Inside the Anatomy of a Modern Crisis
The Sri Lankan government declared a state of emergency last week as mosques were torched and Muslim-owned businesses destroyed in attacks by Sinhalese mobs that left at least three people dead. A curfew was imposed in the Kandy District, the center of the of the violence, which was lifted late last week. A ban on some social media sites remains in effect as does increased police and military presence in the district.
The GeoSpark Analytics Activity GRiD, an artificial intelligence component of the BlueGlass risk and threat assessment platform alerted users to the increased levels of activity and the decrease in social media in the region. We produced a SparkCast titled Sri Lanka: State of Emergency in the Tourist Destination of Kandy on this event. This Blog is a quick peak inside how BlueGlass can dissect events like this and provide deeper insight enabling you to take action.
The graphic to the left shows a time lapse view of the Activity GRiD in action over Sri Lanka from 1–9 March. The yellow and red grids indicate activity deviations from the norm.
Every 15 minutes the model continuously assesses activity levels, defines normal patterns and identifies anomalous activity alerting users to actions that may pose a threat or risk to their operations, their people or their investments. The GRiD instantaneously looks at news and social media in the location of anomaly, analyzes trending topics in the media and provide users with an estimate of what caused the anomaly. Alerts to heightened activity generate email warnings.
BlueGlass’ social media integration enables a deep understanding of both geotagged and non-geotagged social media from Twitter, Facebook, Instagram, VK, Sino Weibo and other popular social media outlets. The graphic to the left illustrates a marked decline in geotagged social media posts prior to and during the government imposed social media blackout in the Kandy District.
The marked decline in social media correlated with a steep increase in traditional press reporting of the crisis. The BlueGlass Activity GRiD alerted to both the unusual decrease in social media and increased press reporting.
In addition to illustrating the significant changes in social media and news reporting, the graphic illustrates a steep increase in the volume of mobile devices in Kandy and Digana (the two main towns where the violence occurred). These increases correlate with the increase in police and military presence in the region.
BlueGlass incorporates Natural Language Generation algorithms that instantaneously generate written reports of structured content. The reports not only describe the trends and patterns in the data but look for correlation of significance. This graphic illustrates a small portion of the report generated from the news reporting in the Kandy District from the first ten days in March.
Machine Driven Human Finished Reporting
This brief example illustrates a number of the tools and content that we deliver to every user of the BlueGlass platform. These capabilities enable our users with machine driven content that they incorporate into human finished reports.
If you would like to learn more about how GeoSpark Analytics is developing machine learning models and artificial intelligence in the BlueGlass platform that can help your operations identify risk, threat and opportunities please contact us at email@example.com.