How does the Re:Match algorithm work?
Re:Match is a project that pilots and tests an algorithm-based matching process in order to enable an informed, individualized and needs-oriented relocation of protection seekers to welcoming cities.
Together with our partner Pairity, we’ve developed a unique and specialized algorithm to implement this. Due to this algorithm the Re:Match process is able to include preference-ranking and therefore to give the protection seekers more agency in the settlement decision-making process. On the other side of the match, welcoming cities participate in this innovative approach to relocation by feeding data about their current services and capacities into the process. Find out more about the Re:Match algorithm on this page!
The algorithm used by the Re:Match project optimises the collective welfare of Re:Match participants by assigning best possible matches given their attributes and preferences, and the services and capacities of participating cities. This procedure ensures fairness without prioritizing a certain participant’s preferences ahead of the overall group welfare, while maximizing the returns from scarce municipal resources.
The Re:Match algorithm hence finds the best direct match between participating cities and individuals/families within a participant group (cohort) based on
- Personal data and individually weighted preferences of the protection seekers, and
- Municipal data on local conditions, services, and needs, as well as their dynamic accommodation types/capacities.
The Re:Match multi-dimensional matching procedure systematically incorporates cultural and social factors critical for protection seekers overall welfare and integration, in addition to factors determining economic integration. It avoids group-level assumptions and biases about what newcomers need, since individuals with similar characteristics hold different priorities and aspirations. City data taken into account by the algorithm includes the number and types of accommodations in municipal facilities, housing markets, educational, employment, and training opportunities, available cultural services, and capacity to support medical vulnerabilities.
Practically, the algorithm simulates all potential assignments to identify the scenario achieving the highest overall quality of matches for an entire cohort. Computationally, a group of ten participant households/individuals and 6 cities requires simulating approx. 7 million match scenarios.
Want to learn more about the algorithm? Check out chapter 4 in our interim evaluation report!
