Introduction The Data Material and the Demolink System

2. Demolink Compared to Automatic and Manual Systems

This paper looks at the application of an interactive record linkage system, Demolink, to a 19th century Norwegian data set, from a historian's point of view [Note 2]. The system resembles earlier semi-automatic systems in some respects. Most important of these is that the linkage decisions are taken by the historian and not by predefined computer algorithms. It differs from them by the more efficient way the computer is used, by the flexibility with which different historical sources can be included, and by its user interface which enables the historian to work in a familiar environment, without expertise in computer science. Like them, Demolink offers an alternative to fully automatic systems, not only in respect to the costs but also to the methodology. The interactive technique is not merely a step towards a fully automatic system; it is considered a method in its own right.

In traditional family reconstruction, events are copied from the sources onto cards, sorted in different ways several times, and the information for each family is collected on a family card or sheet. The work is done manually, following rather mechanically fixed rules. Information is compiled chronologically and from one source at a time. At the very end, loose ends and problematic cases are considered and decided.

In automatic record linkage, event records are compared and possibly linked two by two. The degree of manual inspection of the results varies. The Canadian IREP institute uses such high quality sources that a basically automatic record linkage has proved successful [Bouchard 1986], but it also has a procedure for manual control [Bouchard 1996,33]. In the English sources the information is much poorer. The Cambridge Group's record linkage system chooses according to probabilities or randomly when the information is not sufficiently discriminative [Schofield 1992]. The results are not revised manually. Because many of the cases are impossible to solve no matter what method is used, such a revision may seem unnecessary. Norwegian 19th century sources lie between the two extremes. There is insufficient information for a fully automatic linkage system to be used successfully, but the Cambridge approach would not exploit all the information found in the sources.

Demolink is a general system which can handle nominative, individual information from different sources. Its most important advantage is that it enables the historian to link records from several sources simultaneously. This more dynamic and flexible approach differs both from traditional family reconstruction and from earlier semi-automatic or automatic processes. Demolink enables the historian to approach the data in a more dynamic way. The most secure links are selected first, problematic cases are evaluated throughout the record linkage process, thus producing more secure life histories. The main disadvantage of Demolink is the time required to accomplish the linkage. Some results will be discussed towards the end of the paper to illustrate the success of the method.


Introduction The Data Material and the Demolink System

Interactive Record Linkage: The Cumulative Construction of Life Courses
Eli Fure
© 2000 Max-Planck-Gesellschaft ISSN 1435-9871