Artificial Intelligence for Humanitarian Action: Restoring Family Links Project

The project is implemented by the Turkish Red Crescent in Türkiye. It began in August 2020 and ended in February 2021. 

ABYET is a case management system that is usually used by Turkish Red Crescent (TRC) for the  tracing of missing persons, family reunification and family message transmission. ABYET's artificial intelligence integrations can scan all in-house databases to determine if missing persons previously received services from TRC, allowing easy communication if a record exists. In addition, if a missing person’s photo is uploaded to the system, artificial intelligence can easily match the photo of this person regardless of age if there is a previous photo of the person in other TRC databases.

The project began with TRC’s efforts to align its activities with the current technological developments and Microsoft's request for partnership as part of its Artificial Intelligence for Humanitarian Action Social Responsibility Programme, the project aims to achieve more effective results by using artificial intelligence technologies in tracing of missing persons, family reunification and family message transmission services provided by the Restoring Family Links unit of the TRC.

Main activities of the Good Practice

• Searching for missing migrants whose names may be spelled in different ways.
• Blockchain technology has been integrated into ABYET database for secure searching in case of cooperation with external stakeholders (public institutions and NGOs).
• Missing person search matching success has been increased by using photo aging system.

Partners involved

• Microsoft Corporation
• Mart Yazılım

What challenges were encountered in delivering the project and how were they overcome?

Challenges

• Existing integrated AI technology is not active for external database searches of missing people as external stakeholders (public institutions and NGOs) are not integrated into these systems.
• High level of similarity of names creates loss of time in the process of finding the right person among them.
• Not every missing person's family has the photos, thus photo matching may not always be possible.

How they were overcome

• Informing external stakeholders about the security of the system used (this is a suggested solution, and no positive development has been recorded yet).
• Including personal information other than name and surname in the search.
• If possible, asking the family for a photo of the missing person, regardless of whether it is an old or new one.

Results of the Good Practice

• Searching for missing persons in a short time and shortening the response time for beneficiaries.
• Increasing the likelihood of finding missing persons who could not be found in database searches before due to spelling errors.

In what way does the good practice meet one or more of the four objectives of the Global Compact on Refugees?

Objective 1: Ease the pressures on host countries

Türkiye is a transit country, therefore there is a very high caseload related to migration. While it takes months and sometimes even years to get results in conventional missing person searches, digitalization and artificial intelligence technologies enable results to be obtained in a much shorter time, preventing case backlogs. 

Next steps

It is expected that we will hold cooperation meetings with other institutions in order to obtain better results by applying the photo matching and missing person search mechanisms through integration of the system and information into their databases. Furthermore, improvement works will be carried out to address some deficiencies and errors of the system.

Are there areas in which support would be required to continue and/or scale up your good practice?

There is a need to make improvements in the database to ensure better execution.

Kamil Erdem Güler, Head of Programme Development and Coordination, Turkish Red Crescent 

[email protected]

[email protected]