Data analysis through social media according to the classified crime

Abstract

The amount and variety of data generated through social media sites has increased along with the widespreaduse of social media sites. In addition, the data production rate has increased in the same way. The inclusion of personalinformation within these data makes it important to process the data and reach meaningful information within it. Thisprocess can be called intelligence and this meaningful information may be for commercial, academic, or security purposes.An example application is developed in this study for intelligence on Twitter. Crimes in Turkey are classified accordingto Turkish Statistical Institute criminal data and keywords are defined according to this data. A total of 150,000 tweetdata in the Turkish language are collected from Twitter between specified dates and processed by Turkish Zembereknatural language processing. It is seen that 56% of the people are talking about terrorist attacks and bombing attackson the study dates. The words “bomb,” “terror,” “attack,” “organization”, and “explode” have percentages of 24%,12%, 8%, 6%, and 6%, respectively. Moreover, associations between words and situations are found. Correlations areimportant to create new subclusters like “terror” and “rape” in this study with 0.90 correlation. Bigger masses can beaccessible by expanding keyword groups to have a clear picture of the real situation.