SynonameTM: A Personal Name-Matching Program for Use in the Humanities
Susan L. Siegfried · Literary and Linguistic Computing · 1992
Names are a key entry point for researching and indexing historical information. However, the format and spelling of personal names varies greatly from one institution to the next, reflecting traditional differences in practice. If historical information is to be compiled or shared, matching different versions of personal names is a necessity. The computer program described here automatically matches many possible forms of a single personal name by using an ordered sequence of twelve algorithms for pattern matching that include both character- and word-matching techniques. The matched pairs of names are considered to be ‘candidate matches’ until confirmed by a human name-authority editor. Run against a merged file of artists' names from museum collections data, the program performed with an accuracy rate of 97. 4% and an optimum efficiency rate of 90. 8%. Accuracy can increase to nearly 99% at the expense of some efficiency. The concepts behind the algorithms and their implementation may be useful to others merging data in different contexts