An emerging paradigm of volunteered geographic information (VGI) in the field of Geography was resulted from recent advances of GPS equipped mobile devices such as the iPhone, and the amount of geographic information and its promptness has been dramatically increased with the participation of volunteers. Although there are many fields that can utilize the capabilities of VGI, conventional quality control processes are often not applicable to VGI. Therefore, it is essential to balance the prompt availability of VGI with the unverified character of its contents. For this purpose, a credibility model approach measuring the credibility of VGI is presented in this thesis to promote the democratization of geographic information from a traditional top-down process, to a revolutionary bottom-up process. For a plausible inference of credibility, a computational credibility model was constructed based on a Bayesian network (BN), providing a convenient way to tackle diverse problems in which conclusions are not warranted logically but rather predicted probabilistically. This research also has led to the development of an integrated framework to acquire, infer, store, and visualize detailed credibility information. Using a case study designed to simulate the process of credibility assessments of the volunteered geographic data from the Haiti earthquake, this thesis examines the behavioral characteristics and the benefits of the developed credibility model. In a set of experiments, damage assessments with and without the credibility assessments are conducted and compared with an official assessments by the Haitian government and international organizations. The experiments have not only demonstrated microscopic effects on the individual data, but also showed the macroscopic variation of the overall damage patterns by the credibility model. The case study and experiments serve as a proof-of-concept to validate the benefits of the credibility assessment framework. This credibility model approach is implemented using an open-source spatial database system and application programing interface (API), and can be further developed to be used as an API in social network services to better support collaborative workflow tasks.
【 预 览 】
附件列表
Files
Size
Format
View
Credibility assessment of volunteered geographic information for emergency management: a Bayesian network modeling approach