Bootstrapping Text Anonymization Models with Distant Supervision

TL;DR

This paper introduces a method to train text anonymization models using distant supervision from knowledge graphs, eliminating the need for manual labeling and enabling scalable privacy-preserving text processing.

Contribution

It presents a novel approach that leverages knowledge graphs for automatic annotation to bootstrap text anonymization models without manual data labeling.

Findings

Effective training of anonymization models using knowledge graph annotations.

Challenges due to noise and incompleteness in knowledge graphs.

Multiple valid anonymization solutions can exist for the same text.

Abstract

We propose a novel method to bootstrap text anonymization models based on distant supervision. Instead of requiring manually labeled training data, the approach relies on a knowledge graph expressing the background information assumed to be publicly available about various individuals. This knowledge graph is employed to automatically annotate text documents including personal data about a subset of those individuals. More precisely, the method determines which text spans ought to be masked in order to guarantee $k$-anonymity, assuming an adversary with access to both the text documents and the background information expressed in the knowledge graph. The resulting collection of labeled documents is then used as training data to fine-tune a pre-trained language model for text anonymization. We illustrate this approach using a knowledge graph extracted from Wikidata and short biographical texts from Wikipedia. Evaluation results with a RoBERTa-based model and a manually annotated collection of 553 summaries showcase the potential of the approach, but also unveil a number of issues that may arise if the knowledge graph is noisy or incomplete. The results also illustrate that, contrary to most sequence labeling problems, the text anonymization task may admit several alternative solutions.