Improved browsable displays: an experimental test
Bryce Allen · Information Technology and Libraries · 1993
Browse searching of online catalogs and electronic indexes is a popular means of retrieving information, but some subject heading lists are so extensive that browsing them can be difficult and time consuming. This research tested one way of presenting a browse interface to reduce the amount of scanning of subject headings required. It found that a hierarchical presentation of heading reduced the amount of scanning required by more than 50 percent without any reduction in the effectiveness of the search. Better browsable display of this sort can improve efficiency of searching but appear not to alter effectiveness. Browsing in automated catalogs and indexes is a popular means of finding information, probably because it is relatively uncomplicated. Typically, a user enters a search expression, such as an author's name, a title, or a subject heading. The information system then displays a sorted list of authors, titles, or subject terms, with the user's search expression highlighted. The user is able to scan up and down the list, selecting one or more items from the list for further processing. This additional processing usually involves retrieving and scanning the references that are indexed by the selected terms from the browse list. The simplicity of this kind of access can be compared with the relative complexity of search systems requiring the use of Boolean operators or identification of building blocks and synonyms. As a result, many users of CD-ROM indexes and OPAC systems rely upon browse searching to retrieve information. There are, however, problems with browse access. One of these problems is that some information systems have so many index entries that it is a time-consuming and difficult task to scan through them to find appropriate terms or headings. This seems particularly true in subject access. In most cases, browse access to subjects means scanning the subject index of the OPAC system or the the-suarus of the CD-ROM system. With large databases such as those found in OPACs, the subject index may be so large that scanning even a relatively restricted section of it is difficult. Massicotte, noting the results of OPAC evaluations and research into subject access by Atherton Cochrane and others, pointed out that even for a relatively restricted topic it may be necessary for a user to scan through hundreds of subject headings in order to obtain an overview of the topic.[1] She suggested that it would be better to restrict the number of subject headings that users have to scan initially, ideally to a single screen of headings for any topic. Her suggestion was that the browse display be reduce by replacing many headings with broad conceptual categories. For example, the many geographical subdivisions note, Subdivided by geographical areas. McGarry and Svenonius followed up this suggestion by exploring two in which restricted lists of subject headings could be created to minimize the scanning of subject headings that was necessary for subject browsing in online catalogs.[2] The first was Massicotte's geographical compression, and the second was an even more radical compression they called blanket compression. In blanket compress, all subheadings were to be suppressed in initial displays of subject headings by using an algorithm that would recognize repeating first elements of headings. In addition, they suggested that the problem of divided displays be resolved by incorporating into a single alphabetical sequence the different ways that subject headings are punctuated. As part of a larger series of investigations, this project explored experimentally the effects of providing this type of enhancement to browsable subject access. In particular, the first element of McGarry and Svenonius' blanket compression was implemented in an experimental system in such a way that subdivisions were initially suppressed, but could be displayed on demand. As McGarry and Svenonius noted, this type of compression is easy to implement in a semi-automatic manner, as opposed to the conceptual compression suggested by Massicotte, which would require more detailed programming to implement. …