INTACT generates abstraction-sorted replacement candidates for sensitive spans and selects the most specific candidate that resists LLM-based inference attacks, achieving a strong privacy-utility trade-off on the Text Anonymization Benchmark.
Evaluating the stability of embedding- based word similarities
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CL 1years
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Truthful Text Sanitization Guided by Inference Attacks
INTACT generates abstraction-sorted replacement candidates for sensitive spans and selects the most specific candidate that resists LLM-based inference attacks, achieving a strong privacy-utility trade-off on the Text Anonymization Benchmark.