@inproceedings{meisenbacher-etal-2024-collocation,
title = "A Collocation-based Method for Addressing Challenges in Word-level Metric Differential Privacy",
author = "Meisenbacher, Stephen and
Chevli, Maulik and
Matthes, Florian",
editor = "Habernal, Ivan and
Ghanavati, Sepideh and
Ravichander, Abhilasha and
Jain, Vijayanta and
Thaine, Patricia and
Igamberdiev, Timour and
Mireshghallah, Niloofar and
Feyisetan, Oluwaseyi",
booktitle = "Proceedings of the Fifth Workshop on Privacy in Natural Language Processing",
month = aug,
year = "2024",
address = "Bangkok, Thailand",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.privatenlp-1.5",
pages = "39--51",
abstract = "Applications of Differential Privacy (DP) in NLP must distinguish between the syntactic level on which a proposed mechanism operates, often taking the form of *word-level* or *document-level* privatization. Recently, several word-level *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 *between* the word and sentence levels, namely with *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.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="meisenbacher-etal-2024-collocation">
<titleInfo>
<title>A Collocation-based Method for Addressing Challenges in Word-level Metric Differential Privacy</title>
</titleInfo>
<name type="personal">
<namePart type="given">Stephen</namePart>
<namePart type="family">Meisenbacher</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maulik</namePart>
<namePart type="family">Chevli</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Florian</namePart>
<namePart type="family">Matthes</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-08</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Fifth Workshop on Privacy in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ivan</namePart>
<namePart type="family">Habernal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sepideh</namePart>
<namePart type="family">Ghanavati</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Abhilasha</namePart>
<namePart type="family">Ravichander</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vijayanta</namePart>
<namePart type="family">Jain</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Patricia</namePart>
<namePart type="family">Thaine</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Timour</namePart>
<namePart type="family">Igamberdiev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Niloofar</namePart>
<namePart type="family">Mireshghallah</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Oluwaseyi</namePart>
<namePart type="family">Feyisetan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Bangkok, Thailand</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Applications of Differential Privacy (DP) in NLP must distinguish between the syntactic level on which a proposed mechanism operates, often taking the form of *word-level* or *document-level* privatization. Recently, several word-level *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 *between* the word and sentence levels, namely with *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.</abstract>
<identifier type="citekey">meisenbacher-etal-2024-collocation</identifier>
<location>
<url>https://aclanthology.org/2024.privatenlp-1.5</url>
</location>
<part>
<date>2024-08</date>
<extent unit="page">
<start>39</start>
<end>51</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T A Collocation-based Method for Addressing Challenges in Word-level Metric Differential Privacy
%A Meisenbacher, Stephen
%A Chevli, Maulik
%A Matthes, Florian
%Y Habernal, Ivan
%Y Ghanavati, Sepideh
%Y Ravichander, Abhilasha
%Y Jain, Vijayanta
%Y Thaine, Patricia
%Y Igamberdiev, Timour
%Y Mireshghallah, Niloofar
%Y Feyisetan, Oluwaseyi
%S Proceedings of the Fifth Workshop on Privacy in Natural Language Processing
%D 2024
%8 August
%I Association for Computational Linguistics
%C Bangkok, Thailand
%F meisenbacher-etal-2024-collocation
%X Applications of Differential Privacy (DP) in NLP must distinguish between the syntactic level on which a proposed mechanism operates, often taking the form of *word-level* or *document-level* privatization. Recently, several word-level *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 *between* the word and sentence levels, namely with *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.
%U https://aclanthology.org/2024.privatenlp-1.5
%P 39-51
Markdown (Informal)
[A Collocation-based Method for Addressing Challenges in Word-level Metric Differential Privacy](https://aclanthology.org/2024.privatenlp-1.5) (Meisenbacher et al., PrivateNLP-WS 2024)
ACL