Computerized Literature Searching: Research Strategies and Databases

Research Strategies And Databases

Charles L. Gilreath

• Publisher: Routledge
• Year: 2019
• Print ISBN: 9780367019686
• eBook ISBN: 9780429725289
• Edition: 1
• Topic: Social Sciences
• Subtopic: Sociology
• Index: Social Sciences

The computer terminal is well on its way to being as commonplace as the telephone, and its usefulness to the scholar and scientist is so great that already computer screens and disks are seen frequently in academic offices. The value of computers in research is well established, with vast amounts of data being processed daily by all sizes of computers. Computers also have had dramatic effects on the researcher's literature-searching options: Scientists and scholars can now query enormous databases containing tens of millions of citations to published literature and can extract bibliographies tailored to their specific questions. The power and flexibility afforded the user of information by these literature-searching systems ease the burden of library work, but in order to use the systems effectively, it is necessary to understand both their capabilities and their limitations. Mr. Gilreath describes the principles underlying online bibliographic systems, the databases available, and the factors a researcher must consider in using them. He explores in some depth the relationship of the structure and terminology of publications in various fields to the literature-searching process and provides detailed guidelines for research in the life, health, agricultural, and social sciences, chemistry, physics, mathematics, geology, meteorology, engineering, education, psychology, business, law, current affairs, and the humanities. A glossary of literature-searching terms is included.

1
The Basics of Computer Literature Search Systems

In 1946, when the 19,000 tubes of the ENIAC computer began to glow dimly at the University of Pennsylvania, a new era of information processing was born. Originally conceived as high-speed calculating machines to help scientists solve complex equations, the generations of computers descending from ENIAC have evolved into powerful general purpose logic machines capable of manipulating enormous banks of data. The pervasiveness of this technology in society is nearly impossible to fathom. Today computers touch us in almost every aspect of our lives, from calculating and printing electric utility bills to controlling the flow of traffic on city streets, from recording vital statistics of newborns to editing the text of newspaper obituary columns. Computer technology has forced us to develop new skills virtually undreamed of by our great-grandparents and to do traditional jobs in radically different ways. Like the rest of the world, libraries have been significantly affected by the computer revolution, some having joined eagerly and others finding themselves swept along by the succeeding waves of technological change.

Generally the evolution of automation in research libraries has been led by those activities most heavily utilizing clerical time, that is, the purchasing, cataloging, and circulation of books and other library materials. Automated circulation systems, usually run on in-house or at least locally owned computers, have relieved both library staff and patrons of much of the drudgery required for maintaining records of book circulations. Large national or regional computer systems, such as those developed and operated by OCLC, Inc., the Research Libraries Group, and the Washington Library Network, have been highly successful in speeding the cataloging process for their member libraries. Using one of these shared cataloging systems, a person sitting at a computer terminal in Texas can locate cataloging copy for a book, enter whatever changes are desired to make that record meet local library standards, order a set of cataloging cards, and share with member libraries in Kentucky or Rhode Island the fact that that item is now available for loan in Texas—all in a fraction of the time it used to take to type a set of cards to file in the local catalog.

Used in their traditional way, circulation and cataloging systems are essentially behind-the-scenes operations that affect only minimally the way people use libraries. New developments in the nature of these record-keeping systems, however, will soon have a dramatic impact on how information can be retrieved from libraries. In the next few years it will be possible for patrons of many libraries to go to a terminal—perhaps even in their homes or offices—and dial up the library's computer. Once connected, they can query the system for books on any subject. Having identified the ones they want, they can determine whether those books are available for checkout. In some instances, the library users may even be able to request that the books be held for them until they can get to the library to charge them out.

Perhaps more dramatic in its potential impact on library research methods has been the development of computer systems to search indexes, abstracting services, directories, and other traditional reference tools. Over the past fifteen years the publishers of indexing and abstracting services have invested heavily in automation. Entries for such services are today typically prepared and edited in machine-readable form. A computer tape is prepared, which is subsequently used to drive typesetting equipment or to photocompose directly camera-ready pages for printing. These edited computer tapes have become an increasingly valuable resource as the basis of large machine-readable databases. The mere existence of a machine-readable file does not, of course, have much impact on the retrieval of information by individuals conducting research. Slow and inefficient computer programs to search machine-readable files are no better—and sometimes worse—than manual search techniques.

The mid-1960s, however, saw the development of computer systems that made possible the fast retrieval and manipulation of bibliographic information. These systems allowed users to query bibliographic databases for words and phrases that might appear in the citations and, in a conversational mode, to modify the results of the search. Thus it became possible for researchers to go to one of these systems, pose questions and, based on the results of the queries, adjust the answers so that they got out of the system a list of references tailored to their information needs. Because online systems such as these require such enormous storage capacity, they are too expensive for most libraries to consider operating locally, but through subscription to any of a number of computer services that provide dial-up access to timeshared computing centers, many libraries can gain access to literally millions of machine-readable records in order to provide interactive searching of vast stores of information.

Two of the earliest and most successful computer retrieval systems are System Development Corporation's (SDC) ORBIT and Lockheed (now DIALOG) Information Service's DIALOG. Both of these retrieval packages were developed in the late 1960s under government contract. SDC developed its ORBIT system to make the National Library of Medicine's MEDLARS (Medical Literature Automated Retrieval System) searchable in a conversational mode, while Lockheed developed the prototype of its DIALOG system as the RECOΝ (Remote Console) system for NASA. The years since these early systems began operation have seen the development of many other retrieval systems throughout the world. By 1982 it was estimated that there were actually dozens of retrieval systems in operation providing online access to tens of millions of bibliographic and factual records.

In operation each retrieval system may look quite different to the uninitiated observer, but whatever system is used— DIALOG'S, SDC's or another system from any of the burgeoning number of information services now in the field— the basic organization and operation is the same. Figure 1-1 (a hypothetical search sequence) illustrates many of these features.

In this example the computer system prompts the user each time it is ready to accept additional commands through the cues for each statement and the word "USER." Its own responses

Figure 1-1. Hypothetical search sequence

are labelled "PROG." After each system cue, the user enters a part of the search strategy, in this instance using the commands "FIND," "MATCH," "LIMIT," and "PRINT." After each instruction from the user, the system responds with a number, indicating how many references in the database meet the requirements of that statement. Thus, the first response from the system shows that there are 1,500 references in which either "automobiles" or "cars" appears in the parts of the reference searched by the computer. The dollar signs in statement 3 are truncation symbols, signaling the system that all words beginning with the preceding characters are acceptable; thus words with many possible forms can be searched with minimal keying. The vast numbers of citations retrieved can be matched against each other in order to narrow search results to that much smaller number meeting the various aspects of the search logic. Usually in such a match the numbers drop dramatically, as in the results of statement 4, which represents those items in the hypothetical database which should be on automobile sales forecasts.

Retrieval systems, by one means or another, usually allow users to manipulate search results in other ways as well, as illustrated by statement 5, in which the eighteen citations in the previous statement are limited to those items published in 1981. Once narrowed sufficiently, the results are sampled by requesting a printout of selected titles. If the sample is acceptable, the user can choose to have a full bibliography of the results printed out either immediately at the terminal while it is still connected to the system or later at the computer facility and sent through the mail.

The flexibility that these systems provide represents a quantum leap for the researcher who must use bibliographic reference tools. Some of the differences between manual and online searches of literature are obvious; after all, with one, printed volumes are consulted and with the other, a computer terminal is employed. Other aspects of this new technology for literature searching are more subtle, but their impact on the way one carries out a library research project is significant. The chapters that follow will provide more detailed discussions of online research methods in specific disciplines, but it will be useful here to look briefly at the major aspects of both manual and computerized versions of indexing and abstracting services.

Organizing Information

Before turning to the more specific problem of how information retrieval systems are put together, one needs to consider for a moment the more general issue of information organization. Figure 1-2 represents one conception of how information is produced and subsequently organized for retrieval. At one extreme is the author who has discovered something about reality and, wishing to communicate with others, writes up the discovery as an article and submits it for publication. If the article is accepted, it will be edited and eventually published and distributed to the subscribers of the journal. A recipient of that journal may then read the article and make use of its contents right away or, if the information is not needed immediately, may file the journal away for future reference. A researcher with an unusually keen memory may be able to come back directly to that article in the future but will more likely have to rely on some secondary tool as a memory aid. These secondary tools can take many forms, ranging from a simple card file on the researcher's desk to the elaborate multivolume abstracting and indexing services such as Chemical Abstracts or Zoological Record.

The primary point to be remembered from this illustration is that communication between authors and readers is highly complex. At each stage in the flow of information from the author or producer of the information to the library researcher or user of information there are several opportunities for the original message to become garbled. As one progresses along the continuum from author to reader, the further away from the author's manuscript one proceeds the fewer become the methods of retrieving information. For example, a journal editor may have to edit a manuscript severely in order to make it fit space requirements, and most certainly an indexing or abstracting service is going to use only a skeletal version of the original article to point the reader back to the source. The fewer access points a reader has for retrieving a reference, of course, the greater are the chances of missing relevant citations in a search

Figure 1-2. Production and transmission of information.

of the literature. The bottom portion of Figure 1-2 shows the types of access commonly provided at each stage of the information flow.

The author's own title for an article can provide many access points, provided that care is given in selecting terminology that accurately reflects the content of the paper. Depending on the journal for which the article is intended, the author may also provide a brief abstract and/or a list of keywords that may affect subsequent retrieval of articles via indexes. Finally, indexers and abstracters create document surrogates, that is, bibliographic citations that try to represent in a very brief form the most significant aspects of the original publication. Indexing citations may include the addition of keywords from a controlled list of such terms, specially prepared abstracts, and/or various other codes identifying aspects of the article such as, subject, publication type (review, theoretical paper, etc.), language, and so forth.

Printed Indexes

Hierarchical Indexes

Manual literature-searching techniques rely on the structure of printed retrieval tools. Printed indexes take many forms, but all are linear devices, that is, all references are arranged in a locked sequence on the printed page. Most printed indexes are hierarchical, presenting citations in a logically arranged series of subject headings and related topical subdivisions. Users of such a work look under those headings that appear most aptly to describe the area of investigation. Having located the appropriate heading, they then read through the titles appearing under that heading to identify the ones they are interested in using for their research. Hierarchical indexes are particularly good in providing access to large volumes of literature under a carefully controlled set of categories, thereby freeing the user from having to search large numbers of individual, and sometimes idiosyncratic, terms to locate desirable references. In most instances, however, such indexes are slow to reflect changes in terminology. When new research areas emerge, some substantial period of time is usually required before new entries are introduced into the list of subject headings of printed hierarchical indexes; so the user working on a new area of specialization has to seek information under headings not quite on target for the research topic.

KWIC and KWOC Indexes

In order to overcome the shortcomings of hierarchical indexes, other approaches to indexing have been developed. KWIC (keyword-in-context) and KWOC (keyword-out-of-context) indexes provide subject access via computerized rotations of keywords and/or title words. This approach, as shown in Figure 1-3, enables the reader to search for articles that contain specific words or phrases in the title or in strings of index terms. The KWIC index also displays terms in such a way that the user can easily check the context in which the desired term appears.

Coordinate Indexes

A third type of indexing scheme, adaptable for both card files and printed retrieval aids, is the coordinate index. It is the indexing principle, as well, upon which computer literature-searching systems are based. An index based on coordinate principles first identifies each bibliographic entry by means of some unique number or letter-number tag, as shown in Figure 1-4. Indexes are built for each significant word by listing together document numbers for every item containing that word. These document number lists that constitute the index's access points are then filed or printed in alphabetic or numeric sequence. When users d...

Table of contents

• Cover
• Half Title
• Series Page
• Title
• Copyright
• Contents
• List of Figures
• Acknowledgments
• Introduction
• 1 The Basics of Computer Literature Search Systems
• 2 The Researcher-Search Analyst Partnership
• 3 Agricultural and Life Sciences
• 4 Social Sciences and Education
• 5 Physical Sciences and Engineering
• 6 Business Literature
• 7 Humanities, the Arts, and Architecture
• 8 Legal Research
• Glossary
• Index