--- title: "STAMP: Selective Task-Aware Mechanism for Text Privacy (2026)" description: "For the noise itself, the authors propose a polar mechanism that keeps an embedding's norm fixed and… 1 reference, cited by 1 paper here." authors: ["Fengwei Tian", "Payel Bhattacharjee", "Heidi Hanson", "Geoffrey D Rubin", "Joseph Y. Lo", "Ravi Tandon"] year: 2026 venue: "Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)" source: "https://aclanthology.org/2026.eacl-long.61/" doi: "10.18653/v1/2026.eacl-long.61" licence: "CC BY 4.0 (ACL Anthology)" licence_url: "https://creativecommons.org/licenses/by/4.0/" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/en/papers/tian-2026-stamp-selective-task-aware-mechanism/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/tian-2026-stamp-selective-task-aware-mechanism/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/17/) / STAMP: Selective Task-Aware Mechanism for Text Privacy # STAMP: Selective Task-Aware Mechanism for Text Privacy **Authors:** Fengwei Tian, Payel Bhattacharjee, Heidi Hanson, Geoffrey D Rubin, Joseph Y. Lo and Ravi Tandon · **Year:** 2026 · **Venue:** [Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)](https://md.research.elamur.ai/free/en/venues/eacl/) ([EACL 2026](https://md.research.elamur.ai/free/en/venues/eacl/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/) **Links:** [Publisher's page](https://aclanthology.org/2026.eacl-long.61/) · [PDF](https://aclanthology.org/2026.eacl-long.61.pdf) · [DOI 10.18653/v1/2026.eacl-long.61](https://doi.org/10.18653/v1/2026.eacl-long.61) · [OpenAlex](https://openalex.org/W7140145616) **Citations:** 0 citations (OpenAlex, 2026-10-14) · 1 in our library, of the 149,439 papers whose bibliographies we read ## Summary The paper addresses client-side privatization of user text before it reaches a remote LLM or other downstream consumer, under local differential privacy (LDP). Earlier approaches, such as randomized response or isotropic Laplace/Gaussian noise on embeddings, typically apply the same budget to every token and damage meaning. STAMP instead splits tokens into groups along two axes. One is how relevant a token is to the task, measured by similarity to a task- or query-specific representation. The other is whether the token is sensitive, as flagged by NER/PII detection of items like names, dates and identifiers. Each group then receives its own privacy budget. For the noise itself, the authors propose a polar mechanism that keeps an embedding's norm fixed and randomizes only its direction on the unit sphere using von Mises-Fisher sampling. Text is recovered by cosine nearest-neighbor lookup, so the decoding geometry matches the noise geometry. On SQuAD, Yelp and AG News, STAMP gives better utility than uniform budgeting at equal per-token budgets, and the polar mechanism beats Laplace noise. The extra computation is negligible, which makes task-aware text privacy practical to deploy. ## Key points - Tokens are sorted into four groups (G1–G4) by crossing task importance with privacy sensitivity. Noise is concentrated on content that is sensitive but unimportant to the task. Utility falls steadily as more groups are perturbed, and Uniform allocation is the lowest-utility case. - The gains are largest when importance is conditioned on a question, as in SQuAD. They are smaller with fixed prompts, as in Yelp and AG News. - Overhead is small: - The grouping step adds about 0.002 s per SQuAD example (roughly 180 tokens). - STAMP-Polar runs at 35.16 s per example (195 ms/token), versus 34.54 s (192 ms/token) for the Laplace baseline. - vMF sampling scales as O(d), comparable to Gaussian or Laplace sampling. - The authors list several limitations: - The method needs a meaningful task description. - Grouping relies on static embedding similarity. - Results depend on the quality of PII detection. - With 768-dimensional embeddings, fairly large per-token ε values are needed to keep utility. *Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.* ## Institutions - [Duke University](https://md.research.elamur.ai/free/en/universities/duke-university/) — University, Durham, United States · 28 works · 15,077 citations - [University of Arizona](https://md.research.elamur.ai/free/en/universities/university-of-arizona/) — University, Tucson, United States · 38 works · 10,603 citations - [Oak Ridge National Laboratory](https://md.research.elamur.ai/free/en/labs/oak-ridge-national-laboratory/) — Research institute, Oak Ridge, United States · 3 works · 44 citations 📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works · [🤝 Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/) · 1,772 works · [🤝 Sociology and Political Science](https://md.research.elamur.ai/free/en/shelves/social-sciences/sociology-and-political-science/) · 577 works ## Cite Cite as: Fengwei Tian, Payel Bhattacharjee, Heidi Hanson, Geoffrey D Rubin, Joseph Y. Lo and Ravi Tandon. 2026. STAMP: Selective Task-Aware Mechanism for Text Privacy. Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers). The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @inproceedings{tian2026stamp, title = {{STAMP: Selective Task-Aware Mechanism for Text Privacy}}, author = {Fengwei Tian, Payel Bhattacharjee, Heidi Hanson, Geoffrey D Rubin, Joseph Y. Lo and Ravi Tandon}, booktitle = {Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)}, year = {2026}, doi = {10.18653/v1/2026.eacl-long.61}, url = {https://doi.org/10.18653/v1/2026.eacl-long.61} } ``` - [← previous SSA: Improving Performance With a Better Scoring Function](https://md.research.elamur.ai/free/en/papers/naim-2026-ssa-improving-performance-better-scoring/) - [next → Start Making Sense(s): A Developmental Probe of Attention Specialization Using Lexical Ambiguity](https://md.research.elamur.ai/free/en/papers/riviere-2026-start-making-sense-s-developmental/) - [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/17/) ## References 1 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. Pathum Chamikara Mahawaga Arachchige, Peter Bertok, Ibrahim Khalil, Dongxi Liu, Seyit Camtepe, and Mohammed Atiquzzaman. 2019. Local differential privacy for deep learning. IEEE Internet of Things Journal, 7(7):5827–5842. \#\# Ethical Considerations Stefan Arnold, Dilara Yesilbas, and Sven Weinzierl.… — [arXiv:2306.01457](https://arxiv.org/abs/2306.01457) [find](https://scholar.google.com/scholar?q=Pathum+Chamikara+Mahawaga+Arachchige%2C+Peter+Bertok%2C+Ibrahim+Khalil%2C+Dongxi+Liu%2C+Seyit+Camtepe%2C+and+Mohammed+Atiquzzaman%2E+2019%2E+Local+differential+privacy+for+deep+learning%2E+IEEE+Internet+of+Things+Jou) *Updated 31 October 2026*