Paper Type

Complete Research Paper

Description

Social networks offer vast potential for marketing agencies, as members freely provide private information, for instance on their current situation, opinions, tastes, and feelings. The use of social networks to feed into crime platforms has been acknowledged to build a kind of a virtual neighborhood watch. Current attempts that tried to automatically connect news from social networks with crime platforms have concentrated on documentation of past events, but neglected the opportunity to use Twitter data as a decision support system to detect future crimes. In this work, we attempt to unleash the wisdom of crowds materialized in tweets from Twitter. This requires to look at Tweets that have been sent within a vicinity of each other. Based on the aggregated Tweets traffic we correlate them with crime types. Apparently, crimes such as disturbing the peace or homicide exhibit different Tweet patterns before the crime has been committed. We show that these tweet patterns can strengthen the explanation of criminal activity in urban areas. On top of that, we go beyond pure explanatory approaches and use predictive analytics to provide evidence that Twitter data can improve the prediction of crimes.

Share

COinS
 

INVESTIGATING CRIME-TO-TWITTER RELATIONSHIPS IN URBAN ENVIRONMENTS - FACILITATING A VIRTUAL NEIGHBORHOOD WATCH

Social networks offer vast potential for marketing agencies, as members freely provide private information, for instance on their current situation, opinions, tastes, and feelings. The use of social networks to feed into crime platforms has been acknowledged to build a kind of a virtual neighborhood watch. Current attempts that tried to automatically connect news from social networks with crime platforms have concentrated on documentation of past events, but neglected the opportunity to use Twitter data as a decision support system to detect future crimes. In this work, we attempt to unleash the wisdom of crowds materialized in tweets from Twitter. This requires to look at Tweets that have been sent within a vicinity of each other. Based on the aggregated Tweets traffic we correlate them with crime types. Apparently, crimes such as disturbing the peace or homicide exhibit different Tweet patterns before the crime has been committed. We show that these tweet patterns can strengthen the explanation of criminal activity in urban areas. On top of that, we go beyond pure explanatory approaches and use predictive analytics to provide evidence that Twitter data can improve the prediction of crimes.