My answer on Quora
An interesting case study involving Big Data – Quantifying Information Flow During Emergencies.
Source:[1401.1274] Quantifying Information Flow During EmergenciesAbstract:
Liang Gao @ Beijing Jiaotong University in China and a few other reseachers created a studied mobile phone usage and human behavior during extraordinary conditions, such as during earthquakes, armed conflicts or terrorist incidents. The brainchild of the study was that much of big data is based on events that are consistent, familiar and slow moving.Metadata:
The metadata that the researchers used were voice calls and text messages from over 10 million people during a 4 year period in an unknown European country. The metadata included the caller and receiver, a timestamp and the location of the tower that routed the call (see probe requests @ http://frivolousutility.com/2013…). The study consisted of two distinct groups: the affected group (first group), the group that received the first call after the event [i.e. friends and relatives] (second group).Google News was used to identify three extraordinary events that were referred to as “Jet Scare”, “Plane Crash” and “Bombing” that were near the the time and location of the voice calls/text messages. The control event (consistent, familiar and slow moving) is referred to as “Concert”. The team wanted to see how communications during extraordinary events differed from a controlled event.Findings:
During an extraordinary event, phone usage spikes in the first group and subsequently spikes in the second group. As with most networks, the normal line of thinking is that information will spread. However, information did not spread from the second group to other groups. Instead, the second group called the affected group back! The second group seemed to want to know more about the emergency. Compare this to the Concert where the second group more times than not did not call the first group back if a call was dropped.Conclusion:
The study concluded that “information flow through the social networks of affected individuals becomes an important means to spread situational awareness and information to the general population.” In other words, in emergency situations, the best source of information are those closest to the affected group. However, the need for more information within these tight networks is greater than the need to spread the news to others groups. This may give authorities new ways of using big data for emergency management.
Source:[1401.1274] Quantifying Information Flow During EmergenciesAbstract:
Liang Gao @ Beijing Jiaotong University in China and a few other reseachers created a studied mobile phone usage and human behavior during extraordinary conditions, such as during earthquakes, armed conflicts or terrorist incidents. The brainchild of the study was that much of big data is based on events that are consistent, familiar and slow moving.Metadata:
The metadata that the researchers used were voice calls and text messages from over 10 million people during a 4 year period in an unknown European country. The metadata included the caller and receiver, a timestamp and the location of the tower that routed the call (see probe requests @ http://frivolousutility.com/2013…). The study consisted of two distinct groups: the affected group (first group), the group that received the first call after the event [i.e. friends and relatives] (second group).Google News was used to identify three extraordinary events that were referred to as “Jet Scare”, “Plane Crash” and “Bombing” that were near the the time and location of the voice calls/text messages. The control event (consistent, familiar and slow moving) is referred to as “Concert”. The team wanted to see how communications during extraordinary events differed from a controlled event.Findings:
During an extraordinary event, phone usage spikes in the first group and subsequently spikes in the second group. As with most networks, the normal line of thinking is that information will spread. However, information did not spread from the second group to other groups. Instead, the second group called the affected group back! The second group seemed to want to know more about the emergency. Compare this to the Concert where the second group more times than not did not call the first group back if a call was dropped.Conclusion:
The study concluded that “information flow through the social networks of affected individuals becomes an important means to spread situational awareness and information to the general population.” In other words, in emergency situations, the best source of information are those closest to the affected group. However, the need for more information within these tight networks is greater than the need to spread the news to others groups. This may give authorities new ways of using big data for emergency management.