National Technical University of Athens (NTUA)
Προστασία της ιδιωτικότητας στην δημοσίευση μη σχεσιακών δεδομένων
Abstract
dc:descriptionThis doctoral thesis addresses issues of privacy preservation on the publications of non-relationaldatasets. The research focuses on proposing and implementing anonymization algorithms whichtransform the datasets to be published, so that the identity of each record and the personal informationof individuals are not revealed.A broad spectrum of human activities, such as medical examinations, credit card purchases, internetsearching, webpage browsing, etc. produce massive amounts of data. Their publication or sharing isposing a threat to the privacy of users; even after the elimination of direct identifiers (SSN, full name)the combinations of other attributes (such as gender, age, or zip code) may be rare enough to act asquasi-identifiers, thus revealing the identity of records. Anonymization methods aim at preventing theidentification of records by malicious attackers, so that the sensitive information concerning individualswill not be revealed. This can be achieved by transforming the data in order to ensure that a givenprivacy guarantee is not violated. This process causes an information loss of the released data as italters or conceals part of the quasi-identifier values. The goal of anonymization algorithms is to achievethe best balance between the privacy preservation of individuals and the utility of the released data.Several steps have been made in the related literature on the privacy of relational data. The main focusof this thesis is the anonymization of (a) tree-structured data, such as XML data, as well as joinedrecords from a relational database which includes many tables linked together via foreign keys, (b)graph-structured data such as linked and RDF data which are popular on the Web, and (c)unstructured sparse multidimensional data where every record is a bag of continuous numerical values,such as financial payment data or credit card purchases. Furthermore, alternative attack models werestudied, such as (i) the scenario where the aggregate knowledge of the numerical data values (e.g.sum, average) can lead to the identification of a personal record, and (ii) more complex attackscenarios which include the knowledge of multiple functions, each being defined on any subset of thenumerical attributes of a record.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2015
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Gkountouna, Olga
- Γκουντούνα, Όλγα
Subjects
dc:subject × 18- Ιδιωτικότητα;
- Ανωνυμία
- Προστασία ιδιωτικότητας
- Ανώνυμη δημοσίευση δεδομένων
- Μη-σχεσιακά δεδομένα
- Privacy
- Anonymity
- Privacy preservation
- Anonymous data publishing
- Non-relational data
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Natural Sciences
- Computer and Information Sciences
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/43013