What comes first: 112 call or first tweet? Which social media content help emergency respondents and which content instead biases the response?
The power of collectively generated data was demonstrated during the large scale crisis events such as Haiti earthquake, the tsunami in Japan or the Arab Springs revolts. Many agencies and organizations are currently trying to include social media into their crisis management practices. However, it is still unclear what is the quality and reliability of social media information in an emergency context, as well as which type of events and which phases of emergency management can benefit from such information.
In this project (2014) we have explored the potential of social media data to complement existing information sources of incident responders in the Netherlands. In particular, we explored and compared data streams from Twitter, Flickr and Foursquare, addressing questions such as:
What are the features of social media as information source for Incident management and crisis management?
What is the information content of each data source in terms of data volume and data structure?
What are the space-time attributes of the data feed?
What are the features of the information feed during normal and special events?
What evidence can we collect from the analysis of incidents in the Netherlands?
We tested the applicability of social media for several events, ranging from highway and railway accidents, country-wide weather anomalies and large-scale social events. The animations below present the response of social media community for two examples of emergency situations.
On August 7, 2012 a severe car crash took place on the A12 highway near Arnhem, Netherlands. It involved 7 cars and engaged the emergency services for the following six hours. The reaction of the Twitter community was immediate, providing observations on the emergency circumstances, as well as the development of traffic conditions. The data generated by social media activity offered a broad proxy for traffic flow estimation, based on the content of the messages, as well as the activity of users. The increased social media activity was visible not only at the local scale, but also at the national level, where the accident was commented until the following day.
On February 3 the Netherlands experienced unusually heavy snowfall. It caused over 800 km of traffic jams and serious disruption to train services. Many flights from Schiphol were delayed or cancelled. The unusual snowfall event was broadly commented on Twitter creating significant positive anomalies of the usual social media activity. Analysis of tweets’ semantics enabled to determine which parts of the country were affected by heavy weather conditions, also in terms of problems with traffic jams and trains
Distinct pick of increased activity concerned also Flickr. Relatively fast upload of pictures related to snowfall determined quick availability of data that could serve as visual indication of the situation in different parts of the country.
Fro more information: email@example.com