Recent Technologies in Differential Privacy for NLP Applications
Yingyi Wu, Xinpeng Xie, Zhaomin Xiao, Jinran Zhang, Zhuoer Xu, Zhelu Mai · 2024
Differential Privacy (DP) has become a crucial framework for preserving data privacy, especially in the field of Natural Language Processing (NLP). This survey provides a comprehensive overview of existing DP applications in NLP, focusing on key aspects such as application scenario classification, diversity of privacy mechanisms, utility-performance trade-offs, data granularity, empirical studies versus theoretical analysis, and emerging challenges with future directions.