Uneven Coverage of Natural Disasters in Wikipedia: the Case of Flood

Valerio Lorini, Javier Rando, Diego Sáez-Trumper, Carlos Castillo

Abstract

Wikipedia, a collaboratively-produced encyclopedia, is a vital source of real-time information during crises. However, its coverage is subject to the biases of its contributors. This study focuses on the English version of Wikipedia to analyze its coverage of global flood events between 2016 and 2019. While Wikipedia effectively documents high-fatality disasters, there is a significant skew toward wealthy, English-speaking nations. Floods in high-income countries are nearly 3.5 times more likely to be covered than those in low-income regions, even when human impact is comparable. These findings highlight a digital divide that could impact automated disaster response systems relying on non-authoritative data.

Introduction

Disaster management increasingly relies on "non-authoritative" data sources, such as social media and Wikipedia, to provide situational awareness when official channels are slow. Wikipedia is particularly effective at capturing information in real-time as crises unfold. However, these platforms are not neutral; they reflect the geographic and socio-economic biases of their editors. This study quantifies the "self-focus" bias in English Wikipedia, investigating whether a flood’s likelihood of being documented depends more on its human impact or the wealth of the country where it occurs.

Methodology

The researchers developed a pipeline to compare "ground truth" disaster data against Wikipedia entries. They consolidated three major disaster databases—Floodlist, EM-DAT, and the Dartmouth Flood Observatory (DFO)—to create a comprehensive global list of floods from 2016 to 2019.

Global distribution of floods recorded in the DFO and the final merged ground truth dataset.

Using a snapshot of English Wikipedia, the team extracted flood-related sentences and used a machine learning classifier to ensure they described specific events. Natural Language Processing (NLP) tools were employed to match these mentions to ground truth events based on location and date. The primary metric, the "Hit Rate," represents the percentage of verified floods that were successfully documented on the platform.

Results: The Wealth and Language Gap

The analysis revealed a near-linear correlation between a country’s income and its representation on Wikipedia. Events in "Very High Income" countries had a hit rate of approximately 68%, compared to just 19% for "Low Income" countries.

Geographic distribution of flood coverage on English Wikipedia, illustrating the significant skew toward the United States and other high-income regions.

Geographically, North America enjoyed the highest coverage (49% hit rate), while South America had the lowest (10.5%), despite frequent flooding in the region. The United States emerged as a massive outlier with a hit rate of 86%. Language also acts as a gatekeeper; countries with over 80% English speakers had a hit rate of ~57%, while those under 20% hovered around 27%. While Wikipedia editors are "fatality-driven"—covering 74% of events with over 100 deaths—high-fatality events in poor countries still frequently go unrecorded.

Conclusion

The digital divide in Wikipedia's disaster coverage has serious implications for global equity in emergency response. If automated detection systems are trained on this biased data, they may disproportionately direct attention toward developed nations while neglecting the most vulnerable populations. Bridging this gap requires a conscious effort to source and translate disaster information from non-English regions and lower-income countries.

Summary updated

Abstract

The usage of non-authoritative data for disaster management presents the opportunity of accessing timely information that might not be available through other means, as well as the challenge of dealing with several layers of biases. Wikipedia, a collaboratively-produced encyclopedia, includes in-depth information about many natural and human-made disasters, and its editors are particularly good at adding information in real-time as a crisis unfolds. In this study, we focus on the English version of Wikipedia, that is by far the most comprehensive version of this encyclopedia. Wikipedia tends to have good coverage of disasters, particularly those having a large number of fatalities. However, we also show that a tendency to cover events in wealthy countries and not cover events in poorer ones permeates Wikipedia as a source for disaster-related information. By performing careful automatic content analysis at a large scale, we show how the coverage of floods in Wikipedia is skewed towards rich, English-speaking countries, in particular the US and Canada. We also note how coverage of floods in countries with the lowest income, as well as countries in South America, is substantially lower than the coverage of floods in middle-income countries. These results have implications for systems using Wikipedia or similar collaborative media platforms as an information source for detecting emergencies or for gathering valuable information for disaster response.
BibTeX
@inproceedings{lorini2020uneven,
  author = {Valerio Lorini and Javier Rando and Diego Sáez-Trumper and Carlos Castillo},
  title = {{Uneven Coverage of Natural Disasters in Wikipedia: the Case of Flood}},
  booktitle = {ISCRAM},
  year = {2020},
}
Figure