Matching of citations between non‐standardized databases

The work presents the results of a novel approach to automatic matching and linkage of citations between nonstandardized databases, that is, databases which contain untagged attributes. In order to provide a reasonably high incidence of overlap and therefore fertile ground for citation‐matching, a databank was created following on‐line retrievals from different databases but on the same subject area. Matching was achieved through coding of pertinent attributes and the derivation of fixed‐length retrieval keys. A total of 14 different coding techniques were applied and their effectiveness was measured in terms of their efficiency in retrieval of ‘duplicates’ with minimal false hits, their reliability in retrieving all matching citations known to be present, and their performance relative to the ideal. © 1990 John Wiley & Sons, Inc.