How to Link Housing/Homelessness Agency Client Records and Preserve Privacy

Canadian Alliance to End Homelessness Conference, October, 2023.

A person interacting with the homelessness and housing system of care often interacts with many different agencies. These agencies usually record data documenting that interaction but these data sets are often not linked together. However, linking is important to understand how people interact with the system as a whole. Since housing and homelessness systems typically do not assign people with a unique identification number, the only way to link records across agencies is using names and birthdates. This presentation describes how to scramble names and birthdates to protect privacy but also still link records where names or birthdates may have been mispelled or entered incorrectly.

NOTE: Bloom filters are the scrambling technique used in this presentation. The basic Bloom filter has known security vulnerabilities and should not be considered a strong form of encryption. Data scrambled with basic Bloom filters should be protected with additional encryption and security measures. Stronger forms of Bloom filters do exist as described in more detail here.