Browsing by Author "Karimi Adl, Rosa"

Now showing 1 - 3 of 3
  • Thumbnail Image
    ItemOpen Access
    A Negotiation Game: Establishing Stable Privacy Policies for Aggregate Reasoning
    (2012-10-31T14:47:10Z) Barker, Ken; Denzinger, Joerg; Karimi Adl, Rosa
    The process of personal information collection and exchange is associated with ever-growing privacy concerns. To resolve the issue, data provider's consent on the usage of private information is sought through privacy policy speci cations. The parameters of such privacy policies in uence the quantity and quality of gathered information. Choosing the right privacy policy parameters can potentially increase the revenues to a data collector and the rms (third-parties) interested in accessing the database for data analysis purposes. In this work we use an extensive form game model to examine the decisions made by a data collector and a third-party to maximize their bene ts from collecting and accessing data. We have found the game's subgame perfect equilibria for various problem settings and provide the details of game analysis for a simpli ed scenario and two case studies. The equilibrium solutions demonstrate steady states of the game where collecting personal information at a speci c privacy level is advantageous to the data collector and the third-party. Consequently the results de ne a realistic boundary on collecting personal information.
  • Thumbnail Image
    ItemOpen Access
    Privacy Consensus in Anonymization Systems Via Game Theory
    (2012-03-01T18:26:47Z) Karimi Adl, Rosa; Askari, Mina; Barker, Ken; Safavi-Naini, Reihaneh
    Privacy protection appears as a fundamental concern when personal data is collected, stored, and published. Several anonymization methods have been proposed to protect individuals' privacy before data publishing. Each anonymization method has at least one parameter to adjust the level of privacy protection. Choosing a desirable level of privacy protection is a crucial decision because it affects the volume and usability of collected data differently. In this paper, we demonstrate how to use game theory to model different and conflicting needs of parties involved in making such decision. We describe a general approach to solve such games and elaborate the procedure using k-anonymity as a sample anonymization method. Our model provides a generic framework to find stable values for privacy parameters within each anonymization method, to recognize the characteristics of each anonymization method, and to compare different anonymization methods to distinguish the settings that make one method more appealing than the others.
  • Thumbnail Image
    ItemOpen Access
    Stable Privacy Parameter Settings Using Game Theory
    (2013-04-16) Karimi Adl, Rosa; Barker, Ken
    Privacy protection appears as a fundamental concern when personal data is collected, stored, and published. Several privacy protection methods have been proposed to address privacy issues in private datasets. Each method has at least one parameter to adjust the guaranteed level of privacy protection. As the privacy protection level increases, the dataset loses more information utility due to further application of data manipulation methods and/or access restriction rules. Consequently, balancing the trade ff between privacy and utility is a crucial step and so far no systematic mechanism exists to provide directions on how to establish values for privacy parameters such that a balanced privacy/utility tradeff is induced. A balanced privacy/utility tradeoff can be described as a level on which the stakeholders of data reach a consensus (in the sense that no single party would be wiling to act diff erently to change the agreed upon level). Game theory provides a natural solution to finding such balanced tradeoff s. In this thesis, we capture the essence of establishing balancing values for privacy parameters as an extensive-form game with incomplete and imperfect information. A high-level step-by-step guideline is provided on how to solve the generic game. We instantiate the generic game model for three different privacy protection methods and analytically solve each game. The games' solutions are further simulated for sample problem settings to study the effects of various problem parameters on the balancing values of privacy parameters. The game model and its solution contribute to the fulfillment of our objective of establishing balancing values for privacy parameters (of a chosen privacy protection method). In addition to our main objective, the proposed game model can be consulted to choose the most pro fitable privacy protection method based on the problem requirements. Benchmarking frameworks can also benefi t from our game solutions by using the balancing privacy parameter values as the reference points for the comparisons between different privacy protection methods. We believe that a first step towards improving the data collection and privacy protection procedures is to understand how much privacy is currently sacrificed to achieve information utility (at the steady states). The game-based solution provided in this thesis promotes a deeper understanding of how privacy and utility reach a balanced tradeoff within the current privacy protection methods.