Using Twitter to Enhance Traffic Incident Awareness

Abstract

Automatic incident detection is an important component of intelligent transportation management systems that provides information for emergency traffic control and management purposes. Social media are rapidly emerging as a novel avenue for the contribution and dissemination of information that has immense value for increasing awareness of traffic incidents. In this paper, we endeavor to assess the potential of the use of harvested tweets for traffic incident awareness. A hybrid mechanism based on Latent Dirichlet Allocation (LDA) and document clustering is proposed to model incident-level semantic information, while spatial point pattern analysis is applied to explore the spatial patterns. A global Monte Carlo K-test indicates that the incident-topic tweets are significantly clustered at different scales up to 600m. Then a density-based algorithm successfully detects the clusters of tweets posted spatially close to traffic incidents. The experiments support the notion that social media feeds act as sensors, which allow enhancing awareness of traffic incidents and their potential disturbances.