A Study of the Correlation between the Spatial Attributes on Twitter

In recent years, much attention has been given to the topics on social network analysis. Previous research has shown that each Twitter user can work as a sensor to detect the target events such as the earthquakes. They used the spatial attributes on Twitter to estimate the location of the target event. Although the precision of the location information may affect the performance of the estimation, they did not consider its reliability. In this paper, we investigate the correlation between the profile locations on Twitter and the GPS coordinates in tweets. A text-based grouping method is applied to the spatial attributes to cluster them with considering of the administrative districts. The analysis result shows that nearly 50% of users post the most of their tweets in the profile locations while 30% of users, who may have high mobility in a wide range, do not have any tweets in their locations.