Genealogical Record Linkage

Record Linkage is the process of integrating information from two or more independent sources, ensuring that records believed to represent the same ob ject are matched and treated as a single entity. In the context of genealogical databases, record linkage seeks specifically to identify whether or not individuals belonging to different pedigrees refer to the same person.

As the custodian of the largest source of genealogical data, the Church of Jesus Christ of Latter-day Saints' Family History Department (FHD) has a keen interest in accurate record linkage to (a) guarantee the integrity and overall quality of the available data, and (b) facilitate researchers' work by merging together complementary information.

We have designed a solution, MAL4:6 (Mining And Linking for Successful Information eXchange), based on a two hidden-layer structured neural networks, which makes use of both individual data as well as family relationships. The results were encouraging, especially in the case where individuals were underrepresented but had rich pedigrees.

We are working on improving this solution, as well as on studying the applicability of Bayesian networks in this context.

Click here to view our publications on geneological record linkage.