Crimes: reportes oficiais vs. postagens no Twitter

• Gilvan O. dos Reis UFAM
• Eduardo F. Nakamura UFAM

Resumo

The Social Network Sites (SNS) has become a major source of communication. Through SNS, we can communicate with people who are in many part of the planet and by using the information contained in these sites, we can turn users into social sensors. Based on this, this project aims to compare crime reports collected through social sensors to official police reports. As a case study, we validate our model with real data from New York City. The main results obtained were the training of a classifier with precision of 86% and a classified database with 491.415 tweets. Even with this result, it was not possible to find a correlation between the official dataset and our dataset.