Privacy-Preserving Data Re-publishing (PPDR) deals with publishing microdata in dynamic scenarios. Due to privacy concerns, data must be disguised before being published. Research in privacy-preserving data publishing (PPDP) has proposed many such methods on static data. In PPDR, multiple appeared records can be used to infer private information of other records. Therefore, inference channels exist among different releases. To understand the privacy property of data re-publishing, we need to analyze the impact of these inference channels. Previous studies show such analysis when data are updated or disguised in special ways, however, no general method has been proposed. Using the Maximum Entropy Modeling method, we have developed a general solution. Our method can conduct inference analysis when data are arbitrarily updated or arbitrarily disguised using either generalization or bucketization, two most common data disguise methods in PPDR. Through analysis and experiments, we demonstrate the advantage and the effectiveness of our method.

Privacy-Preserving Data Publishing (PPDP) deals with the publication of microdata while preserving people' private information in the data. To measure how much private information can be preserved, privacy metrics is needed. An essential element for privacy metrics is the measure of how much adversaries can know about an individual's sensitive attributes (SA) if they know the individual's quasi-identifiers (QI), i.e., we need to measure P(SA|QI). Such a measure is hard to derive when adversaries' background knowledge has to be considered.We propose a systematic approach, Privacy-MaxEnt, to integrate background knowledge in privacy quantification. Our approach is based on the maximum entropy principle. We treat all the conditional probabilities P(SA|QI) as unknown variables; we treat the background knowledge as the constraints of these variables; in addition, we also formulate constraints from the published data. Our goal becomes finding a solution to those variables (the probabilities) that satisfy all these constraints. Although many solutions may exist, the most unbiased estimate of P(SA|QI) is the one that achieves the maximum entropy.

To provide students with hands-on exercises in computer security education, we have developed a laboratory environment (SEED) for computer security education. It is based on VMware, Minix, and linux, all of which are free for educational uses. Based on this environment, we have developed ten labs, covering a wide range of security principles. We have used these labs in our three courses in the last four years. This paper presents our SEED lab environment, SEED labs, and our evaluation results.

Recently K-anonymity has gained popularity as a privacy quantification against linking attacks, in which attackers try to identify a record with values of some identifying attributes. If attacks succeed, the identity of the record will be revealed and potential confidential information contained in other attributes of the record will be disclosed. K-anonymity counters this attack by requiring that each record must be indistinguishable from at least K-1 other records with respect to the identifying attributes. Randomization can also be used for protection against linking attacks. In this paper, we compare the performance of K-anonymization and randomization schemes under linking attacks. We present a new privacy definition that can be applied to both k-anonymization and randomization. We compare these two schemes in terms of both utility and risks of privacy disclosure, and we promote to use R-U confidentiality map for such comparisons. We also compare various randomization schemes.