Real-Time Detection of Traffic From Twitter Stream Analysis

摘要: Social networks have been recently employed as a source of information for event detection, with particular reference to road traffic congestion and car accidents. In this paper, we present real-time monitoring system detection from Twitter stream analysis. The fetches tweets according several search criteria; processes tweets, by applying text mining techniques; finally performs the classification tweets. aim is assign appropriate class label each tweet, related or not. was areas Italian network, allowing events almost in real time, often before online news web sites. We support vector machine model, achieved an accuracy value 95.75% solving binary problem (traffic versus nontraffic tweets). were also able discriminate if caused external not, multiclass obtaining 88.89%.