Volunteered Geographic Information (VGI) has been widely adopted to assist in disaster management, yet its characteristics of uncertainty and requirements of large amounts of manual manipulation for data validation and interpretation hinder VGI applications. In this study, we aimed to develop an effective method to assess the credibility of VGI for time-critical conditions, such as disaster response. We collected datasets from two extreme flood events in 2011 and 2013 from Brisbane, Australia. According to the defined geo-location factors, we built a binary logistic regression with the 2011 event dataset to measure the credibility scores of the VGI instances. At the threshold of 0.917, the overall accuracy of the model in the 2011 training dataset was 90.5%. Next, the performance of this probability model was evaluated by the 2013 testing instances. We found that our model could categorize the credibility classes with 80.4% accuracy. These results suggest great potential for our model to be used by emergency management sectors to sort credibility of VGI for efficient and rapid response, decision-making, and coordination.
Financed by the National Centre for Research and Development under grant No. SP/I/1/77065/10 by the strategic scientific research and experimental development program:
SYNAT - “Interdisciplinary System for Interactive Scientific and Scientific-Technical Information”.