Nominal Record Linkage:
the development of computer strategies to achieve the family-based record linkage of nineteenth century demographic data.
A thesis submitted for the PhD degree by John Anthony Welford, the Faculty of Social Sciences, the Open University, July 1989
Previous research in the field of historical nominal record linkage has focused on the linkage of individuals identified in parish register and other personal event records. The major contribution of this thesis is to explore a more advanced, family-based concept of record linkage, in which the unit of linkage can encompass the co-residing family, rather than merely the individual. The techniques developed are to be appropriate for nineteenth century demographic data, and specifically for household census data for the period 1851-71 and for baptism, marriage and burial records from English parish registers.
While certain aspects of the problem of nominal record linkage, such as the need to handle name variations, are well understood, there remains the overriding problem of defining the total problem in a coherent way. The present research uses the development of an operational record linkage system as a vehicle for enabling the total problem to be explored in depth. This development also provides an opportunity to explore the use of novel computing techniques, and, in particular, the contribution made by database management facilities.
A significant change in perspective which is observed is the transition from a concern with 'algorithm' to a concern with 'architecture'. The central task for the developer of a record linkage system is seen to be the design of a number of 'data structure states'. The task of developing 'mechanisms' for providing the necessary transformations between these states is regarded as a secondary and subservient one.
The implementation of viable record linkage strategies requires appropriate computing 'machinery'. The crucial contribution made by the introduction of the database management system is acknowledged, and, in particular, the ease with which it enables complex genealogical structures to be modeled. The development of increasingly generalised linkage strategies will require the provision of appropriate rule-based computing techniques.
Population Index on the Web, Volume 57, Number 2, Summer 1991, Office of Population Research, Princeton University