The Best Way to a Strong Defense is a Strong Offense: Mitigating Deanonymization Attacks via Iterative Language Translation

Mack, Nathan H.; Bowers, Jasmine; Williams, Henry; Dozier, Gerry; Shelton, Joseph

doi:10.7763/ijmlc.2015.v5.543

Cited by 16 publications

(20 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…transforming writing style but not semantic content. Back-and-forth machine translation provides a simple but highly limited technique [3,7,10,14,44], as it does not allow targeting or avoiding any particular style. Another classical alternative is rule-based paraphrase replacement from knowledge bases [12, 36-40, 45, 61], 1 https://gitlab.com/ssg-research/mlsec/parchoice/ Techniques Transformations applied [40,46] synonym replacement from WordNet [45,61] word embedding neighbour replacement [38] word replacement from GNU Diction [26], hand-crafted rules [12] synonym replacement from FreeLing [54], hand-crafted rules [36] synonym/hypernym/definition replacement from WordNet or PPDB, hand-crafted rules Table 1.…”

Section: Introductionmentioning

confidence: 99%

Effective writing style transfer via combinatorial paraphrasing

Gröndahl

Asokan

2020

Proceedings on Privacy Enhancing Technologies

View full text Add to dashboard Cite

Stylometry can be used to profile or deanonymize authors against their will based on writing style. Style transfer provides a defence. Current techniques typically use either encoder-decoder architectures or rule-based algorithms. Crucially, style transfer must reliably retain original semantic content to be actually deployable. We conduct a multifaceted evaluation of three state-of-the-art encoder-decoder style transfer techniques, and show that all fail at semantic retainment. In particular, they do not produce appropriate paraphrases, but only retain original content in the trivial case of exactly reproducing the text. To mitigate this problem we propose ParChoice: a technique based on the combinatorial application of multiple paraphrasing algorithms. ParChoice strongly outperforms the encoder-decoder baselines in semantic retainment. Additionally, compared to baselines that achieve nonnegligible semantic retainment, ParChoice has superior style transfer performance. We also apply ParChoice to multi-author style imitation (not considered by prior work), where we achieve up to 75% imitation success among five authors. Furthermore, when compared to two state-of-the-art rule-based style transfer techniques, ParChoice has markedly better semantic retainment. Combining ParChoice with the best performing rulebased baseline (Mutant-X [34]) also reaches the highest style transfer success on the Brennan-Greenstadt and Extended-Brennan-Greenstadt corpora, with much less impact on original meaning than when using the rulebased baseline techniques alone. Finally, we highlight a critical problem that afflicts all current style transfer techniques: the adversary can use the same technique for thwarting style transfer via adversarial training. We show that adding randomness to style transfer helps to mitigate the effectiveness of adversarial training.

show abstract

Section: Introductionmentioning

confidence: 99%

Effective writing style transfer via combinatorial paraphrasing

Gröndahl

Asokan

2020

Proceedings on Privacy Enhancing Technologies

View full text Add to dashboard Cite

show abstract

“…In addition to using ILT for obfuscation (see above), Day et al [36] applied iterative paraphrasing, creating the paraphrase dataset with the online tool Plagiarisma. Paraphrasing decreased the author identification rate with Hybrid-II [105] from 54% to 7% in the first iteration, 1% with the second iteration, and 6% with the third iteration. The LSA-value for paraphrased text was 0.80, indicating relatively high lexical overlap with the original text.…”

Section: Khosmood and Levinsonmentioning

confidence: 93%

“…Results on the effects of iterations are also indecisive. Almishari et al [4] decreased identification accuracy by adding iterations, whereas Mack et al [105] and Day et al [36] did not. More systematic comparative research would be needed to properly evaluate the effects of the languages, the number and direction of iterations, and the translation method.…”

Section: Automatic Obfuscationmentioning

confidence: 99%

“…Iterative language translation The easiest and most common method used for automatic style alteration and hence author obfuscation is machine translation (MT) across one or more languages back to the source language. This can also be called Iterative Language Translation (ILT) [105], as its effect can be increased by multiple iterations. It was first suggested by Rao and Rohatgi [142] and has since been applied in multiple studies, with divergent results.…”

Section: Automatic Obfuscationmentioning

confidence: 99%

See 1 more Smart Citation

Text Analysis in Adversarial Settings

Gröndahl

Asokan

2019

ACM Comput. Surv.

View full text Add to dashboard Cite

Textual deception constitutes a major problem for online security. Many studies have argued that deceptiveness leaves traces in writing style, which could be detected using text classification techniques. By conducting an extensive literature review of existing empirical work, we demonstrate that while certain linguistic features have been indicative of deception in certain corpora, they fail to generalize across divergent semantic domains. We suggest that deceptiveness as such leaves no content-invariant stylistic trace , and textual similarity measures provide a superior means of classifying texts as potentially deceptive. Additionally, we discuss forms of deception beyond semantic content, focusing on hiding author identity by writing style obfuscation . Surveying the literature on both author identification and obfuscation techniques, we conclude that current style transformation methods fail to achieve reliable obfuscation while simultaneously ensuring semantic faithfulness to the original text. We propose that future work in style transformation should pay particular attention to disallowing semantically drastic changes.

show abstract

“…Another work, from Nathan et al [121], evaluates Bowers's work by introducing a feature selection approach, namely Generative and Evolutionary Feature Selection (GEFES), over the set of predefined features that mask out non-salient previously extracted features. Both Reference [36] and Reference [121] are tested over a set of blog posts by users and the results show the efficiency of ILT-based anonymization. A recent work is also proposed by Zhang et al [191] that anonymizes users' textual information before publishing user-generated data.…”

Section: Authors In Social Media and Privacymentioning

confidence: 99%

A Survey on Privacy in Social Media

Beigi

Liu

2020

ACM/IMS Trans. Data Sci.

View full text Add to dashboard Cite

The increasing popularity of social media has attracted a huge number of people to participate in numerous activities on a daily basis. This results in tremendous amounts of rich user-generated data. These data provide opportunities for researchers and service providers to study and better understand users' behaviors and further improve the quality of the personalized services. Publishing user-generated data risks exposing individuals' privacy. Users privacy in social media is an emerging research area and has attracted increasing attention recently. These works study privacy issues in social media from the two different points of views: identification of vulnerabilities and mitigation of privacy risks. Recent research has shown the vulnerability of user-generated data against the two general types of attacks, identity disclosure and attribute disclosure. These privacy issues mandate social media data publishers to protect users' privacy by sanitizing user-generated data before publishing it. Consequently, various protection techniques have been proposed to anonymize user-generated social media data. There is vast literature on privacy of users in social media from many perspectives. In this survey, we review the key achievements of user privacy in social media. In particular, we review and compare the state-of-the-art algorithms in terms of the privacy leakage attacks and anonymization algorithms. We overview the privacy risks from different aspects of social media and categorize the relevant works into five groups: (1) social graphs and privacy, (2) authors in social media and privacy, (3) profile attributes and privacy, (4) location and privacy, and (5) recommendation systems and privacy. We also discuss open problems and future research directions regarding user privacy issues in social media.

show abstract

The Best Way to a Strong Defense is a Strong Offense: Mitigating Deanonymization Attacks via Iterative Language Translation

Abstract: Abstract-In

Cited by 16 publications

References 8 publications

Effective writing style transfer via combinatorial paraphrasing

Effective writing style transfer via combinatorial paraphrasing

Text Analysis in Adversarial Settings

A Survey on Privacy in Social Media

Contact Info

Product

Resources

About