Websites in German are a major hurdle for refugees. To better support these people from abuse in the labor market and with questions about housing, the ZHAW has developed a multilingual website with a chatbot. Unlike many digital dialog systems, it works according to the logic of the user.
The war in Ukraine and the increasing number of people seeking protection present the authorities with the challenge of accommodating and caring for a large number of refugees within a reasonable period of time. Most of the refugees from Ukraine are women and children. These are particularly vulnerable groups of people. Especially in the areas of housing and work, this increases the risk of refugees being exploited. ZHAW researchers have developed the “Safety for Refugees” website so that those affected can find their way around better, recognize and classify potentially dangerous situations, and report cases of exploitation immediately.
Help for self-help
The website consists of the rule-driven chatbot EMA, short for “Empowerment Means All.” The goal of this bot is to make the numerous pieces of information distributed across various sources available in technical language in a user-centered manner. EMA triages queries and provides answers in Ukrainian, Russian, German, and English.
The chatbot handles topics related to employment and housing and is limited to the canton of Zurich and protection seekers with refugee status S. “This spatial and staffing limitation is because this is a pilot project for now,” explains project manager Judith Bühler from the ZHAW Department of Social Work. “We need a manageable level of complexity in order to develop a solution and to be able to learn from experience. With EMA, we are testing the first minimally functional product, but we hope to expand it to other cantons, other subject areas and residence permit types.”
Educating about rights and obligations
In contrast to the chatbot of the Migration Office of the Canton of Zurich, EMA is not about an assistance function for the work of an organization, but rather about helping users find their way through the cross-organizational interplay of a wide variety of information and advice services in the areas of housing and work. “Our chatbot empowers with information and thus strengthens self-protection,” says Judith Bühler.
A major challenge in the development of the chatbot was how to optimize the interaction, explains the project’s technical manager, Don Tuggener of the School of Engineering: “Central to this was always that the answers belong in a specific context and are not ambiguous. This avoids misunderstandings and prevents contextually incorrect responses and, in the worst case, mistakenly failing to identify an exploitation situation.” That’s why the project team decided to develop a decision tree and guide users through a prefabricated dialogue flow.
Developed participatively with stakeholders
The development of EMA was supported with an accompanying group of practice organizations as well as a group of protection seekers from Ukraine. These included the specialist organization AOZ, the Fachstelle Frauenhandel und Frauenmigration FIZ, the association Beratungs- und Schulungszentrum Menschenhandel und sexuelle Ausbeutung Act212, the Swiss Refugee Agency SFH and the State Secretariat for Migration SEM.
Safety-for-Refugees was launched as part of the Rapid Action Call “Dealing with Emergency Situations as Consequences of the War on Ukraine” of the Digitalization Initiative (DIZH). The pilot project also serves to learn more about the opportunities and challenges of automated interface management.