Florimond Houssiau, Researcher at Imperial College London - Using Data While Protecting Privacy in the Digital Era
1.
2. Computational Privacy
Using Data while Protecting Privacy in the Digital Era
Florimond Houssiau
Imperial College London
PhD student, Algorithmic Society Lab
7. Anonymization: Safely using data for
statistical purposes
Name
Katerine Enter
Luella Perret
Dong Rice
Carroll Stiner
Ken Alamo
Yulanda Parikh
Janee Lundell
...
Income
[$/an]
100.000
35.678
45.000
325.000
125.000
23.459
75.008
G
F
F
M
M
M
F
F
DOB
01/1936
04/1960
12/1982
03/1970
05/1969
11/1997
09/1995
(Disaggregated by age
and gender)
8. Step 1: remove direct identifiers
(pseudonymization)
Income
[$/an]
100.000
35.678
45.000
325.000
125.000
23.459
75.008
G
F
F
M
M
M
F
F
Name
vF0m6JGQ
p0nYRG91
LgRLdjaA
uH4sUWLU
zfyv9PRY
qbu8Us1P
SrQ4sonIn
...
DOB
01/1936
04/1960
12/1982
03/1970
05/1969
11/1997
09/1995
(Disaggregated by age
and gender)
9. Step 2: blur indirect identifiers
(de-identification)
Income
[$/an]
100.000
35.678
45.000
325.000
125.000
23.459
75.008
G
F
F
M
M
M
F
F
Name
vF0m6JGQ
p0nYRG91
LgRLdjaA
uH4sUWLU
zfyv9PRY
qbu8Us1P
SrQ4sonIn
...
ICO - “Data protection law does not apply to data rendered anonymous”
(here 2-anonymous). No consent. No purpose.
DOB
80
60
30
50
50
20
20
(Disaggregated by age
and gender)
15. Unicity of mobile phone data
European country
1.5 M people
15 months
Points:
antenna and hour
de Montjoye, Y. A., Hidalgo, C. A., Verleysen, M., & Blondel, V.
D. (2013). Unique in the Crowd: The privacy bounds of human
mobility. Nature SRep, 3.
21. Predicting personality from
metadata
de Montjoye, Y. A., Quoidbach, J., Robic, F., & Pentland, A. S. (2013).
Predicting personality using novel mobile phone-based metrics. In
Social Computing, Behavioral-Cultural Modeling and Prediction (pp.
48-55). Springer
22. Predicting gender from
large-scale metadata
Europeancountry
SouthAsian
country
Jahani, E., Sundsøy, P. R., Bjelland, J., Pentland, A., & de Montjoye, Y. A.
Improving Official Statistics in Emerging Markets using Machine Learning
and Mobile Phone Data, EPJ Data Science (2017)
Felbo, B., Sundsøy, P., Pentland, A.S., Lehmann, S. and de Montjoye, Y.A.,
Modeling the Temporal Nature of Human Behavior for Demographics
Prediction (2017) ECML/PKDD
8 channels:
number of unique
contacts, calls,
texts and the total
duration of calls
(in and out)
25. Finding an actual trade-of
Privacy
Utility
Keep the promise:
anonymous use of data
(“your data will not be linked back to
you”)
Change the means:
anonymization
privacy-through-security
26. OPAL: Bringing
the code to the
data
Developed by:
With support from:
Secured question-and-answer
system (API)
To be installed in Senegal and
Colombia by the end of 2018
All open-source software &
published research
28. Florimond Houssiau
Imperial College London
florimond@imperial.ac.uk
In collaboration with Yves-alexandre de Montjoye, Alex Sandy Pentland, Luc Rocher, Cesar Hidalgo, Vincent
Blondel, Latanya Sweeney, Cameron Kerry, Jake Kendall, Michel Verleysen, Erez Shmueli, Arek Stopczynski, Sune
Lehmann, Eaman Jahani, Emmanuel Letouzé, Ali Farzaneh Far, Axel Oehmichen, Thibaut Lienart, Arnaud Tournier,
Andrea Gadotti.