The representativeness and spatial bias of volunteered geographic information: a review

Many applications of volunteered geographic information (VGI) involve inferring the properties of the underlying population from a sample consisting of VGI observations, i.e. VGI sample. The representativeness of VGI sample is crucial for deciding the fitness for use of VGI in such applications. Due to the volunteers' opportunistic observation efforts, spatial distribution of VGI observations is often biased (i.e. spatial bias). This degrades the representativeness of VGI and impedes the quality of inference made from VGI. Extensive research has been conducted on assessing or assuring VGI quality from the perspective of the fundamental dimensions of spatial data quality. Yet, this perspective alone provides limited insights on the representativeness of VGI. Assessing VGI representativeness and developing novel approaches to accounting for spatial bias in VGI is in need for broadening the spectrum of VGI applications. This article offers a comprehensive survey of the scientific literature from various domains (ecology, statistics, machine learning, etc.) to summarize existing endeavors related to sample representativeness assessment and sample selection bias correction for enlightening the treatment of these issues in VGI applications.