Thought off-line sanitization methods for bank transactions

Isaias Hoyos, Miguel Nunez-del-Prado

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In the digital era, people generate a lot of digital traces ranging from posts on social networks, call detail records and credit or debit banks transactions among others. These data could help society to understand different urban phenomena such as what citizens are talking about, how they commute or what are their spending behaviors. Therefore, the use of such data trigger privacy issues. In the present effort, we study four different Statistical Disclosure Control filters to sanitize off-line credit or debit bank transactions. Consequently, we analyze Noise Addition, Microaggregation, Rank Swapping and Differential Privacy filters concerning Disclosure Risk, Information Loss, and utility. We observed that Microaggregation and Different Privacy perform very well for minimizing Disclosure Risk while providing a good utility for statistics of spending amounts per industry type.
Original languageEnglish
Title of host publicationInformation Management and Big Data - 5th International Conference, SIMBig 2018, Proceedings
EditorsHugo Alatrista-Salas, Denisse Muñante, Juan Antonio Lossio-Ventura
Place of PublicationCham
Pages257-264
Number of pages8
ISBN (Electronic)9783030116798
DOIs
StatePublished - 1 Jan 2019
EventCommunications in Computer and Information Science -
Duration: 1 Jan 2019 → …

Publication series

NameCommunications in Computer and Information Science
Volume898
ISSN (Print)1865-0929

Conference

ConferenceCommunications in Computer and Information Science
Period1/01/19 → …

Bibliographical note

Publisher Copyright:
© 2019, Springer Nature Switzerland AG.

Keywords

  • Differential Privacy
  • Microaggregation
  • Privacy filters
  • Statistical Disclosure Control (SDC)

Fingerprint

Dive into the research topics of 'Thought off-line sanitization methods for bank transactions'. Together they form a unique fingerprint.

Cite this