Location based information systems have shown a sustained growth trend in the last decade, thanks to the appearance of smartphones, capable of locating a user in real time and with great accuracy. Current users of mobile apps are normally sharing their location which is being stored by the service providers, either for internal use, to be sold later to third party companies, or to be released for public use as open data. The last two uses require the data to be stripped of information of individuals so it cannot allow identification; however, if the location is not altered sufficiently, it still could be used to find a single person. In the literature there are many location privacy protection mechanisms - LPPMs, usually expected to be applied on real time over the location information. This work introduces VoKA, a, offline privacy protection technique that focuses at the moment in which the data is released outside a protected environment, proposing a new aggregation mechanism based on Voronoi diagrams and K anonymity that preserves geographical value of the data like density and geographical distribution. The study uses Twitter data from the Lombardia area of Italy, and compares unaltered data, obfuscated data, grid-based aggregation and VoKA. Results show that VoKA can aggregate data in a more organic manner than grid-based data, and reduces the possibility to identify individuals compared to simple location obfuscation, in geostatistical analysis techniques like entropy and heatmap.