COMRADES · Project

AI Platform That Filters Crisis Information from Social Media in Real Time

digitalTestedTRL 5Thin data (2/5)

Imagine a major flood or earthquake hits and thousands of people start posting on social media at once — calls for help, rumors, photos, updates in different languages. Right now, emergency teams drown in this noise and can't tell what's real from what's fake. COMRADES built a smart platform that automatically sorts through social media messages during a crisis, flags what's important, verifies information, and tracks small emergency events as they unfold. Think of it as a spam filter, but for disaster response — it keeps the signal and throws out the noise.

By the numbers

consortium partners across 5 countries

total project deliverables produced

demo deliverables including platform and AI classifiers

countries represented in the consortium

The business problem

What needed solving

During emergencies, organizations face a firehose of unverified social media posts, SMS messages, and citizen reports — with no fast way to separate critical intelligence from noise. Existing tools like Ushahidi collect and map data but cannot automatically analyze, verify, or prioritize information in real time. This gap costs lives, delays response, and leaves decision-makers flying blind during the hours that matter most.

The solution

What was built

The project built the COMRADES platform (reaching a third version), which extends the open-source Ushahidi crisis mapping tool with AI capabilities. Key components include web services for classifying social media and SMS message relevance and informativeness, classifiers for detecting and categorizing micro emergency events, and multilingual crisis information processing — all backed by 15 deliverables including accuracy evaluations.

Audience

Who needs this

Civil protection and emergency management agencies needing real-time social media intelligenceInsurance companies requiring faster catastrophe damage assessment from citizen reportsNews organizations and wire services verifying breaking crisis information at scaleHumanitarian NGOs coordinating disaster response across multiple regionsSmart city platforms integrating citizen-generated crisis data into operations centers

Business applications

Who can put this to work

Emergency Management & Civil Protection

enterprise

Target: National civil protection agencies and disaster response organizations

If you are a civil protection agency dealing with information overload during emergencies — this project developed an AI-powered platform that automatically classifies social media and SMS messages by relevance and informativeness. It filters unverified content and tracks micro-events in real time, so your response teams act on verified intelligence instead of drowning in noise.

Insurance & Risk Assessment

enterprise

Target: Property and casualty insurers with catastrophe exposure

If you are an insurance company dealing with slow damage assessment after natural disasters — this project developed classifiers for emergency events that can detect and categorize crisis micro-events from social media data. This means faster situational awareness during catastrophes, enabling quicker claims triage and more accurate loss estimates while events are still unfolding.

Media & News Agencies

mid-size

Target: News organizations and wire services covering breaking events

If you are a news organization dealing with verifying breaking crisis reports from citizen journalists — this project developed content informativeness classification services that assess the relevance and reliability of social media posts during emergencies. Instead of manually checking hundreds of tips, your newsroom can prioritize verified, high-value reports automatically.

Frequently asked

Quick answers

What would it cost to license or deploy this platform?

The COMRADES platform is open source, built on top of the widely-used Ushahidi crisis mapping tool. Deployment costs would involve infrastructure, customization, and integration rather than licensing fees. Based on available project data, no commercial pricing model was established during the project.

Can this scale to handle a national-level emergency with millions of social media posts?

The platform was designed for real-time social data analysis and multilingual crisis information processing. It was deployed and evaluated with multiple local and distributed communities during the project. Scaling to national-level volumes would require additional infrastructure investment beyond what was tested.

What is the IP situation — can we use this commercially?

COMRADES was funded as a Research and Innovation Action (RIA) and the platform is open source, extending the existing Ushahidi tool. The classification algorithms and AI services developed across 15 deliverables may have specific IP terms. Contact the consortium for commercial licensing details on the proprietary AI components.

Does this work with our existing emergency management systems?

The platform uses Linked Open Data standards and was designed to go beyond Ushahidi's existing data collection and mapping functions. Based on available project data, it provides web services for content classification and event detection, suggesting an API-based architecture that could integrate with other systems.

How accurate are the AI classifiers for crisis messages?

The deliverables include accuracy tests and evaluation against state-of-the-art methods for both content informativeness classification and emergency event classification. Specific accuracy percentages are not available in the project summary, but independent evaluation reports were produced for each classifier.

Is this ready to deploy today or does it need more development?

The project delivered a third version of the COMRADES platform by its end in 2018. While functional and tested with communities, the project closed as a research initiative without evidence of commercial deployment. The technology would likely need updating and productization for current operational use.

Consortium

Who built it

The COMRADES consortium brings together 7 partners from 5 countries (Belgium, Kenya, Netherlands, Norway, UK), led by The Open University in the UK. The team is heavily academic with 4 universities and 1 research organization, complemented by just 1 industry partner (14% industry ratio) and no SMEs. The inclusion of Kenya is notable — it suggests real-world testing in crisis-prone regions, adding credibility to the platform's practical relevance. However, the low industry participation and zero SME involvement signal this was primarily a research effort, meaning commercial readiness will require significant business development beyond what the consortium delivered.

THE OPEN UNIVERSITYCoordinator · UK
USHAHIDI NETWORKparticipant · KE
THE UNIVERSITY OF SHEFFIELDparticipant · UK
UNIVERSITETET I AGDERparticipant · NO
TECHNISCHE UNIVERSITEIT DELFTparticipant · NL
GOVERNMENT TO YOUparticipant · BE

How to reach the team

The coordinator is The Open University (UK). SciTransfer can facilitate a warm introduction to the research team.

Order a report on this consortium See THE OPEN UNIVERSITY profile →

Next steps

Talk to the team behind this work.

Want to explore how COMRADES' crisis intelligence AI could work for your organization? SciTransfer connects businesses with EU research teams — contact us for a briefing.

Order a report on this topic More in Digital & ICT