- Determine where human rights violations are occurring and identify spikes in activity across the globe
- Better understand how the Benetech database of violations has changed over time and where there may be early indicators of impending strife
- Predict where future violations are likely to occur based on patterns in the data, and potentially identify violations in near-real time
- Created reusable scripts for data summarization and visualization alllowing Benetech to interpret their data in new ways
- Built automated maps to show sudden outbreaks in violations, including hotspots in South America and Asia around conflict regions
- Helped Benetech understand how their database is changing over time and where there may be early indicators of impending strife
Benetech is a nonprofit dedicated to using technology to root out human rights violations. Using a tool called Martus, Benetech empowers people worldwide to submit anonymous posts about violations. Activists and organizers use this data to battle injustice and strengthen human rights internationally.
Benetech wanted to find out what they could learn about the patterns of violations that had been uploaded to their database. Specifically, they wanted to understand where the violations occurred and identify spikes in activity. Could they use their massive data source to see how often people post?
Our expert volunteers, led by DataKind co-founder Drew Conway, worked with representatives from Benetech during a San Francisco DataDive in 2011.
The team created a set of codes used to recognize patterns in data, called scripts, to automatically generate daily reports for Benetech about the type of public human rights data that had been uploaded to Martus. The scripts can then be re-run against new data as it flows into the Martus servers, allowing Benetech to see the time and location of sudden bursts in activity. The team also found ways to build automated maps to show sudden outbreaks and patterns about spikes in activity on a regional level.
The automatically generated reports and maps have enabled Benetech to better identify early indicators of impending strife.
Even though the data from Martus is (intentionally) quite opaque to protect the users involved, the Benetech and DataKind volunteers were able to pull out a huge amount of public information. The automatic reports and maps can alert Benetech to sudden changes in activity and adapt the maps to help identify high-activity regions in near-real time.