Logo des Repositoriums

Auflistung nach Schlagwort "anonymization"

1 - 2 von 2
  • Textdokument
    Anonymization Is Dead – Long Live Privacy
    (Open Identity Summit 2019, 2019) Zibuschka, Jan; Kurowski, Sebastian; Roßnagel, Heiko; Schunck, Christian H.; Zimmermann, Christian
    Privacy is a multi-faceted, interdisciplinary concept, with varying meaning to different people and disciplines. To most researchers, anonymity ist he “holy grail” of privacy research, as it suggests that it may be possible to avoid personal information altogether. However, time and time again, anonymization has been shown to be infeasible. Even de-facto anonymity is hardly achievable using state-of-the-art cryptographic anonymization techniques. Furthermore, as there are inherent tensions between the privacy protection goals of confidentiality, availability, integrity, transparency, intervenability and unlinkability, failed attempts to achieve full anonymization may make it impossible to provide data-subjects with transparency and intervenability. This is highly problematic as such mechanisms are required by regulation such as the General Data Protection Regulation (GDPR). Therefore, we argue for a paradigm shift away from anonymization towards transparency, accountability, and intervenability.
  • Konferenzbeitrag
    Pseudonymizing Log Entries with time-selective Disclosure
    (Workshops der INFORMATIK 2018 - Architekturen, Prozesse, Sicherheit und Nachhaltigkeit, 2018) Sonntag, Michael
    Centralized logging of entries containing personally-identifiable data, like IP addresses, is common. However, this chances that persons other than the operator of the individual server might obtain access to these logs and then disclose or use them. Additionally, the GDPR recommends as a security measure pseudonymization, i.e. splitting the information into two parts. This article describes a method to pseudonymize personal information in elements stored in a time series. After a predetermined time, the information can be automatically anonymized without requiring any changes in the stored entries themselves. Additionally, some statistical analyses remain possible, as the same values are encoded with the same pseudonym. It is also possible to disclose an arbitrary time period from within the log file: everything after the start time and before the end time can be de-pseudonymized, but the rest of the data remains anon-/pseudonymous.