Anonymizing transaction databases for publication

Abstract

This paper considers the problem of publishing "transaction data" for research purposes. Each transaction is an arbitrary set of items chosen from a large universe. Detailed transaction data provides an electronic image of one's life. This has two implications. One, transaction data are excellent candidates for data mining research. Two, use of transaction data would raise serious concerns over individual privacy. Therefore, before transaction data is released for data mining, it must be made anonymous so that data subjects cannot be re-identified. The challenge is that transaction data has no structure and can be extremely high dimensional. Traditional anonymization methods lose too much information on such data. To date, there has been no satisfactory privacy notion and solution proposed for anonymizing transaction data. This paper proposes one way to address this issue.

Personalized Privacy Preserving Publication of Transactional Datasets Using Concept Learning

“…Yabo Xu et al [6] model the power of attackers by the maximum size of public itemsets that may be acquired as prior knowledge, and proposed a novel privacy notion called "coherence" suitable for transactional databases.…”

Towards Identify Anonymization in Large Survey Rating Data

“…The structure of large survey rating data is different from relational data, since it does not have fixed personal identifiable attributes. The lack of a clear set of personal identifiable attributes makes the anonymisation challenging [23], [7]. In addition, survey rating data contains many attributes, each of which corresponds to the response to a survey question, but not all participants need to rate all issues (or answer all questions), which means a lot of cells in a data set are empty.…”

“…Privacy-preservation of transactional data has been acknowledged as an important problem in the data mining literature [3], [4], [21], [7], [23]. The privacy threats caused by publishing data mining results such as frequent item sets and association rules is addressed in [3], [4].…”

“…In contrast, our work addresses the privacy threats caused by publishing a large survey rating data. Recent work [7], [23] targets anonymisation of transaction data. Our work aims to prevent individual identity disclosure in a large survey rating data set.…”

ρ-uncertainty Anonymization by Partial Suppression

“…We choose to solve the set-valued data anonymization problem by partial suppression because global suppression tends to delete more items than necessary, and the removal of all occurrences of the same item not only changes the data distribution significantly but also makes mining association rules about the deleted items impossible. The problem of anonymization by suppression (global or partial) is very challenging [1,18], exactly because, (i) the number of possible inferences from a given dataset is exponential, and (ii) the size of the search space, i.e. the number of ways to suppress the data is also exponential to the number of data items.…”