Identifying incidents of threat to public life from news articles can be used in various applications like traffic management, disaster management etc. In this paper we are identifying threat to public life by processing RSS feed of news articles and we investigate the tradeoff between speed Vs accuracy of predicting the severity of the incidents which is treat to public life. It can also be used to collect information on geographical locations where threat has occurred. We have proposed a NLP-based approach that would help identify features of the data that are salient for classifying it as road incident like accident or bomb blast etc and its severity. © Research India Publications.