Socio-Spatial Intelligence: social media and spatial cognition for territorial behavioral analysis

Area of Science:

• Investigative analysis
• Spatial cognition
• Social media analytics

Background:

• Investigative analysts utilize diverse data sources, including social media (SM), to understand criminal spatial behavior.
• Previous research demonstrated the utility of tweets for estimating private points of interest.

Purpose of the Study:

• To extend previous findings by analyzing a larger dataset, demonstrating the application of social maps and web applications in investigative analysis and spatial cognition research.
• To test the hypothesis that user areas of activity and residence are predictable using SM data.

Main Methods:

• Analysis of 100 Twitter accounts (approx. 250 tweets each) using geographical techniques (Convex-Hull, Mean-Center, Median-Center, Standard-Deviation-Ellipse).
• Validation of predictions using explicit references to activity areas and user residences within the tweets.
• Development of simple algorithms to predict user home locations.

Main Results:

• Predictive algorithms successfully indicated home locations in approximately 80% of cases.
• All identified home positions were within the Convex-Hull, and most within the Standard-Deviation-Ellipse.
• Up to 80% of residences were located within a 1.500 m buffer zone using Median-Center, with a minimum effectiveness threshold of 12-13 tweets.

Conclusions:

• Social media analysis, combined with geographical data, effectively predicts user mobility, spatial cognition, and home locations.
• Geographical Information System (GIS) tools and SM analysis provide valuable spatial and territorial information for understanding social dynamics.
• These findings support applications in criminal behavior prevention and the modeling of specific human behaviors.