Incidents are first detected through proprietary data gathering tools, scanning thousands of information sources. The system also has a built-in auto-translation capacity to manage sources in a variety of languages.
These are a few of the types of data sources used to gather information on incidents:
- News (International and Regional)
- Seismological institutes
- Meteorological agencies
- Departments of Foreign Affairs
- Other Government Agencies
- Social Networks and forums (Crowdsourcing)
- Aid organizations and other NGO’s
- Meta-data aggregation
- Health and Disease Surveyors
- Safeture Incident and Event Database (all events and incidents that Safeture process are tagged and labeled)
- Strategic Partners
- Social Media
After data gathering, the information first goes through automatic filtering by an Artificial Intelligence (AI) system (semantic relationships) plus a list of keywords to search for. A secondary manual filtering is done by analysts. They then verify the data, define the impact radius, severity level, and finally, publish the alert.
The impact radius is dynamic and varies depending on the incident. For instance, a shooting is seen as a very local incident while an earthquake might have countrywide effect. The radius of the incident is set based on the incident category and how widespread the effect is. They are set with an interval between 10 km and 1000 km. Event radius crosses borders and may be set as nationwide. Minor incidents, such as traffic disruptions, civil unrest, and events, start within a 10 km radius. Natural disasters span a large radius interval, with the most widespread tropical storms ranging upwards 1000 km.
If the incident is seen as critical, an initial message will be published and sent to clients in the affected location within 5-15 minutes from when a source initially reported about the incident. We are capable to do so due to the integrated information mining system, message publication system, and our highly automated process. Sometimes robot publication is used where the time between source input to message (SMS, push notification) publication and send-out is required to be short and where the source is highly credible, such as automated tsunami warnings.
On publication of an alert, the Safeture Platform will automatically send out an alert about the incident to all end-users that are within the impact radius, or that are traveling into the region where there has been a published alert during the last three days.
By “Favoriting” a region, end-users will receive newly published alerts for that country even if they are not traveling to, or within, that country. The alert will show as a notification, and when pressing the link or notification, the end-user is presented with an alert overview. A map will show the incident location, the risk level, and the distance from their current position to the incident location.
The Safeture Platform will also notify relevant Administrators in the Safeture Web Portal (own/assistance providers) if any employees were impacted by the alert. To Administrators, alert notifications are sent via email and can be configured to send via SMS also.