PIIvot: A Lightweight NLP Anonymization Framework for Question-Anchored Tutoring Dialogues

TL;DR

PIIvot is a lightweight NLP framework designed for effective PII anonymization in educational dialogues, leveraging data context to improve detection, supported by a new large-scale tutoring dataset.

Contribution

Introduces PIIvot, a novel lightweight framework for PII anonymization that uses data context, and provides QATD-2k, the largest open-source tutoring dialogue dataset.

Findings

PIIvot improves PII detection accuracy with lower computational complexity.

QATD-2k enables better research and development in educational dialogue anonymization.

Framework demonstrates effectiveness on real-world tutoring data.

Abstract

Personally identifiable information (PII) anonymization is a high-stakes task that poses a barrier to many open-science data sharing initiatives. While PII identification has made large strides in recent years, in practice, error thresholds and the recall/precision trade-off still limit the uptake of these anonymization pipelines. We present PIIvot, a lighter-weight framework for PII anonymization that leverages knowledge of the data context to simplify the PII detection problem. To demonstrate its effectiveness, we also contribute QATD-2k, the largest open-source real-world tutoring dataset of its kind, to support the demand for quality educational dialogue data.