2022-07-01 - Using GDELT to Monitor Xenophobic Violence Towards Migrants and Refugees project funded by ODU Data Science Seed Funding Program

We are grateful to have received one of four Data Science Seed Funding awards from ODU to begin an investigation into using worldwide news data from GDELT to monitor xenophobic violence towards migrants and refugees. This is joint work with Erika Frydenlund from VMASC and will support one PhD student for one year.  The goal of the Data Science Seed Funding program was to foster collaborations between on-campus ODU faculty and VMASC researchers in the area of data science.

Michele C. Weigle and Erika Frydenlund, Data Science for Social Good: Mining and Visualizing Worldwide News to Monitor Xenophobic Violence Towards Migrants and Refugees, ODU Data Science Seed Funding, Jul 2022 - Jun 2023, $38,000.

Monitoring the prevalence of xenophobic violence towards migrants and refugees is an important requirement for policymakers to provide support and resources to the areas most in need. Previous approaches have largely involved manual curation of news articles and data or have been focused on a specific location. We will mine the GDELT worldwide news database to develop a dynamically updated visualization dashboard to monitor and highlight reports of xenophobic violence globally. Coupled with publicly available population data, this work can assist policymakers in making informed decisions about the allocation of resources and support. This work will bring together big data from GDELT with innovative data visualizations to highlight how data science can be applied to address real-world problems.

During the course of the project, we will develop visualizations to explore the following initial research questions:

1. Which xenophobic “hotspots” are the most pressing when considering the relative number of refugees and migrants hosted in that country?
2. How do we prioritize the most pressing xenophobic outbreaks of concern before they reach a critical point (i.e., how do we know when they are escalating)?
3. Can we identify triggering events that precipitate increases in xenophobic violence?

We have identified four main tasks for carrying out this exploratory work:

• Task 1: Determine what data is required to address the research questions. We will query the GDELT APIs and databases to identify known xenophobic events through time using the relevant GCAM theme (e.g., “DISCRIMINATION_INTEGRATION” and related subthemes of "XENOPHOBIA" and "ATTACKS_ON_IMMIGRANTS") and CAMEO events (e.g., "182: Physically assault"; "201: Engage in mass expulsion" on Actor code "REF" (for refugees)). We will join this data with publicly available data about country-level and migrant populations from IOM and UNHCR. Analyzing known past events will help us to discover potentially triggering events to monitor for future violent outbreaks.
• Task 2: Consider why the data is needed. We will perform initial exploratory data analysis on the datasets and map the domain research questions to abstract tasks (e.g., compare, summarize, identify).
• Task 3: Develop candidates for how the data can be presented in visualizations. Based on the data and abstract tasks developed in Tasks 1 and 2, we will create an initial visualization dashboard. This will be inspired by existing tools, but guided by modern visualization dashboard design. We strive to develop alternatives beyond heatmaps to clearly present the data to serve the needs of decision makers.
• Task 4: Iterate on the visualization design process. Data analysis through initial visualizations will generate additional domain questions that will then lead to refinements and/or new visualizations. During this task, we will engage humanitarian practitioners and scholars to get feedback about the effectiveness of the developed visualizations.

Erika and I are excited to continue our research collaboration, and I'm looking forward to working with her both on this project and as a Co-PI on her $1.7M grant from the Minerva Research Initiative.

-Michele