Privacy-aware Scheduling
Project Description
Privacy is an important human right that is constantly challenged by modern information technology. Several regulatory frameworks, such as the GDRP from the European Union, compel organizations to protect the privacy of individuals. Many organizations have protocols and technologies in place to protect obviously sensitive data such as medical records. However, some business activities indirectly reveal private information, e.g., a typical example from everyday life is that many people might suspect that a woman might is pregnant if she is not drinking alcohol at a party. Understanding how organizations indirectly reveal private information is for upmost importance and should be further investigated.
In previous work, I was shown that private information can be revealed from published schedules as they are commonly used for example in healthcare settings. One common goal of privacy research is to quantify how much information is leaked in a specific setup. Previous work by Fahrenkrog-Petersen et al. quantified privacy loss for schedules under the assumption that a public schedule is optimal and correct. However, small variations of the schedule made it impossible to quantify the privacy loss, since the inference attack was based in inverse optimization and required an optimal schedule. Therefore, making a potential anonymization of the data trivial.
In this project, we aim to further improve the quantification of privacy losses of published schedules. For this purpose, we want to develop more realistic inference attacks that allow for a more accurate calculation of the privacy loss. Such results can be used to further develop novel privacy protection techniques and to generalize the existing privacy loss to a wider range of scheduling problems.
In previous work, I was shown that private information can be revealed from published schedules as they are commonly used for example in healthcare settings. One common goal of privacy research is to quantify how much information is leaked in a specific setup. Previous work by Fahrenkrog-Petersen et al. quantified privacy loss for schedules under the assumption that a public schedule is optimal and correct. However, small variations of the schedule made it impossible to quantify the privacy loss, since the inference attack was based in inverse optimization and required an optimal schedule. Therefore, making a potential anonymization of the data trivial.
In this project, we aim to further improve the quantification of privacy losses of published schedules. For this purpose, we want to develop more realistic inference attacks that allow for a more accurate calculation of the privacy loss. Such results can be used to further develop novel privacy protection techniques and to generalize the existing privacy loss to a wider range of scheduling problems.
Relevance to Liechtenstein
The research project holds relevance for Liechtenstein and the Alpine Rhine Valley region, particularly because of the potential of a publication in a highly ranked journal, such a publication would support the strategic positioning of the Liechtenstein as a high-tech region. Further, Liechtenstein has strict privacy laws itself and understanding how modern technology impact the privacy of individuals is of importance for the country. If the project leads to an additional SNSF grant this would further strengthens Liechtenstein`s visibility and positioning as a high-tech region and also strengthen the financial basis of the University of Liechtenstein.
