USE OF TRAFFIC AND CITIZEN TWEETS FOR INCIDENT MANAGEMENT FOR DISTRICT DEPARTMENT OF TRANSPORTATION

摘要

Extracting and analyzing social media for incident management and incident collection is a relatively new concept in transportation field. In this research we will study the transportation data collected from social network Twitter regarding in cidents happening in Washington DC. Twitter is a micro blogging platform that contains a large amount of publically accessible user generated content, which contains plenty of information that is related to current transportation events. In this research at District Department of Transportation (DDOT), we developed an algorithm to carefully retrieve tweets from the twitter stream. New algorithms were developed to extract right geographically based key words, rank them based on the relevance to get the traffic information, developed a schema for storing the unstructured data in efficient way. Each tweet is ranked and scored by algorithm. Based on the keywords used, number of keywords, recurrence of keywords and score of tweet they are classified as traffic tweet, based on the patterns extracted from the influential users. An intuitive user interface is built to show the latest traffic tweets along with visualizations tools like histogram. Additionally we developed a search interface for operators to look for particular incidents/delays on a particular day and time. This helps our operators to fill out any missing information which can be supplied by the social media resources to the incident reports.