Public Twitter data and transport network status

Abstract

Twitter data can be collected and analysed to be used for predicting the status of a transport network at a given time and geographic location (e.g. forecasting disruptions, congestions, or road closures). However, this requires geolocating the tweets to define the parts of the transport network which may be related to these tweets. This paper investigates the relationship between the actual transport network status, with that being synthesised using public Twitter data in the Greater Manchester conurbation. Therefore, it answers the following question: are the sentiments of tweets around the incidents and accidents areas (or bounding boxes) different from the sentiments of tweets in the seamless traffic areas?. According to the used research methodology, analysis techniques, and sentiment detection APIs, it has been concluded that there is no significant difference between the sentiments in the tweets regardless the prevailing traffic conditions of the locations the tweets refer to.