In this work, we have made several contributions. We
demonstrated how machine learning can be used to detect
small scale incidents with high precision. In this case, we
contribute the first classifier to detect three different types of
small scale incidents with 82.2% precision and 82% recall.
Furthermore, we examined the user behavior during small
scale incidents and showed that a variety of individual users
share small incident related information and that those users
are reporting faster than official sources.
For future work, it might be interesting to do a qualitative
analysis of the content shared by citizens. E.g., a differentiation
which information really can contribute to increasing
situational awareness is necessary. Furthermore, the analysis
should be extended throughout different cities.