The Stack Archive

Singapore research group highlights difficulties in ‘anonymising’ location data

Mon 22 Sep 2014

A new report [PDF] by a Singapore-based research group has cast doubt upon current practices in anonymising location-based mobile data, suggesting that individuals can be easily re-identified from large mobile datasets, and that such detailed data poses ‘serious privacy risks’.

The dataset used by the group, which is partially funded by the National Research Foundation of Singapore and the Economic Development Board, contains a filtered and optimised set of 0.63mn users, whose path in space and time is interpreted as a ‘trajectory’. The group found that traditional anonymising methods, including the random substitution of a location within the trajectory, are unable to provide sufficient anonymity.

The group’s proposed alternative is to cut up identifying trajectories into sub-trajectories, which preserves the analytical value of the data without impacting upon the privacy of individuals.


Controversies regarding the anonymisation of data continue to make news headlines, particularly where privacy issues are most sensitive, such as the the sale of the health records of nearly 50mn hospital patients as reported in February of this year. The sale occurred less than a week after a six-month moratorium was announced by the government on the proposed sharing of doctors’ medical records.

In 2006 a woman was publicly associated with her internet search history with information derived from supposedly anonymised data from AOL.

A 2013 MIT study posited that it is possible to identify individuals in an anonymised mobile dataset with 95% accuracy from just four points of data on the trajectory.

The paper’s authors are Yi Song, Daniel Dahlmeier and Stephane Bressan from the National University of Singapore and also SAP Research & Innovation, the research arm of European enterprise software organisation SAP SE. The paper is entitled ‘Not So Unique in the Crowd: a Simple and Effective
Algorithm for Anonymizing Location Data’, as a response to the influencing work which it acknowledges, the 2013 paper ‘Unique in the Crowd: The privacy bounds of human mobility’ by Yves-Alexandre de Montjoye, César A. Hidalgo, Michel Verleysen and Vincent D. Blondel.


Big Data news privacy security
Send us a correction about this article Send us a news tip