CLJun 30, 2024

A Collocation-based Method for Addressing Challenges in Word-level Metric Differential Privacy

arXiv:2407.00638v127 citations
Originality Incremental advance
AI Analysis

This addresses a specific challenge in applying differential privacy to NLP tasks, offering an incremental improvement for researchers and practitioners in privacy-preserving text analysis.

The paper tackles the problem of poor semantic coherence in word-level metric differential privacy for NLP by perturbing n-gram collocations instead of single words, resulting in higher semantic coherence and variable length outputs as demonstrated in utility and privacy tests.

Applications of Differential Privacy (DP) in NLP must distinguish between the syntactic level on which a proposed mechanism operates, often taking the form of $\textit{word-level}$ or $\textit{document-level}$ privatization. Recently, several word-level $\textit{Metric}$ Differential Privacy approaches have been proposed, which rely on this generalized DP notion for operating in word embedding spaces. These approaches, however, often fail to produce semantically coherent textual outputs, and their application at the sentence- or document-level is only possible by a basic composition of word perturbations. In this work, we strive to address these challenges by operating $\textit{between}$ the word and sentence levels, namely with $\textit{collocations}$. By perturbing n-grams rather than single words, we devise a method where composed privatized outputs have higher semantic coherence and variable length. This is accomplished by constructing an embedding model based on frequently occurring word groups, in which unigram words co-exist with bi- and trigram collocations. We evaluate our method in utility and privacy tests, which make a clear case for tokenization strategies beyond the word level.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes