Spatial Vision releases “Geo-Indistinguishability” plugin for QGIS

The plugin contains a simple tool to add noise to spatial data for the purposes of maintaining “geo-indistinguishability”, i.e. an individual’s true location shouldn’t be able to be determined by the location of the point, but it should be accurate enough to make a useful study of the data. This is done by computing a random number from a 2d laplacian distribution and migrating the origin point by this much. It’s an implementation of the algorithm implemented in [1], and inspired / based on the implementation in the location-guard browser extension.

The first version of the plugin that’s gone up only implements the algorithm as if each location was completely spatially independent (thus points in the same place will be moved in different directions), and a quick tool to draw the displacement lines from the original features to the new locations. E.g. in the screenshot below, the original points are yellow, the resulting points are blue, and the displacement line between each is easily visible.

Why is it important? If we’re dealing with sensitive data, for example survey respondents’ homes, we may want to give people an idea of where they are from (accurate to say the suburb), but not their actual locations. Consider for example data related to health, heritage, etc.

The plugin’s homepage is freely available and works in any version of QGIS from version 2.0 and above (latest QGIS version is 2.12).


Rate this item
(0 votes)
Login to post comments
Go to top