This study explores xenophobic events related to refugees and migration using the GDELT 2.0 database and APIs through visualizations. We conducted two case studies -- the first being an analysis of refugee-related news following the death of a two-year-old Syrian boy, Alan Kurdi, and the second a surge in news articles in March 2021 based on the data obtained from GDELT API. In addition to the two case studies, we present a discussion of our exploratory data analysis steps and the challenges encountered while working with GDELT data and its tools.
Exploring Xenophobic Events through GDELT Data Analysis
1. Exploring Xenophobic Events through
GDELT Data Analysis
Modeling and Simulation Student Capstone Conference 2023
Track: Data Science
Himarsha R. Jayanetti, Erika Frydenlund, Michele C. Weigle
Presented By:
Himarsha R. Jayanetti
Department of Computer Science, ODU
Web Science & Digital Libraries Research Group
@HimarshaJ @WebSciDL @oducs
1
2. Global Migration: Seeking Opportunities and/or Escaping Conflicts
2
https://unis.unvienna.org/unis/en/topics/refugees-migration.html
3. International Organization like UNHCR and IOM: Safeguarding Migration and
Fighting Xenophobia
3
https://www.iom.int/
https://www.unhcr.org/us/
4. The Need for a Global Tool to Track Xenophobic Attacks on Migrants
4
https://acleddata.com/dashboard/#/dashboard
https://www.xenowatch.ac.za/
5. We Used the GDELT 2.0 Database to Obtain the Data
5
https://www.gdeltproject.org/
Database Updates every
15 minutes and translates
articles from around the
world from 65 different
languages into English.
A massive dataset of
online and TV and news
reporting.
6. We Identified Three Key Tables in the GDELT Database
● Event table contains data
about events happening
globally.
● Event Mentions table each
mention of the event in a
news article or other
source.
● Global Knowledge Graph
(GKG) form an extensive
interconnected network
(events, context, actors,
sentiment of the article,
themes)
6
Event
Event Mentions
Global Knowledge
Graph (GKG)
7. Relationship Among the Three Database Tables: Events, Event Mentions, and
GKG
7
MentionIdentifier field to
connect the Mentions
table with the GKG
table.
A One-to-Many relationship
between the GLOBALEVENTID
in the Event table and the
Event Mentions table.
8. We Used Google BigQuery to Collect Data
8
https://cloud.google.com/bigquery
GDELT users can access
the data using BigQuery,
which uses SQL-like
queries.
GDELT database is a large,
open-source database and
is supported by Google.
A simple query to get 10
GLOBALEVENTIDs of
event data from the
Event table
9. We Used Two Separate Data Collection Criteria
Criteria 1: Events where Actor2Code is REF.
Criteria 2: Events where Actor2Code is REF AND has Refugee
related GKG themes
Eight GKG themes relevant to refugees:
• DISCRIMINATION_IMMIGRATION_XENOPHOBIA
• DISCRIMINATION_IMMIGRATION_ANTIIMMIGRANTS
• DISCRIMINATION_IMMIGRATION_OPPOSED_TO_IMMIGRANTS
• DISCRIMINATION_IMMIGRATION_AGAINST_IMMIGRANTS
• DISCRIMINATION_IMMIGRATION_ATTACKS_ON_IMMIGRANTS
• DISCRIMINATION_IMMIGRATION_ATTACKS_AGAINST_IMMIGRANTS
• DISCRIMINATION_IMMIGRATION_XENOPHOBE
• DISCRIMINATION_IMMIGRATION_XENOPHOBES 9
Actor2Code is REF is
indicative of Actor 1
performing the act
on ‘refugee’.
These eight themes
were named as the
“GKGthemes_REF”.
10. We Conducted Two Case Studies
Case Study 1: Alan Kurdi Incident.
● September 2015
● Alan and his family (refugees from Syria)
● Travel to Europe from Turkey
● Lost life by drowning in the Mediterranean Sea
Case Study 2: Spike in Number of Articles in March 2021.
● A surge in the number of articles in March 2021.
● 1st hypothesis: due to spa shooting in Atlanta, GA.
● GDELT API to gain insight into the data before
querying for the data itself.
10
We downloaded the data
six months before and
after the incident (March
2015 to March 2016)
where Actor2Code is REF.
We downloaded the data
for March of 2021 where
Actor2Code is REF and the
theme was one in the
“GKGthemes_REF” set.
11. GDELT DOC 2.0 API To Query the GDELT Database
11
https://api.gdeltproject.org/api/v2/doc/doc?query=xenophobia
&mode=artlist&maxrecords=100×pan=1week
● Full text search API
● Search back to Jan 2017
● Search across 65 languages
● Instant Visualizations
● Support JSON output
https:/blog.gdeltproject.org/gdelt-doc-2-0-api-debuts/
Queried on April 18, 2023
12. We Used the GDELT Doc API Python Client to Fetch Data From the GDELT API
12
https://github.com/alex9smith/gdelt-doc-api
1. Article search
● A list of news articles
that match the filters.
1. Timeline
● A timeline of the
volume of news
coverage matching
the filters (number of
articles and a total).
13. Percentage of Number of Articles with Refugee Related Themes From January
2017 to December 2022
13
21. Future Work
● Develop a dashboard to monitor xenophobic violence against refugees and
migrants.
● Developing impactful visualizations that can aid us in addressing the
following:
○ Identify xenophobic “hotspots”
○ Identify when xenophobic outbreaks are escalating
○ Triggers of Xenophobic Violence
21
22. Key Takeaways
22
● Understanding xenophobic events
● GDELT database
● BigQuery and GDELT APIs to access GDELT data
● Case Study 1
○ The period surrounding the death of Alan Kurdi
○ Increase in media attention
○ Mostly negative sentiment
○ A shift in the range of emotions
● Case Study 2
○ March 2021 spike
○ Increase of African migrants on the Canary Islands
○ A choropleth map for the location-based analysis.