Detection and Labeling of Personal Identifiable Information in E-mails

Abstract

Part 8: Identifiability and Decision MakingInternational audienceThe protection of personal identifiable information (PII) is increasingly demanded by customers and data protection regulation. To safeguard PII a organization has to find out which incoming communication actually contains it. Only then PII can be labeled, tracked, and protected. E-mails are one of the main means of communication. They consist of unstructured data difficult to classify. We developed an automated detection system for PII in e-mails and connected it to a usage control infrastructure. Our concept is based on previous findings in the area of spam detection. We tested our approach with a data set in a customer service scenario. The evaluation shows that the utilization of Bayes-classification is very promising to detect PII

Make personal information security great again: A case of users’ perspectives on personal identifiable information in South Africa

“…Information elicited from these platforms (such as where they spend time off work or who they socialise with) can be aggregated to build a user profile, which then serves as a basis for a targeted phishing attack because, over time, the cyberattacker has painted a picture of the user. Personal identifiable information can then be used to craft a specific email deliberately designed to manipulate the user and make them think it is genuine Bier and Prior (2014). At times, the use of PII can be adverse when the cyberattack leads to identity theft or social engineering to exploit the user's emotions to gain information (Abawajy 2014).…”

Towards a Visual Privacy Advisor: Understanding and Predicting Privacy Risks in Images

“…Aura et al [3] explore this in the context of electronic documents, where they propose a tool to remove user names, identifiers, organization names and other private information from text-based documents with metadata. [5,13] study this in the context of textual email-content. Bier et al [5] model this as a privacy-classification problem, whereas Geng et al [13] detect four types of personal information -email addresses, telephone numbers, addresses and money.…”

“…[5,13] study this in the context of textual email-content. Bier et al [5] model this as a privacy-classification problem, whereas Geng et al [13] detect four types of personal information -email addresses, telephone numbers, addresses and money. The closest related work to ours is [42], who are also motivated by unwanted disclosure and privacy violation on social media.…”