RedactBuster: Entity Type Recognition from Redacted Documents

TL;DR

The paper tackles privacy risks in redacted documents by showing that sentence context can enable deanonymization of redacted entities. It introduces RedactBuster, a sentence-context-aware framework that uses fine-tuned transformer embeddings and supervised classifiers to predict anonymized entity types from redacted text, achieving up to $0.985$ accuracy on the TAB dataset. The study provides a baseline of $0.958$ accuracy without fine-tuning and demonstrates a strong performance uplift with fine-tuning, while also proposing a homoglyph-based countermeasure called character evasion. These findings underscore the need for more robust redaction strategies and offer a practical evaluation framework and open-source tools for assessing redaction resilience in real-world documents.

Abstract

The widespread exchange of digital documents in various domains has resulted in abundant private information being shared. This proliferation necessitates redaction techniques to protect sensitive content and user privacy. While numerous redaction methods exist, their effectiveness varies, with some proving more robust than others. As such, the literature proposes several deanonymization techniques, raising awareness of potential privacy threats. However, while none of these methods are successful against the most effective redaction techniques, these attacks only focus on the anonymized tokens and ignore the sentence context. In this paper, we propose RedactBuster, the first deanonymization model using sentence context to perform Named Entity Recognition on reacted text. Our methodology leverages fine-tuned state-of-the-art Transformers and Deep Learning models to determine the anonymized entity types in a document. We test RedactBuster against the most effective redaction technique and evaluate it using the publicly available Text Anonymization Benchmark (TAB). Our results show accuracy values up to 0.985 regardless of the document nature or entity type. In raising awareness of this privacy issue, we propose a countermeasure we call character evasion that helps strengthen the secrecy of sensitive information. Furthermore, we make our model and testbed open-source to aid researchers and practitioners in evaluating the resilience of novel redaction techniques and enhancing document privacy.

Figures (3)

• Figure 1: Examples of redaction techniques.
• Figure 2: General pipeline for document redaction.
• Figure 3: RedactBuster framework overview.