Data-driven intelligence in crisis: The case of Ukrainian refugee management
本研究基于瑞士58场半结构化访谈，开发了R2G应用，利用社区数据和自然语言处理技术（包括聊天机器人）为政策制定者提供实时难民需求识别，并提炼出四项设计原则。
The ongoing conflict in Ukraine has triggered a humanitarian crisis, leading to a substantial increase in refugees. This situation presents a significant challenge for European countries, emphasizing the urgent need for effective refugee management strategies. Hence, effective decision-making is needed for the public sector to create a better livelihood for refugees. In this study, we propose using the concept of intelligence defined by Herbert Simon for effective refugee management. Following the Design Science Research Methodology, we utilize 58 semi-structured stakeholder interviews within Switzerland to identify problems and define design goals that facilitate intelligence in refugee management. Based on the design goals, we developed R2G – “Refugees to Government”, an application that utilizes community data and state-of-the-art NLP, including a chatbot interface, to offer an interactive dashboard for identifying refugee needs. The chatbot allows policymakers to interact with refugee data through dynamic, conversational queries, enabling real-time identification of refugee needs and providing data-driven intelligence. Our assessment of R2G, facilitated through 28 semi-structured interviews, resulted in four design principles for data-driven intelligence in refugee management: community-driven insight, spatial-temporal knowledge, multilingual data synthesis and visualization, and interactive data querying through chatbots. Additionally, we provide policy recommendations emphasizing the ethical use of community data, the integration of advanced NLP techniques in government processes, and the need for shifting governmental roles towards data analytics. • Our study applies Herbert Simon's concept of intelligence—the first step in decision-making—to refugee management. • We identified problems in current refugee management through 58 stakeholder interviews and developed solution objectives. • We created R2G ("Refugee to Government"), a dashboard app generating bottom-up intelligence for refugee management. • R2G uses community data a state-of-the-art natural language processing in the form of topic modeling and chatbot functionality. • We evaluated R2G with 28 interviews, abstracting design principles aligning with Herbert Simon's concept of intelligence. • We propose policy implications that the use of a tool like R2G brings.