xTRoad: The Tweet Extraction Method for Profiling Road and Traffic Conditions

Twitter contains many natural language text elements that can be used for many purposes. One purpose is to produce the information about road and traffic conditions. The goal of this research is to propose a new extraction method for Indonesia Twitter text called xTRoad. By applying the xTRoad method, the national roads at Jakarta city could be identified by the other roadways nearby those which are connected in the particular region. The identification of the road and traffic conditions is completed in less than four minutes.
The obtained information has the attributes of time, date, day, road names, incidents, weather and traffic conditions. All elements can be used to form the profile of road and traffic conditions. The results  of this study showed that the obtained profiles have many dimensions such as the congestion conditions per thirty minutes, the types of barriers that occur on national roads, the trend of traffic jam in 24 hours, etc. The similarity level for comparison between the traffic conditions from the extracted text and video data has an accuracy of 62.8%.