Detection of Barriers to Mobility in the Smart City Using Twitter

We present a system that analyzes data extracted from the microbloging site Twitter to detect the occurrence of events and obstacles that can affect pedestrian mobility, with a special focus on people with impaired mobility. First, the system extracts tweets that match certain predefined terms. Then, it obtains location information from them by using the location provided by Twitter when available, as well as searching the text of the tweet for locations. Finally, it applies natural language processing techniques to confirm that an actual event that affects mobility is reported and extract its properties (which urban element is affected and how). We also present some empirical results that validate the feasibility of our approach.