Database indexing: yesterday and today

Broadly speaking, if the primary objective of the back-ofthe-book indexer is to help the reader locate specific topics within a single, usually lengthy, document, the primary objective of the database indexer is to help the reader find entire documents, typically journal articles, on specific topics within some large document collection. While a back-of-the-book index is usually part of the document that it indexes, the index terms in the literature database are usually part of a separate database record, which functions as a sort of document surrogate, much like a library catalog card, for the purposes of sorting and searching. For a more comprehensive comparison of the techniques and cultures of these two indexing fraternities, see the recent overview in The Indexer (Klement, 2002). For a closer view of database indexing itself see Greenhouse (2000). I am speaking here of text databases, rather than numeric databases. The latter have quite different purposes and are optimally implemented in quite different software.1 But even with these qualifications it must be admitted that database indexing is a protean activity, not easily defined, serving a wide range of purposes with a variety of approaches an activity whose success often seems to depend on the degree to which the indexer recognizes and adapts to the special circumstances of the indexing task. I have accordingly chosen to emphasize the changing contexts within which database indexing has developed.