citations in seven lead- nals requested. If electronic data process- ing journals of physiology and found ·:>.ing equipment is available, the library that fifty-eight journals yielded 90 per 1 · as a third step, might distribute weekly cent of the citations; publications in a permuted title index to the content of chemistry clustereft around even fewer scientific journals received. A logical fi- journals. The citation count in six rep- nal step might be the establishment of a resentative journals in chemistry reveal- system known as Selective Dissemination ed that thirty-eight journals produced of Information. This system requires that 90 per cent of the citations. Circulation participants profile their research and statistics reinforce the implications of teaching interests in terms of weighted the citation count. Fleming and Kil- keywords. The keywords can be authors' gour10 found that sixty-seven journals in names, subject headings, names .9f or- the Yale medical library represented 50 ganisms, reactions, or even citations. per cent of serials borrowed and 262 titles Each participant may submit as many accounted for more than 80 per cent of profiles as are necessary to describe all loans. An earlier study by Urquhart11 at aspects of his interests and needs. The the Science Library in London estab- publica~ons in selected journals are in- lished that forty journals provided half dexed, uS>ing terms in the participants' the circulation. It is conceivable that profiles as \ a subject authority. The pro- the spread of primary information is not files of the participants are matched, by so narrow as the citation count and the computer, against the profiles of the in- circulation statistics would indicate. It dexed publications. If a preselected level does seem likely, however, that inspec- of correlation is obtained between the tion of a relatively small number of well- profile of the participant and the pro- chosen journals will retrieve a large frac- file of a publication, an abstract is sent tion of the most important scientific pub- to the participant, who, in turn, is ex- lications pertinent to personal and spe- pected to furnish feedback on the ac- cific interests. tual relevance of the publication. The Since current awareness is vital to the feedback results in a continuous re- academic teacher and scientist as an in- vision of the user's profile and increases dividual and since both theoretical and the ratio of relevant information to noise empirical reasons indicate that a small delivered by the system. This informa- scale, i.e., university-wide, service can be tion system was devised by the · late effective, the science library could en- Hans Peter Luhn of IBM, who also hance its value by providing a current wrote a program for it, which is avail- awareness service. This service might able from IBM upon application. The take, as a first step, the form of a daily system has proved itself in many non- list of accessions, mailed to each par- academic environments, among them IBM itself, NASA, and the Ames Lab- oratories. One major difficulty arises with the maintenance of a Selective Dis- semination of Information system, viz., finding and keeping capable indexers. Graduate students can be used effective- ly, but student help usually implies a fast turnover. The most practical way, 8 Charles Harvey Brown, Scientific Serials ( "ACRL Monograph No. 16," rChicago: Association of College and Reference Libraries, 1956J), p. 116-19; 98-99. 10 Thomas P. Fleming and Frederick G. Kilgour, "Moderately and Heavily Used Biomedical Journals," Medical Library Association Bulletin, LII (January 1964), 234-41. 11 D. J. Urquhart, "Use of Scientific Periodicals," in International Conference on Scientific Information, 1958. Preprints of Papers (Washington: National Academy of Science, 1958), p. 277-90. Selective Dissemination of Information I 57 and possibly the final step in improve- ment toward an individualized system at the university level, seems to be the di- rect participation of potential users,_ Each participant could make · himself available for the indexing of five or six journals in his field of specialization. Judging from the Cranfield reports, 1~ this effort should require not more than five to six hours per month, and would seem a small investment for a hand-tai- lored service prepared competently and covering an area limited only by the num her of participants and their diverse interests. With little more trouble and expense the library could maintain this u Cyril W. Cleverdon, Report on the First Stage of an Investigation into the Comparative Efjiciency of Indexing Systems (Cranfield, England: College of Aeronautics, 1960 ) . system as a retrospective device for lit- erature searches. The progressive build-up through the steps discussed could be established and maintained by the academic science li- brary. The service should improve mark- edly the signal-to-noise ratio at the input to a personalized information storage and retrieval system. Admittedly, this would be a departure from the tradi- tional role of an academic science li- brary, but special libraries have been at- tempting services like this for several years. The availability of computers, the growth of knowledge, and the accumu- lating emphasis upon research have com- bined to make it all but mandatory that the science library assume a more active role in the research process. • •