Crowdsourcing Incident Information for Disaster Response Using Twitter

Social media data have the potential to be used as a source of valuable information for real-time traffic operations to supplement existing systems such as 511. This paper analyzed incident data from two different data sources: 1) A traditional data provider that collects incident reports from multiple agencies, and 2) User posts from Twitter during Hurricane Sandy that flooded many areas in New York metropolitan area in 2012. A text classifier, built by utilizing extracted keywords from actual incident reports, is trained to find incident related Twitter data. The keywords are identified by Term Frequency–Inverse Document Frequency (TF-IDF) and naïve Bayesian method. The filtered Twitter data are cleaned and classified into various incident types to be compared geographically with that from the traditional data provider. The result showed that Twitter could provide detailed location information of a specific incident along with its intensity, duration and. Furthermore, it also provides information about incidents such as gas shortage that may not be easy to be obtained by traditional detection systems. It is not recommended to use Twitter as the only data source since it is biased and can be misleading depending on the type of analysis, yet it can be very powerful as a complementary data source. It is not only a real-time and inexpensive data provider but it also has a wide geographical coverage. It is worth to mention that Twitter data also contains incidents that are not available in TRANSCOM data set such as long lines at gas stations, crowdsourced traffic and closure conditions, but more accidents were reported by TRANSCOM. Therefore, merging these two sources will be useful especially for building models predicting incidents and generating resiliency maps.

• Supplemental Notes:
  • This paper was sponsored by TRB committee ABJ50 Standing Committee on Information Systems and Technology.
• Corporate Authors:

  Transportation Research Board

  500 Fifth Street, NW
  Washington, DC  United States  20001
• Authors:
  • Kurkcu, Abdullah
  • Zuo, Fan
  • Gao, Jingqin
  • Morgul, Ender Faruk
  • Ozbay, Kaan
• Conference:
  • Transportation Research Board 96th Annual Meeting
  • Location: Washington DC, United States
  • Date: 2017-1-8 to 2017-1-12
• Date: 2017

Language

• English

Media Info

• Media Type: Digital/other
• Features: Figures; References; Tables;
• Pagination: 17p
• Monograph Title: TRB 96th Annual Meeting Compendium of Papers

Subject/Index Terms

• TRT Terms: Classification; Crowdsourcing; Machine learning; Real time information; Social media; Traffic data; Traffic incidents
• Candidate Terms: Geospatial analysis
• Identifier Terms: Twitter
• Subject Areas: Data and Information Technology; Highways; Operations and Traffic Management;

Filing Info

• Accession Number: 01627649
• Record Type: Publication
• Report/Paper Numbers: 17-04082
• Files: TRIS, TRB, ATRI
• Created Date: Feb 27 2017 5:12PM