Papers
Topics
Authors
Recent
Search
2000 character limit reached

Silencing the Risk, Not the Whistle: A Semi-automated Text Sanitization Tool for Mitigating the Risk of Whistleblower Re-Identification

Published 2 May 2024 in cs.CY, cs.CL, cs.HC, cs.IR, and cs.SE | (2405.01097v1)

Abstract: Whistleblowing is essential for ensuring transparency and accountability in both public and private sectors. However, (potential) whistleblowers often fear or face retaliation, even when reporting anonymously. The specific content of their disclosures and their distinct writing style may re-identify them as the source. Legal measures, such as the EU WBD, are limited in their scope and effectiveness. Therefore, computational methods to prevent re-identification are important complementary tools for encouraging whistleblowers to come forward. However, current text sanitization tools follow a one-size-fits-all approach and take an overly limited view of anonymity. They aim to mitigate identification risk by replacing typical high-risk words (such as person names and other NE labels) and combinations thereof with placeholders. Such an approach, however, is inadequate for the whistleblowing scenario since it neglects further re-identification potential in textual features, including writing style. Therefore, we propose, implement, and evaluate a novel classification and mitigation strategy for rewriting texts that involves the whistleblower in the assessment of the risk and utility. Our prototypical tool semi-automatically evaluates risk at the word/term level and applies risk-adapted anonymization techniques to produce a grammatically disjointed yet appropriately sanitized text. We then use a LLM that we fine-tuned for paraphrasing to render this text coherent and style-neutral. We evaluate our tool's effectiveness using court cases from the ECHR and excerpts from a real-world whistleblower testimony and measure the protection against authorship attribution (AA) attacks and utility loss statistically using the popular IMDb62 movie reviews dataset. Our method can significantly reduce AA accuracy from 98.81% to 31.22%, while preserving up to 73.1% of the original content's semantics.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (74)
  1. Vigjilenca Abazi. 2020. The European Union whistleblower directive: a ‘game changer’for whistleblowing protection? Industrial Law Journal 49, 4 (2020), 640–656.
  2. AnonyMate: A toolkit for anonymizing unstructured chat data. In Proceedings of the Workshop on NLP and Pseudonymisation. Linköping University Electronic Press, Linköping, 1–7.
  3. The topic confusion task: A novel evaluation scenario for authorship attribution. In Findings of the Association for Computational Linguistics: EMNLP 2021. Association for Computational Linguistics, Punta Cana, Dominican Republic, 4242–4256. https://doi.org/10.18653/v1/2021.findings-emnlp.359
  4. Which anonymization technique is best for which NLP task?–It depends. A Systematic Study on Clinical Text Processing. arXiv e-prints (2022), arXiv–2209.
  5. Bettina Berendt and Stefan Schiffner. 2022. Whistleblower protection in the digital age-why “anonymous” is not enough.: From technology to a wider view of governance. The International Review of Information Ethics 31, 1 (2022).
  6. Going public: Researching external whistleblowing in a new media age. In International handbook on whistleblowing research. Edward Elgar Publishing, 250–272.
  7. An Efficient Big Data Anonymization Algorithm Based on Chaos and Perturbation Techniques. Entropy 20, 5 (2018), 373. article no.: 373; https://www.mdpi.com/1099-4300/20/5/373.
  8. Scaling instruction-finetuned language models. arXiv preprint arXiv:2210.11416 (2022).
  9. AI Now 2019 Report. https://ainowinstitute.org/publication/ai-now-2019-report-2
  10. Mary Cummings. 2004. Automation Bias in Intelligent Time Critical Decision Support Systems. In Proc. of the AIAA 1st Intelligent Systems Technical Conference. doi:10.2514/6.2004-6313.
  11. Rita de Sousa Costa and Inês de Castro Ruivo. 2020. Preliminary Remarks and Practical Insights on How the Whistleblower Protection Directive Adopts the GDPR Principles. In Privacy Technologies and Policy - 8th Annual Privacy Forum, APF 2020, Lisbon, Portugal, October 22-23, 2020, Proceedings (Lecture Notes in Computer Science, Vol. 12121), Luís Antunes, Maurizio Naldi, Giuseppe F. Italiano, Kai Rannenberg, and Prokopios Drogkaris (Eds.). Springer, 95–109. https://doi.org/10.1007/978-3-030-55196-4_6
  12. De-identification of patient notes with recurrent neural networks. Journal of the American Medical Informatics Association 24, 3 (2017), 596–606.
  13. Documenting large webtext corpora: A case study on the colossal clean crawled corpus. arXiv preprint arXiv:2104.08758 (2021).
  14. Josep Domingo-Ferrer. 2007. A three-dimensional conceptual framework for database privacy. In Secure Data Management: 4th VLDB Workshop, SDM 2007, Vienna, Austria, September 23-24, 2007. Proceedings 4. Springer, 193–202.
  15. Database anonymization: privacy models, data utility, and microaggregation-based inter-model connections. Synthesis Lectures on Information Security, Privacy, & Trust 8, 1 (2016), 1–136.
  16. Cynthia Dwork. 2006. Differential privacy. In Automata, Languages and Programming: 33rd International Colloquium, ICALP 2006, Venice, Italy, July 10-14, 2006, Proceedings, Part II 33. Springer, 1–12.
  17. European Data Protection Supervisor (EDPS) and Spanish Data Protection Agency (AEPD). 2021. 10 Misunderstandings Related to Anonymisation. https://edps.europa.eu/system/files/2021-04/21-04-27_aepd-edps_anonymisation_en_5.pdf. Joint statement on anonymisation of personal data according to EU GDPR.
  18. BertAA: BERT fine-tuning for Authorship Attribution. In Proceedings of the 17th International Conference on Natural Language Processing (ICON). 127–137.
  19. Ronen Feldman and James Sanger. 2007. The text mining handbook: advanced approaches in analyzing unstructured data. Cambridge university press.
  20. Generalised differential privacy for text document processing. In Principles of Security and Trust: 8th International Conference, POST 2019, Held as Part of the European Joint Conferences on Theory and Practice of Software, ETAPS 2019, Prague, Czech Republic, April 6–11, 2019, Proceedings 8. Springer International Publishing, 123–148.
  21. Whistleblowing Report 2019. http://whistleblowingreport.org/
  22. IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems. 2019. Ethically aligned design: A vision for prioritizing human well-being with autonomous and intelligent systems. Technical Report. IEEE. https://standards.ieee.org/wp-content/uploads/import/documents/other/ead_v2.pdf
  23. Deidentification of free-text medical records using pre-trained bidirectional transformers. In Proceedings of the ACM Conference on Health, Inference, and Learning. 214–221.
  24. Textwash–automated open-source text anonymisation. arXiv preprint arXiv:2208.13081 (2022).
  25. A survey on stylometric text features. In 2019 25th Conference of Open Innovations Association (FRUCT). IEEE, 184–195.
  26. Alaor Leite. 2021. Whistleblowing und das System der Rechtfertigungsgründe Das erlaubte Whistleblowing nach dem Geschaftsgeheimnisgesetz als, fürdernder Rechtfertigungsgrund”. Goltdammer’s Archiv für Strafrecht 168, 3 (2021), 129–146.
  27. t-closeness: Privacy beyond k-anonymity and l-diversity. In 2007 IEEE 23rd international conference on data engineering. IEEE, 106–115.
  28. Anonymisation models for text data: State of the art, challenges and future directions. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 4188–4203.
  29. Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692 (2019).
  30. De-identification of clinical notes via recurrent neural network and conditional random field. Journal of biomedical informatics 75 (2017), S34–S42.
  31. l-diversity: Privacy beyond k-anonymity. ACM Transactions on Knowledge Discovery from Data (TKDD) 1, 1 (2007), 3–es.
  32. The biases of pre-trained language models: An empirical study on prompt-based sentiment analysis and emotion detection. IEEE Transactions on Affective Computing (2022).
  33. Tanya M Marcum and Jacob Young. 2019. Blowing the whistle in the digital age: are you really anonymous? The perils and pitfalls of anonymity in whistleblowing law. DePaul Bus. & Comm. LJ 17 (2019), 1.
  34. Brian Martin. 2003. Illusions of whistleblower protection. UTS L. Rev. 5 (2003), 119.
  35. The Limits of Word Level Differential Privacy. In Findings of the Association for Computational Linguistics: NAACL 2022. 867–881.
  36. Use fewer instances of the letter “i”: Toward writing style anonymization. In Privacy Enhancing Technologies: 12th International Symposium, PETS 2012, Vigo, Spain, July 11-13, 2012. Proceedings 12. Springer, 299–318.
  37. Joseph McGlynn III and Brian K Richardson. 2014. Private support, public alienation: Whistle-blowers and the paradox of social support. Western Journal of Communication 78, 2 (2014), 213–237.
  38. Towards privacy preserving unstructured big data publishing. Journal of Intelligent & Fuzzy Systems 36, 4 (2019), 3471–3482.
  39. Clinical dialogue transcription error correction using Seq2Seq models. In Multimodal AI in healthcare: A paradigm shift in health intelligence. Springer, 41–57.
  40. Arvind Narayanan and Vitaly Shmatikov. 2010. Myths and fallacies of” personally identifiable information”. Commun. ACM 53, 6 (2010), 24–26.
  41. Janet P Near and Marcia P Miceli. 1985. Organizational dissidence: The case of whistle-blowing. Journal of business ethics 4, 1 (1985), 1–16.
  42. Hoang Nguyen and Sandro Cavallari. 2020. Neural multi-task text normalization and sanitization with pointer-generator. In Proceedings of the First Workshop on Natural Language Interfaces. 37–47.
  43. Generation of Replacement Options in Text Sanitization. In Proceedings of the 24th Nordic Conference on Computational Linguistics (NoDaLiDa). 292–300.
  44. Neural Text Sanitization with Explicit Measures of Privacy Risk. In Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 12th International Joint Conference on Natural Language Processing. 217–229.
  45. Constantinos Patsakis and Nikolaos Lykousas. 2023. Man vs the machine: The Struggle for Effective Text Anonymisation in the Age of Large Language Models. arXiv preprint arXiv:2303.12429 (2023).
  46. Andreas Pfitzmann and Marit Hansen. 2005. Anonymity, unlinkability, unobservability, pseudonymity, and identity management-a consolidated proposal for terminology – v. 0.28. Technical Report. https://dud.inf.tu-dresden.de/literatur/Anon_Terminology_v0.28.pdf
  47. David J. Phillips. 2004. Privacy policy and PETs: the influence of policy regimes on the development and social implications of privacy enhancing technologies. New Media and Society 6, 6 (2004), 691––706.
  48. The text anonymization benchmark (tab): A dedicated corpus and evaluation framework for text anonymization. Computational Linguistics 48, 4 (2022), 1053–1101.
  49. Sravana Reddy and Kevin Knight. 2016. Obfuscating gender in social media writing. In Proceedings of the First Workshop on NLP and Computational Social Science. 17–26.
  50. Joyce Rothschild and Terance D Miethe. 1999. Whistle-blower disclosures and management retaliation: The battle to control information about organization corruption. Work and occupations 26, 1 (1999), 107–128.
  51. Mary Saade. 2023. Women & Whistleblowing. Hastings Journal on Gender and the Law 34, 1 (2023), 43.
  52. Shikha Sachdeva and Narendra Singh Chaudhary. 2022. Exploring whistleblowing intentions of Indian nurses: a qualitative study. International Journal of Organizational Analysis ahead-of-print (2022).
  53. Pierangela Samarati and Latanya Sweeney. 1998. Protecting privacy when disclosing information: k-anonymity and its enforcement through generalization and suppression. Technical Report. Harvard Data Privacy Lab. https://dataprivacylab.org/dataprivacy/projects/kanonymity/paper3.pdf
  54. David Sánchez and Montserrat Batet. 2016. C-sanitized: A privacy model for document redaction and sanitization. Journal of the Association for Information Science and Technology 67, 1 (2016), 148–163.
  55. Detecting sensitive information from textual documents: an information-theoretic approach. In Modeling Decisions for Artificial Intelligence: 9th International Conference, MDAI 2012, Girona, Catalonia, Spain, November 21-23, 2012. Proceedings 9. Springer, 173–184.
  56. Automatic general-purpose sanitization of textual documents. IEEE Transactions on Information Forensics and Security 8, 6 (2013), 853–862.
  57. Utility-preserving privacy protection of textual healthcare documents. Journal of biomedical informatics 52 (2014), 189–198.
  58. The necessary illegitimacy of the whistleblower. Business & Professional Ethics Journal (2010), 85–107.
  59. Authorship attribution with topic models. Computational Linguistics 40, 2 (2014), 269–310.
  60. A4nt: author attribute anonymity by adversarial training of neural machine translation. In 27th {{\{{USENIX}}\}} Security Symposium ({{\{{USENIX}}\}} Security 18). 1633–1650.
  61. On identifying authors with style. In 2013 IEEE International Conference on Systems, Man, and Cybernetics. IEEE, 3048–3053.
  62. Automated systems for the de-identification of longitudinal clinical narratives: Overview of 2014 i2b2/UTHealth shared task Track 1. Journal of biomedical informatics 58 (2015), S11–S19.
  63. Marie Terracol. 2019. Building on the EU directive for whistleblower protection: analysis and recommendations. (2019).
  64. On the state of the art in authorship attribution and authorship verification. arXiv preprint arXiv:2209.06869 (2022).
  65. A de-identifier for medical discharge summaries. Artificial intelligence in medicine 42, 1 (2008), 13–35.
  66. Meghan Van Portfliet and Kate Kenny. 2022. Whistleblowing advocacy: Solidarity and fascinance. Organization 29, 2 (2022), 345–366.
  67. Progress in machine translation. Engineering (2021).
  68. Benjamin Weggenmann and Florian Kerschbaum. 2018. Syntf: Synthetic and differentially private term frequency vectors for privacy-preserving text mining. In The 41st International ACM SIGIR Conference on Research & Development in Information Retrieval. 305–314.
  69. Martin Weingardt. 2004. Fehler zeichnen uns aus: Transdisziplinäre Grundlagen zur Theorie und Produktivität des Fehlers in Schule und Arbeitswelt. Julius Klinkhardt.
  70. Simone White. 2018. A matter of life & death: whistleblowing legislation in the EU. In Eucrim: The European Criminal Law Associations’ Forum, Vol. 3. 170–177.
  71. Privacy-aware text rewriting. In Proceedings of the 12th International Conference on Natural Language Generation. 247–257.
  72. Authorship Identification Using Stylometry and Document Fingerprinting. In Big Data Analytics: 8th International Conference, BDA 2020, Sonepat, India, December 15–18, 2020, Proceedings 8. Springer, 278–288.
  73. Differential privacy for text analytics via natural text sanitization. arXiv preprint arXiv:2106.01221 (2021).
  74. Ying Zhao and Jinjun Chen. 2022. A survey on differential privacy for unstructured data content. ACM Computing Surveys (CSUR) 54, 10s (2022), 1–28.
Citations (2)
View on Semantic Scholar

Summary

No one has generated a summary of this paper yet.

Sign Up to Summarize

Paper to Video (Beta)

No one has generated a video about this paper yet.

Sign Up to Generate All Videos Subscribe on YouTube

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Sign Up to Generate

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Generate Now

Continue Learning

We haven't generated follow-up questions for this paper yet.

Generate Now

Collections

Sign up for free to add this paper to one or more collections.

Sign Up

Tweets

Sign up for free to view the 2 tweets with 0 likes about this paper.

Sign Up for Free