Methods Comparison
Here we present the results obtained for each classification app-roach utilized (i.e. Logistic Regression as LR, Dictionary appro-ach as Terms). Table 3 depicts the number of users per attribute who have been classified as expressing negative attitude towards law enforcement and authorities. In line with the flat data thres-holds, and depending on the scenario applied, the security officer may decide which of the above mentioned attributes could be pro-cessed, so as to reveal a user attitude.