Charles R. Hildreth, Ph.D.
1. INTRODUCTION
Not long ago I issued this challenge to my colleagues at an invitational conference on the present and future of the online catalog: "Just say "NO" to present-day, '2nd-generation' online public access catalogs (OPACs). Take a stand for a better way of accessing and browsing library collections online, and help us advance beyond our current position on the developmental plateau of premature satisfaction and myopic complacency. (Hildreth, 1991)
Why was this challenge necessary? Today's online catalogs are very popular with users. This young library access technology may even be, in some sense, addictive - users return to it again and again. In study after study it is reported that users much prefer online library catalogs over card or microfilm catalogs. These same studies also report that users are usually satisfied with the search performance of OPACs and enjoy using them. This development alone is reason enough to pause and celebrate the tremendous OPAC-related accomplishments of the past fifteen years. We all - librarians, researchers, innovative system designers, and vendors - have come a long way in a very short period of time. Reflecting on the early, cautious visions of the card catalog online (Figure 1), we can now point with pride to the achievements gained and the expanded service benefits provided by the best of today's installed OPACs.
In search flexibility and power, today's OPACs are nearly the equal of the conventional, commercial online database search systems, but they are seriously flawed and not nearly good enough access tools for their primary users, those users we simplistically refer to as "end-users." Research findings that document these flaws and shortcomings continue to accumulate. It is true that OPACs have not "stood still" in design and development. Improvements in recent years include expansions in data content and collection coverage, network access, and improved ease of use or user assistance. But little progress has been made in the areas of OPAC functionality and retrieval effectiveness.
Present-day OPACs may be easy and fun to use (compared to the card catalog), but they are unnecessarily restrictive of search behavior and, when measured objectively, they do not perform as well as first thought. To put the case bluntly, when searching today's OPACs, looking is easy; finding what you want presents the difficulty!
Unfortunately, most present-day operational and experimental retrieval systems, including most online catalogs and CD-ROMs, reflect in their design and operation a partial, inadequate conceptual model of information retrieval activity. This model describes the exact or best-match, product-oriented approach of most IR systems. The model assumes the presence of a known, specifiable information need (or subject topic) to start with. Materials that are relevant to that need or topic are represented by index terms such as keywords or subject descriptors, and the need is represented in a well-specified query. These representations are then "best-matched" by the retrieval system (set in motion by the search specialist) to produce the best output set of retrieved materials or information. A second assumption is that for any given query there exists a single best output set that should be targeted for retrieval. Keen (1994) speaks of the "one-shot search" approach of ranked retrieval with fully automatic query expansion, and recommends instead interactive human relevance feedback whereby a "query off target is resighted." Barely concealed here is the image of the search intermediary as the sport rifleman target-shooting specialist.
This "known-subject need, best-match, end product-oriented" information retrieval paradigm accounts for only part of the subject searching story. It is conceptually inadequate for explaining a variety of information seeking situations or for describing different actual subject searching behaviors. For these reasons, a number of researchers, among them Cochrane, Bates, Belkin, Hjerppe, Oddy, Marcus, Markey, Tague, Hancock-Beaulieu, Marchionini, Kuhlthau, Cox, and Cove and Walsh have proposed other retrieval paradigms or conceptual models equally well-suited to guide the design of information retrieval systems. Bates (1986a) has proposed the "exploratory paradigm" to describe unfocused information seeking and other forms of browsing. The insight and assumption shared by these researchers is that browsing, a complex activity in itself, is a primary, frequent or preferred mode of subject searching for many individuals (Hildreth, 1982).
If our emerging IR and OPAC systems are to more fully reflect and support actual searching behavior and a variety of subject searching styles and objectives their design must be informed by models and user requirements analyses that closely describe these kinds of behavior and objectives.
It would be fruitful if we would consider that there are a variety of types of subject searching objectives and requirements searchers may bring to our retrieval systems. When better understood these different aims and requirements may explain the different kinds of subject searching behavior displayed by individual searchers. The lesson is: watch the users in a number of information seeking situations, using a variety of information resources and retrieval systems. Our user research tends to be "tree-focused" rather than "forest-oriented." Armed with this knowledge (rather than assumptions), we will be better qualified to make proposals for specific search features and interfaces appropriate for different kinds of subject searching.
We must not spend the better part of another century - or even another decade - satisfied with or tolerant of this important but flawed "half-way" technology. The Council on Library Resources is to be commended for its efforts to disseminate this knowledge, namely, that present-day OPACs require major improvements and further development, and that these objectives are both practical and obtainable.
This Report will critically examine today's second-generation expanded and extended online catalogs, including the new GUI (graphical user interface) OPACs. The search features and shortcomings of this generation of online catalogs will be discussed in some detail. Alternative design models will be described to reveal their strengths and weaknesses, and a new non-query-based, "exploratory" information seeking model will be recommended as the model to be incorporated in the design of next-generation online catalogs. This Report is written for information system designers, librarians, and all others who have the potential to influence our online future.
