CRIME PREDICTION USING SOCIAL SENTIMENT AND SOCIO-FACTOR

Authors : Sakirin TAM, Ö. Özgür TANRIÖVER

Pages : 11-20

View : 7 | Download : 18

Publication Date : 2018-07-31

Article Type : Research Paper

Abstract :Crime prediction becomes very important trend and a key technique in crime analysis to identify the optimal patrol strategy for police department. Many researchers have found number of techniques and solutions to analyze crime, using data mining techniques. These studies can help to speed up and computerize the process of crime analysis processes.  However, the pattern of crime is flexible, it always changes and grows. With social media, user posts and discusses event publicly. These textual data of every user has contextual information of user’s daily activities. These posts generate unstructured data that can be used for data prediction. As shown by previous research, twitter sentiment enable to predict crime in Chicago, United States. However, existed model on crime prediction was incorporating the use of socio factors. Therefore, the study aims to model crime prediction using social media content with additional socio-factors. The research approach is consisted of a combination of sentiment analysis from Twitter and social-factors with Kernel Density Estimation. Lexicon-base methods will be applied for sentiment analysis, and the model evaluation is measured with the help of logistic regression. 
Keywords : Crime prediction, Data mining, Social media, Sentiment analysis, Socio Factor