Mobile phone data are collected communication logs between human beings. There are two interesting aspects and applications of the data: finding social structures and mobility patterns. The data could not only offer insights about how people make friends with each other but also shed light on how people move around in cities. These questions could help potential applications in security control and smart city planning....[ Read more ]

Mobile phone data are collected communication logs between human beings. There are two interesting aspects and applications of the data: finding social structures and mobility patterns. The data could not only offer insights about how people make friends with each other but also shed light on how people move around in cities. These questions could help potential applications in security control and smart city planning.

One interesting problem in social networks is Social Network Inference problem. Given the original raw communication data, how to accurately infer the relevant social network from the raw data? Are there noisy actors to affect the legitimacy of the social network? We consider the noise removal process as an important issue in Social Network Inference process. In this work, the noise removal problem is formulated and studied. Effective noise removing techniques are proposed to tackle the problem.

Another important application of the data is about the whereabouts of human beings. However, the privacy issue is prohibiting the sharing and study of the data. Recent study shows that more than 50% of the population in the United States could be uniquely identified if a similar mobile phone data as ours is published even with an anonymization on the IDs. We formulate the top location attack and prove that it is an NP-Complete problem to prevent such attack. Then, we propose our novel privacy preserving technique to modify the original data with minimal distortion.