Search results

Many of us are aware of the yawns of waiting for dBase to look sequentially through a 10,000‐record database for a term or terms hidden in a title or descriptor field. I have tried to persuade dBase III to perform a little more smartly with some degree of success. Users will know that dBase retrieves almost instantaneously from indexed fields but it is not logical to index long, ‘multi‐value’ fields like titles or abstracts. What we have done, however, is to extract keywords from these fields or other descriptive parts of the document and build up a parallel keyword file consisting of two fields — keyword and accession number, which provide a link field with the main database. Thus, each entry in the main database may have a number (in our case up to 7) of corresponding entries in the keyword file, which is itself indexed for fast retrieval.