Metadata And Subject Searching
[ Home Page - Surveys - Documents - Presentations - Search ]
Digital collections are only likely to make an impact on the Web if they are presented in such a way that users can retrieve their component parts quickly and easily. This is true even if they have been well selected, digitised to a suitable standard and have appropriate metadata formats. Subject-based access to the collection through searching and/or browsing a tree-like structure can greatly enhance the value of your resource.
Subject-based access can be provided in several ways:
Keywords: A simple but crude method is to anticipate the terms that an unguided searcher might intuitively choose and insert them into a keyword field within relevant records. For instance, the text of Ten days that shook the world [1], a classic narrative of the events of 1917, is more likely to be retrieved if the keywords Russian Revolution are added by the cataloguer (based on his/her analysis of the resource and subject knowledge) and if the keyword field is included in the search. In the absence of an agreed vocabulary, however, variant spellings (labor versus labour), and synonyms or near synonyms (Marxist versus Communist) that distort retrieval are likely to proliferate.
Thesauri and subject schemes: Controlled vocabularies, known as thesauri, can prevent inconsistent description and their use is recommended. They define preferred terms and their spelling. If the thesaurus structure is shown on the search interface, users may be guided through broader-term, narrower-term and associated-term relationships to choose the most appropriate keyword with which to search. Take care to choose a vocabulary appropriate to the scope of your resource. A broad and general collection might require a correspondingly universal vocabulary, such as the Library of Congress Subject Headings (>LCSH) [2]. A subject-specific vocabulary, such as the Getty Art and Architecture Thesaurus (AAT) [3], may provide a more limited but detailed range of terms appropriate for a tightly focused collection.
Classification schemes: Keywords and thesauri are primarily aids to searching but browsing can often be a more rewarding approach - particularly for users new to a given subject area. Thesauri are not always structured ideally for browsing as when related or narrower terms are listed alphabetically rather by topical proximity. Truly effective browsing requires the use of a subject classification scheme. A classification scheme arranges resources into a hierarchy on the basis of their subject but differs from a thesaurus in using a sophisticated alphanumeric notation to ensure that related subjects will be displayed in close, browsable, proximity. A well-designed classification scheme should present a navigable continuum of topics from one broad subject area to another and in this way guide the user related items that might otherwise be missed, as in this example from the Dewey Decimal Classification (DDC) [4].
| 700 | Arts, fine and decorative |
| 740 | Drawing and decorative arts |
| 745 | Decorative arts |
| 745.6 | Calligraphy, heraldic design, illumination |
| 745.66 | Heraldic design |
The notation does not necessarily have to be displayed on screen, however. The subject terms, rather than their respective numbers, may mean more to the user. Another tip is to assign multiple classification numbers to any item that crosses subjects. Digital items can have several 'virtual' locations, unlike a book, which is tied to a single position on a shelf.
Keywords, thesauri and classification can be used in combination or individually.
The most important consideration when choosing a classification scheme is to select the one that best fits the subject, scope and intended audience of your resource.
Universal classification schemes: These are particularly appropriate where collections and their audiences span continents, subjects and languages. Dewey Decimal Classification (DDC) [5], for instance, is the most widely recognised scheme worldwide, whilst UDC (Universal Decimal Classification) [6] is predominant in Europe and Asia. Well-established schemes of this sort are most likely to have user-friendly online implementation tools.
National or subject-specific schemes: More specific collections are usually best served by schemes tailored to a single country (e.g. BC Nederlandse Basisclassificatie) [7], language, or subject (e.g. NLM National Library of Medicine) [8]. If nothing suitable exists, a scheme can be created in-house.
Homegrown schemes: Project-specific schemes can be flexible, easy to change and suited wholly to one's own needs so that there are no empty categories or illogical subject groupings to hinder browsing. However, the development process is costly, time-consuming and requires expert subject-knowledge. Users are sure to be unfamiliar with your categories and, perhaps worst of all, such schemes are unlikely to be interoperable with the broader information world and will hinder wider cross searching. They should be regarded very much as a last resort.
Adapting an existing scheme: A better approach is normally to adapt an existing scheme by rearranging empty classes, raising or lowering branches of the hierarchy, renaming captions, or extending the scheme. Be aware, though, that recurring notation may be found within a scheme at its various hierarchical levels or the scheme might officially be modified over time, both of which can lead to conflict between the official and customised versions. Take care to document your changes to ensure consistency through the lifetime of the project. Some well-known Internet search-services (e.g. Yahoo!) [9] have developed their own classifications but there is no guarantee that they will remain stable or even survive into the medium term.
Double classification: It may be worthwhile classifying your resource using a universal scheme for cross-searching and interoperability in the wider information environment and at the same time using a more focused scheme for use within the context of your own Web site. Cost is likely to be an issue that underpins all of these decisions. For instance, the scheme you wish to use may be freely available for use on the Internet or alternatively you may need to pay for a licence.
[ Home Page - Surveys - Documents - Presentations - Search ]
Web page by Brian Kelly of UKOLN.
Last Modified 29-May-2005
Email comments to webmaster@ukoln.ac.uk