Charles R. Hildreth, Ph.D.
2. OPAC ACCESS MODELS
2.1. Stages in the Development of Online Catalogs
Figure 2 presents a framework for understanding 15 years of OPAC history and the significance of recent developments. Two access models informed and motivated the design of the first OPACs. One track of development reflected attempts to emulate the familiar card catalog in its new online form. Another track of development adopted the model familiar to online database searchers of commercial search services like DIALOG and BRS. Second-generation OPACs, with their multiple access points, search approaches, and user-friendly display formats, represent the marriage of these two access models.
Second-generation online catalogs, which began appearing in the mid-1980s, represent a qualitative leap of progress over first-generation online catalogs. In first-generation online catalogs, access to catalog records was restricted to entry via author or title data. Searching was initiated by derived-key input or by exact word or phrase-matching on at least the first part (left-most) of the word or phrase (as with heading searches in the card catalog). Typically, only short bibliographic records were displayed in a format that resembled the catalog card. No subject access via either keywords or controlled vocabulary was provided, and searching was constrained by a two to three-step linear process. Refining and improving a search in progress, based on an evaluation of intermediate results, was not possible. Without full records, subject access, authority-based searching with cross references, and meaningful browsing facilities, first-generation online catalogs were understandably criticized as inferior to traditional, non-automated library catalogs.
From the earliest days of second-generation OPACs, the challenge that confronted system developers was to combine the ease-of-use (or at least, familiarity) of the card catalog with the powerful search capabilities available to trained online database searchers. This task has not been easy, but significant progress has been made.
Present-day OPACs are powerful, end-user, computer-based interactive retrieval systems that provide greatly enhanced access to the bibliographic citations which represent the materials in one or more library collections. The early model of the card catalog online no longer represents reality.
Today's state-of-the-art, second-generation OPAC is a far more effective form of the library catalog than earlier forms because it provides:
MORE ACCESS POINTS
Any field or word or symbol in the bibliographic record can become an entry point
MORE BROWSING OPTIONS
For those who cannot describe what it is they want, or for those who wish to explore the unknown, perhaps via links and trails which lead to books of interest and related materials
MULTIPLE CATALOG ARRANGEMENTS IN A SINGLE SOURCE
Dictionary, divided, classified, etc.
BETTER SEARCH RESULTS
Broad search results for browsing "mini-catalogs", or very precise search results through the careful use of Boolean search capabilities
REMOTE ACCESS
Consultation from locations near or far from the library
MATERIALS STATUS AND AVAILABILITY INFORMATION
Interfaced to a circulation system, it provides specific-copy loan status; with an acquisitions system, "on order" and "in process" information may be displayed
ACCESS TO THE RESOURCES OF OTHER LIBRARIES
Either through a shared database or system-to-system linkages
INCREASED UTILIZATION OF LIBRARY RESOURCES
Both human and material resources
GREATER USER ENJOYMENT AND SATISFACTION
With the search process and search results
Third-generation online catalogs, not yet widely available, will be described later in this report. The body of this report provides the rationale for the development of this new generation of online catalogs.
2.2. Defining Features of Present-Day Second-Generation OPACs
Although some first-generation OPACs can still be found in libraries, most have been replaced by today's state-of-the-art second-generation systems. The features which define second-generation OPACs are listed below. The majority of installed OPACs satisfy these criteria.
Second-Generation Online Public Access Catalog Features
Improved card catalog-like "main entry" searching and browsing-by-heading capabilities have been joined with the conventional IR keyword and Boolean searching approaches. Much of the power and flexibility familiar to online database search specialists which enables the "post-coordinating" of search concepts and terminology has been brought to the online catalog searcher. Many online catalogs support the ability to restrict searches to specified record fields, to perform character-masking and/or right-hand truncation, and to limit the results by date, language, place of publication, etc. These developments have, we suspect, exacted a price in the usability of these catalogs by untrained users. Acquiring an adequate conceptual understanding of present-day OPACs is further complicated, no doubt, by the fact that they are hybrid card catalog/Boolean search systems. For an excellent critical analysis of the card catalog access paradigm and its shortcomings as a model for online catalog design, see Borgman's forthcoming article, "Why are Online Catalogs Still Hard to Use? Revisited: Some Issues for Training and Design," in a forthcoming issue of the Journal of the American Society for Information Science (1995).
Recent developments with OPACs, including the advent of the networked OPAC, encouraging as they are, must be kept in proper perspective. Progress has been made along some, but not all, dimensions of potentiality. Improvements to the user interface have made second-generation OPACs more usable for the untrained user, and new graphical user interface (GUI) techniques like windows and point-and-click buttons hold the promise of rendering the search process both more intuitive and more direct. Second-generation OPACs are being expanded in information content and coverage almost daily, and access to and from these OPACs has been extended to vast numbers of remote users through network technologies. Why, you may ask, is not this expanded, extended OPAC advanced enough to qualify for the distinction of a new, third generation OPAC? This is the central question addressed in this report.
2.3. The E3OPAC
There has been some confusion in the literature of late as to whether the emerging online catalog is the "expanded" or the "extended" online catalog. Some writers view these terms as synonymous and use them interchangeably; but often there is a lack of clarity or consistency in their use to explain related but different developments. For example, Potter identifies three complementary "expansion" paths along which online library systems will proceed: 1) more indexes to more sets of collections and more online reference databases, 2) the gradual inclusion of more full text of journal articles, and possibly, books, and 3) "greater connectivity from online library systems to other systems, including other library systems, commercial services, bibliographic utilities, local networks, CD-ROM servers, and other information providers in the community" (1989). Mischo puts it this way: "Recently the idea of the 'extended' online catalog has been introduced to describe online catalogs containing specific functional or data extensions. Extended third generation catalogs typically provide value-added access beyond the conventional online catalog by providing expanded entry points, augmented information resources, access to locally mounted and/or remote periodical index databases, and gateway functions to local, regional and national telecommunication networks" (1991).
We introduce the notion of the E3OPAC as a vehicle for elucidating these three concepts:
Enhanced: Functionality and Usability
Expanded: Indexing, data records, collection coverage; i.e., a "full-collection" access tool
Extended: Through linkages, networks, and gateways to additional library collections, information systems and resources
The E3OPAC would have enhanced functionality and usability; its indexing, record data content, and collection coverage would be expanded to make it a "full-collection" access tool; and its access would be extended (through linkages, networks, and gateways) to include the collections and resources of other libraries and information centers.
2.3.1. Expanding the OPAC
The enabling technology of the online catalog makes it possible to expand the scope of the library catalog, and, joined with telecommunications technologies, to extend its range. With the online catalog, it is again feasible to expand the catalog's coverage over, and to deepen access to, all the information materials in the library's collection. As Potter (1989) reminds us, this is not a new ideal:
At one time, a library's catalog was designed to index every intellectual work in the collection - not just books, but also articles in periodicals held by the library, pamphlets, maps, government publications, the whole range of materials acquired by a library. By 1900, all but a few highly specialized libraries found this was an impossible task, and today most catalogs provide access only to the books and the set titles of serials. Readers interested in the other types of materials are compelled to consult indexes that are separate from the catalog such as printed indexes, CD-ROM databases, or commercial online services. While these tools are usually of high quality, they are scattered throughout a library. The reader has lost the unifying function of the catalog, the ability to locate any item from a single source.
It is possible to exceed the ideal of yesterday by providing comprehensive access to the local collection, as well as access to resources beyond the local library's collection. Potter reviews library cases where expanded access to materials and information both within and beyond the library has been "performed in conjunction with an online catalog so that the reader is provided with a single source, a common interface, a unified environment in which to retrieve information." He identifies three trends in current online catalog development and practice: 1) the unification of local collections, 2) providing access to outside resources, and 3) the inclusion of machine-readable reference works and full text (1989).
Access to and knowledge of a work's location and availability status is deepened when the holding library has automated its bibliographic management, circulation, and public catalog functions in an integrated manner of one sort or another. If one views the online catalog as no more than the card catalog replicated online, then it is easy to see how integrating or linking the data records of the circulation and acquisitions systems with the corresponding bibliographic records represents the first stage in the development of the "extended online catalog." One might say, "the extended online catalog begins at home." When the online catalog is interfaced with the local circulation and acquisitions systems, more precise answers can be given online to the user's primary questions: "Does this library have what I want?" and "Can I have it now?" or "When will it be available?" In the locally-extended online catalog on order, in process, or availability information is immediately at hand, along with the bibliographic record.
Two additional developments are leading to an expansion of the local online catalog's full-collection access potential: 1) inclusion in the bibliographic record and the indexing of contents notes opens access doorways to specific authors and titles of essays or short stories published in collections or anthologies, and 2) the mounting (loading, indexing and storing) of commercial indexes and abstracts for periodical journal articles and government publications (Potter, 1989). Some libraries are also loading locally-created indexes to specialized collections and reference materials. A few are adding headings from the tables of contents of newly acquired books and journal issues to the corresponding bibliographic records. An available 'Notes' field in the standard bibliographic record format can easily be used to accommodate this additional book or serials content-rich data. As a result, additional access points to the material are provided to the online catalog user, and when displayed, the table of contents can show the user more accurately than one or two subject headings what is inside a book or periodical issue.
2.3.2. The Extended, Networked OPAC
What is perhaps more revolutionary in Potter's vision is the notion of extending the local online catalog to become a gateway to other library collections, and/or to become a full-text delivery system. An automated library system offers a library the opportunity to access the systems and databases of other libraries through the terminals on its own system. Using a vendor's proprietary network linkages to other libraries using the same vendor's system, or customized software gateways, or dial access public networks, a user at the terminal of one system can access the database of another system. In doing so, the user can search that database to discover if that library owns a desired item, check on the availability of the item, and submit an electronic request for the delivery of the item via interlibrary loan procedures. The advantage of this situation over the "traditional" electronic union catalogs joined with ILL messaging capabilities (e.g., OCLC) should be obvious. The local status, precise location, and availability of specific copies of the item can be determined quickly, and the holding library can be notified instantly that a loan request has been placed on an item that is available. This can lead to major reductions in staff time (and reductions in the associated costs), and dramatically reduces the user's "wait time" to have the requested item in hand.
The value of extending the online catalog to include the collections of several libraries in a district or region seems to have been proven. Experience to date shows that resource sharing increases significantly when library online catalog users can view the holdings of other libraries in their metropolitan area or nearby region. Interlibrary loan and reciprocal borrowing (one library's patrons are permitted to borrow directly from another library by special administrative arrangement) have doubled and tripled in some cases. There are, of course, different methods of extending the local online catalog to include the holdings of other libraries: for example, by joining an online union catalog, by loading the catalogs of other libraries into the local database, and through gateway linkages or interfaces to the other libraries' local systems.
By way of summary, the following list identifies eight ways the conventional library catalog is being extended in a variety of online manifestations in our libraries. Most of these extensions involve adding data to the MARC catalog records, integrating related data files such as customized periodical indexes into the monograph catalog, or adding reference information files to the overall online catalog database or aggregate of databases searchable through the online catalog. However, functional and transactional performance extensions are also being made to today's second-generation online catalogs. This is all to the good, because research and experience have provided us sufficient reason not to be satisfied with the performance of today's online catalogs. Reflecting on all this creative, expansive activity by online catalog designers and librarians, it is clear that in practice no pre-defined "theoretical" boundaries for the proper library catalog (regarding its form, function, or content) are being respected or observed. We are witnessing a shift in emphasis from our usual concern for bibliographic control to expanding access to all the materials and information in our collections. The promise is that the library's primary access instrument, the"catalog", will become its most used and most effective access and discovery tool.
EIGHT WAYS OF ENHANCING AND EXPANDING THE OPAC
Progress along the road to the E3OPAC may be tracked on a three-dimensional scale, as illustrated in Figure 3. One can redefine or reprioritize the axis values, but such a scale is useful for identifying where progress may be lagging and also for comparisons of different systems. Third generation systems are defined primarily by their advanced interface search functionality. It is along this dimension that OPAC progress seems to be stalled. It is important to understand why this is the case.
2.4. The GUI OPAC: Approach With Caution
"GUI" stands for graphical user interface. Williams, et al, (1995) describe three types of user interfaces that have been developed to facilitate interaction between a user and an information system: command-driven, menu-driven, and graphical user interface (GUI). They define a GUI as a "User interface that uses images to represent options. Some of these images take the form of icons, small pictorial figures that represent tasks, functions, or programs." Windows, which divide the display screen into sections, and additional direct-manipulation (e.g., point and click) devices are also usually featured in graphical user interfaces. The authors state that GUIs are "the easiest interface to use." This is a popular, but largely unexamined assumption today.
Pollitt, et al, would have us believe that, "User interfaces which adopt a graphical or window-based approach to search databases should improve access and the effectiveness of end-user searching. The graphic user interface (GUI) is revolutionising the usability of the personal computer in every application." (1994)
Improved access, improved search effectiveness, and greater usability? These are not small claims. Is there evidence to support these claims? To date, there has been little research on the role of GUIs in improving search performance or retrieval effectiveness. The jury is still out on this one. Pollitt, et al, do warn us that "the enthusiastic application of graphical features has sometimes resulted in unnecessary complexity," complexity that may place unreasonable demands on the user.
2.4.1. Understanding User Interface Design for Information Retrieval Systems.
The user interface component of computerized interactive information retrieval systems like online library catalogs is the locus in time and space, typically defined by a particular mix of hardware and software facilities, where the user and the information system interact and communicate to carry out useful information seeking tasks. In today's online catalogs this user interface is primarily manifest through a particular online catalog's input devices and screen displays. However, these tangible components are only part of the story. The user interface in information systems is a complex environment in which system features must match up appropriately with a bewildering variety of users' personal characteristics, cognitive abilities, and task requirements. In the best of cases, this environment, with its brew of tangibles and intangibles, affords the user a comfortable, supportive "space" to carry out information seeking tasks. These tasks require not only appropriate information input and output, but comprehensible decision making support facilities as well.
Looking for documents or other publications in an online catalog is not just a mechanistic information seeking activity. It is a dynamic, decision making activity which requires that careful consideration be given not only to the information to be provided, but to the manner in which that information is presented in displays, and also to the set of decision making facilities available to assist the user in carrying out primary tasks and sub-tasks. Among these tasks are identifying and locating documents, reviewing them, selecting some as suitable to the need or interest, and using retrieved, found data to modify or continue a search strategy. Thus, a major goal of information system design is to develop a user interface that will facilitate the cognitive tasks of user comprehension and decision making. This goal is only partially accomplished by presenting easy-to-use search input screens and legible displays of bibliographic information.
There is much discussion about the "usability" of computer systems designed for and used by "end users." There seems to be agreement that system design features greatly determine the usability of information systems for their primary clients, and, further, that usability is a dimension that may have a profound influence on both search performance and users' satisfaction with the search system. Given the variety of things one might use a computer system to do, usability is surely a relative measure. Furthermore, as Allen (1993) has noted, "Specific design features can combine with specific user characteristics to ensure that information systems are more usable by some people than others." System designers, especially designers of user interfaces, must take into account the primary tasks to be performed with the system and the characteristics brought to the tasks by the users of the system. An understanding of these tasks and characteristics will inform the design of appropriate information search, presentation, review, selection, and related decision making facilities. Too often in online catalog interface design only one or two of these facilities have been optimized. For example, search input may be simplified, but no dynamic review/feedback facility is provided to support search continuation or enhancement based on information that has already been found and displayed.
Although much has been written about the design and use of online catalog user interfaces and screen displays, actual design is still more of an art than a science. There has been surprisingly little research on the sequencing of online catalog display screens appropriate to a dynamic search and review process, or on information requirements of the process beyond what is displayed as bibliographic information. Online catalog user interfaces have been "acceptance tested" more often in the marketplace than in the laboratory or controlled field experiments. Nonetheless, a great deal of research from related areas and experience gained through 15 years of online catalog interface design, use and evaluation can be brought to bear on the design of user-system interaction styles and methods, and on useful, informative screen displays. (For a useful summary of this research and experience, see Shneiderman, 1992.)
Some attention has been given by researchers to the question of how best to display discrete bibliographic records (presumably resulting from a search) on an online catalog's VDU screen (see for example, Reynolds, 1985, Fryser and Stirling, 1984, Matthews, 1986, Shires, 1992, Allen, 1993a). Both content and presentation issues have been addressed. Great effort has been extended to provide online user assistance and "help" features to ease the use of online catalogs. Less concern has been shown for the dynamic aspects of the communicative, decision making interaction between the user and the system during the search process, and the information and display requirements for supporting that interaction. Such requirements include the proper sequencing or formatting of separate screens, and also include a dynamic, proactive role for individual displays of bibliographic information. In traditional library catalogs, the bibliographic record was thought to be the end-point, or some sort of stopping-point in the search process. Some early online catalogs reflected this tradition by displaying "The End" at the bottom of a screen which displayed a complete bibliographic record.
It has become commonplace to label unique data elements in displayed bibliographic records. Other than this practice, there is as yet no uniform or standard practice followed in the presentation of bibliographic records, with regard to choices of labels for data elements, order of data elements, or screen layout and typography. Previous catalog research has indicated that users frequently do not notice the subject descriptors assigned to a work and included in the bibliographic record, and do not understand their collocative function for identifying similar or related works.
On the subject of information displays, Reynolds (1985) has written:
The initial impression created by any display of information can have a strong influence on users' attitudes towards that information. They will almost certainly form judgements about whether the display is likely to be easy or difficult to use or, indeed, whether it will be worth their while attempting to use it at all.
The design of the full bibliographic record displays should be based on research-informed decisions made about data content, format, order of data elements, labeling, and typography. Data in the MARC record judged to be extraneous to the tasks at hand should be omitted from the displays. Considerations of both task and user characteristics must be included in the remaining aspects of the displays. Reynolds goes on to say, "... a good presentation is, first and foremost, one which makes clear the structure and sequence of the information content and which takes into account the way in which the information will be used." Data field labels should be chosen carefully to avoid jargon and to indicate not only the meaning of the data, but, in the case of the subject headings, their use and function (e.g., "SUBJECT GROUP" rather than just "SUBJECT"). With regard to sequence and structure, the MARC format structure, even with its arcane numeric labels disguised, is not suitable for end users. In future online catalog bibliographic record displays, the title field should be the first in order, and like data elements grouped together, unlike MARC which separates "added entries" from the "main entry." Recent research by Allen (1993b) suggests that displaying subject headings first in the display, at the top of the bibliographic record, improves subject searching performance on some search tasks. He attributes this influence to the perceptual speed factor in identifying appropriate elements in a bibliographic display. In a research project described by Hildreth (1993), a different approach was employed to meet this goal. The goal is to bring the subject headings to the notice of the searcher. Special prompts, labeling, and formatting of the subject data in the bibliographic record, seemed to have had a positive influence on the search performance of inexperienced users. (See Figure 4 )
The typographical conventions followed in the bibliographic displays used in Hildreth's research conform to the findings of Fryser and Stirling (1984). This research showed that users preferred labeled displays and conventional upper-lower case typography for the presentation of bibliographic information.
Effective bibliographic displays are influenced by both content and presentation factors. The design goal is to facilitate user comprehension and decision making. Key decisions involved in the bibliographic search process include accurate identification of a work, suitability of a retrieved work for a particular need, and the desirability of modifying a search strategy or expanding a search. The data content of the records in the database is often out of the hands of the system designer. The designer has to use the available tools to present information in the most useful way contemplated. To date, there has been a paucity of empirical research that addresses issues involved in the effective display of bibliographic information. Although Hildreth's research reported here did not directly address these issues, the users of the experimental online catalog expressed general satisfaction with the displays and reported no difficulties in the use of the test versions that could be attributed to factors associated with the bibliographic displays.
Reflections on the online catalog user interface as a complex environment for supporting search, selection, review and related decision-making activities led this author to the articulation of principles and goals which should guide the design and development of the online catalog interface. The first of these is that the online catalog system should never permit a user's search attempt to fail to retrieve one or more bibliographic records for review and action. Many searches in present-day online catalogs fail to retrieve even a single record, and most online catalogs offer little or no assistance to the searcher when this result occurs. The assumption behind this principle (always retrieve something for display and review) is that something in an heterogeneous online catalog database might satisfy the request to some degree, or serve even in its rejection by the user to supply useful information that can be used to further the search.
A second principle is: never assume the display of a bibliographic record is the end of a search, merely to be selected or rejected, then "set aside." Bibliographic records are for use, not just as location devices, but as information-laden devices for furthering the search. This action role of bibliographic displays is often overlooked in system design. Bibliographic records can be generative; they may have a spring-board effect in the search process, or serve as information "seeds" to fertilize subsequent searching.
Searching and browsing are non-deterministic, dynamic processes; it may be best to think of even the most precisely-formed queries in conventional query-oriented systems as dynamic queries, subject to change in the search process. The user may know precisely what he wants and uses the online catalog merely to locate that particular item and determine its availability. Yet, this single-minded user may choose from a variety of ways of searching for the item, may encounter other interesting items while searching for the desired item, or may even lose interest in the original item as alternatives are brought to his attention. For these reasons, found data -- terms, titles, subject descriptors, entire records -- should be able to serve as useful data for expanding a search or revising a search strategy. In short, it ought to be easy for search output to serve as search input. Display formats and prompts, point and click, and linked-record navigation facilities should be employed in online catalogs to satisfy these principles and requirements.
Research has identified several key problem areas in the use of conventional online catalogs that can be alleviated through interface design. Good reviews of these research findings can be found in Larson, 1991a, and Hildreth, 1989b. Several of these problems are listed below:
1. Initial system entry and orientation. Bates (1986) calls it the "docking" problem.
2. Required use of unfamiliar commands or excessive keyboarding (Trenner. 1991).
3. Entering or finding suitable search terms.
4. Modifying a search strategy or query to achieve better results.
5. No way to provide feedback to retrieved information so that it can be exploited to yield enhanced search results.
6. Interpreting and understanding information in bibliographic displays to support decisions regarding the suitability and usefulness of retrieved items.
7. Knowing where one is in the search process or knowing what may be done next.
A common problem with flexible, hypertext retrieval systems which offer many alternative search paths is the feeling of disorientation users experience after searching for a time. Faced with many choices and paths to pursue, users typically begin to wonder where they are, and how they got there. Lacking sufficient markers and prompts, they often feel lost. This experience is exacerbated in non-linear hypertext search systems that have been implemented in earlier screen technologies developed to support only linear modes of searching. With these earlier technologies, one screen is displayed at a time, containing a single logical unit of information which represents a single stage or level in the search process. Related screens that may provide search and browse context, history, or alternative directions to pursue are simply not displayed simultaneously to the user. Newer graphical user interface (GUI) display technologies offer some solutions to this problem through the use of multiple windows and direct-manipulation devices.
2.4.2. What Do GUIs Bring to OPACs?
Before GUIs we not only had inventive menu-driven interfaces, but some OPACs had rudimentary cursor-controlled "point and click" interfaces that permitted a degree of direct-manipulation of data and functions highlighted on the display screens. No doubt, GUIs are generally more attractive and colorful than character-based interfaces, and they hold the promise of making OPAC searching both easier and more richly interactive.
Some basic features of GUI interfaces are:
Such GUI interfaces are familiar to the growing number of Apple MacIntosh and Microsoft Windows personal computer users. Armed with these new GUI, multi-window capabilities, the designers dilemma can be expressed in this query: "What do we do with the windows?"
The march to GUIs by the OPAC designers and vendors will continue unimpeded, so we must be on guard against the very real possibility of throwing out the baby with the bath water. We must not abandon sound principles as we make cosmetic improvements to the user interface. Point and click interfaces pre-date GUIs. More than window-dressing, users need help in understanding the search process as consisting of complex, interrelated stages and levels of interaction between a variety of kinds of data and functions. For example, more useful than icons and direct-manipulation devices would be:
A user interface design goal should be to provide an intuitive interface that permits more direct, informed interaction on the part of the user with the interrelated stages or levels of a search as it is displayed and seen in context. Unlike the linear straight jacket, no rear-window approach of earlier OPACs, GUI online catalogs can simultaneously present multiple levels of the search territory and permit the user to flexibly pursue his or her own course as his or her interest dictates.
2.5. Searching and Browsing in Second-Generation Online Catalogs
In the use of conventional information retrieval systems, including second-generation OPACs, users are faced with the paradox of information retrieval: the need to describe that which you do not know in order to find it. this is because these systems are query-oriented retrieval systems. Query-oriented retrieval systems require the matching of queries and representations of documents or text. The specifications provided in the query must be satisfied to some extent by any document representations that would make up the retrieval or "results" set. Thus, in a query-oriented retrieval system, one must have a pretty good idea of what one is looking for, what one needs to satisfy the information need, the "object of one's interest," and, one must be able to describe that object linguistically, at least partially, in a way that can be "understood" by the system. There is good reason to question whether or not this search and retrieval paradigm reflects the way most information seeking is actually carried out by individuals.
For purposes of analysis, we classify OPAC searching approaches in the following manner.
I. QUERYING
A. Phrase Matching (Text strings or controlled vocabulary)
B. Keyword Matching (Discrete words, with Boolean or proximity formulations)
II. BROWSING
A. Pre-sequenced, linear, inflexible (Typically, lists of index terms, headings, descriptors, or brief titles)
B. Non-linear, multi-directional, flexible ("navigation", "chain", "bridge", "relational", "hypertext", "serendipitous" browsing)
2.5.1. Query Searching
There are two kinds of query searching: phrase matching and keyword matching. A query consists of a term or terms (e.g., a character, number, word or words, or a phrase) and the specification, sometimes called the query "formulation," which defines how the component term(s) of the query are to be interpreted or related for matching purposes (e.g., word truncation, Boolean combinations, word adjacency). The matching function of an online catalog is the mechanism through which the retrieval software makes a comparison between index terms which represent documents and query terms to effect retrieval. The matching criteria are specified through the query by the user or applied automatically by the system. Query searching of either kind (often called just "searching," to distinguish it from an online catalog's Browse mode) utilizes an exact matching function on the part of the system, regardless of the manner in which the matching criteria are specified.
In this all or nothing approach, documents (bibliographic records in online catalogs) will be retrieved in response to a search only if an exact match of the query is found. The query may consist of a pre-coordinated phrase (with or without truncation) or a post-coordinated Boolean expression of keywords. In either case the Query search matching requirements are precise and rigid. The process is purely mechanistic. The burden is on the searcher to enter terms that will match the entry (index) terms in the database and to specify appropriate proximity or term relationship logic. Bates criticizes this predominant approach to subject searching for requiring a "perfect pinpoint match on the one best term." (1986) No match means no retrieval, as viewers of silent online catalog screens witness too often. The search may fail (i.e., not identify relevant documents that are in the collection) unless the searcher knows or guesses the exact way the term (word or phrase) appears in the subject index.
In keyword, Boolean queries, the system's matching mechanism makes a binary (yes/no) split of the database into bibliographic records that conform exactly to the requirements of the query, and all the rest. Only the former are retrieved as "hits." Partial or "closest" matching operations are generally not supported in second-generation online catalogs and conventional retrieval systems.
Query searching is an appropriate, useful search option when the aim of the search is specific, when the searcher knows precisely what he wants, and when this request can be expressed in the language of the database. Even in subject searching for books or articles on a topic the searcher may know his topic exactly and may be able to express it in the language of the system (e.g., the assigned subject headings or descriptors).
2.5.2. Browse Searching
Browsing in online catalogs can take many forms. Typically, the system displays ordered lists of terms, descriptors, or brief bibliographic records for scanning by the searcher. Lists of index terms are usually presented in alphabetical order. The arrangement of brief citation records may be according to date, and some systems support short record browsing in shelf-list order. Usually the only "navigation" option for browsers is to go backward or forward through the list in a constrained, linear manner. Cross references, if included, represent a way of jumping out of the sequence and over to related areas of the database. Hypertext operations, which permit navigation throughout the database's network of related terms and records, and the dynamic definition of "related areas and interests", have not been implemented in second-generation online catalogs. Conventional browsing assumes a vocabulary aim on the part of the searcher. It assists in identifying the correct form of a term and any related terms. Other forms of browsing, rare in today's online catalogs, support related record or document discovery through non-linear explorations of the database.
Browse searching is the most useful and preferred approach when the search aim is not specific (regarding, for example, discipline or topic, type of publication, level of treatment, perspective, etc.), the desired results are not precisely known in advance, or the correct terms for representing the user's query (which may be vague) are not known at the outset. One or more of these circumstances may be present in most subject searching activities.
2.5.3. Present-day OPAC Subject Searching Summarized
Subject searching in today's largely MARC-based OPACs is supported by a variety of exact-match query methods and browsing facilities. Exact-match query searches are basically of two types: 1) phrase searching on precoordinated subject headings, and 2) keyword searches on cataloging or indexing data in the bibliographic record (e.g., titles, notes, subject headings and their subdivisions). Keyword subject searches may be formulated as Boolean expressions to indicate the desired relationship between search terms, and some OPACs also permit explicit proximity designations in the formulation of the query. The trend is clear: most OPACs support both types of exact-match searching. However, these are hybrid systems that do not integrate or link these separate query options in any useful way during a search. It is usually up to the searcher to choose, through commands or menu selection, one query type or the other.
The syntax and mechanics of entering subject searches have been simplified, and the task of precise search statement, query formulation has largely been delegated to the system software. Choice of subject query type is selected from a menu, or invoked by a pressed function key or simple typed command. Phrase searches are automatically processed as straightforward character string matches or word adjacency, same order matches. In most cases the match must begin with the leftmost significant word or character of the indexed entry. Keyword searches with more than one word are automatically processed as Boolean AND queries in most OPACs. Keyword searches may be targeted by the user or the system to one or more field indexes (e.g., title, notes, series statement, subject headings). Some flexibility is permitted in most keyword search OPACs. When searchers choose not to identify in their queries fields for subject keyword searching (or are not permitted by the system to do so), the trend is toward automatically searching both subject and title. ANDed keywords do not have to appear together in either the title field or the subject field to cause a match. Matching records would include those having one word in the title and another word in the subject heading.
Successful subject phrase searching in most OPACs still requires an exact match on at least the initial, main portion of the subject heading in the catalog record. This usually requires a perfect match on a Library of Congress Subject Heading (LCSH). However, if a LCSH heading is entered as a search term and the heading has been modified or replaced in the local catalog record no match will occur. In some OPACs, when no match occurs on the user-entered term, the system displays headings in the alphabetical neighborhood of the term. This may or may not help the searcher find a heading which expresses his search interest.
In present-day OPACs browsing is even less developed and is generally provided as an intermediate or secondary stage in the subject search process. Some keyword OPACs offer the searcher an unpublished (on the screen) option to browse (scan) alphabetical lists of index terms or headings. Phrase search OPACs typically display these lists after a search is constructed and entered, even when an exact match occurs. Research has not shown whether, in the latter case, this helps or confuses the searcher. A few OPACs are beginning to include cross references and related terms in these browsing displays. This will make them more useful browsing tools for certain kinds of subject searching.
There has been a limited introduction of bookshelf-style browsing via
the OPAC display screen. Unfortunately, the small amounts of book data
typically displayed will not support the kind of actual shelf browsing
practiced by library patrons in open stacks. Moving on at the terminal
to additional online information for a selected document, and then returning
(if permitted) to the browse list can be cumbersome. There is no good reason
to restrict shelf list browse entries to one or two lines of display data,
although this is usually done.
