Saved in:
Bibliographic Details
Main Authors: Ye, Junhong, Yuan, Xu, Qiu, Xinying
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.11862
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915721704898560
author Ye, Junhong
Yuan, Xu
Qiu, Xinying
author_facet Ye, Junhong
Yuan, Xu
Qiu, Xinying
contents Accurate recognition of personally identifiable information (PII) is central to automated text anonymization. This paper investigates the effectiveness of cross-domain model transfer, multi-domain data fusion, and sample-efficient learning for PII recognition. Using annotated corpora from healthcare (I2B2), legal (TAB), and biography (Wikipedia), we evaluate models across four dimensions: in-domain performance, cross-domain transferability, fusion, and few-shot learning. Results show legal-domain data transfers well to biographical texts, while medical domains resist incoming transfer. Fusion benefits are domain-specific, and high-quality recognition is achievable with only 10% of training data in low-specialization domains.
format Preprint
id arxiv_https___arxiv_org_abs_2507_11862
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Cross-Domain Transfer and Few-Shot Learning for Personal Identifiable Information Recognition
Ye, Junhong
Yuan, Xu
Qiu, Xinying
Computation and Language
Accurate recognition of personally identifiable information (PII) is central to automated text anonymization. This paper investigates the effectiveness of cross-domain model transfer, multi-domain data fusion, and sample-efficient learning for PII recognition. Using annotated corpora from healthcare (I2B2), legal (TAB), and biography (Wikipedia), we evaluate models across four dimensions: in-domain performance, cross-domain transferability, fusion, and few-shot learning. Results show legal-domain data transfers well to biographical texts, while medical domains resist incoming transfer. Fusion benefits are domain-specific, and high-quality recognition is achievable with only 10% of training data in low-specialization domains.
title Cross-Domain Transfer and Few-Shot Learning for Personal Identifiable Information Recognition
topic Computation and Language
url https://arxiv.org/abs/2507.11862