Wisdom of the Local Crowd: Detecting Local Events Using Social Media Data

Event attendees post about their experiences on social media. We propose a novel approach for analyzing these posts to extract ongoing events. We gather posts from Twitter and Instagram and perform a number of processing steps to identify posts related to events based on hashtags and location information. Our approach detects events only using posts submitted during the past hour which ensures that only ongoing events are detected. The system can detect both large and small events with a high location accuracy, a precision of 0.20, and a recall of 0.60.