Let’s Write Privacy: Survey and Experiment Design on Textual Privacy in Conversation
This research article emphasizes the significance of privacy in textual conversational data, specifically in the context of interactions between virtual assistants and users. While previous studies have primarily focused on named entities containing personally identifiable information, this research focuses on quasi-identifiers.´Due to the limited availability of conversational data in privacyinducing domains like healthcare, police force, and finance, the study employs a discrete prompting technique to engage large language models in generating conversations between hypothetical users and virtual assistants. To facilitate this process, this research introduces a straightforward interface. Additionally, a card-sorting experiment design is presented to involve participants in matching the conversations with persona cards representing hypothetical users. Subsequently, participants are asked to identify the information that aided them in making the appropriate matches. The experiment aims to outline the elements within textual data that can expose the identities of individuals engaged in conversations.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Agents That Know Too Much: A Data-Centric Survey of Privacy in LLM Agents
Large language model agents increasingly query databases, search document collections, call external APIs, remember past interactions, and act on a user's behalf. As they move from answering questions to operating over s…
Need to Know: Contextual-Integrity-Grounded Query Rewriting for Privacy-Conscious LLM Delegation
As LLMs become increasingly woven into everyday workflows, user queries sent to cloud hosted LLMs routinely mix task-essential content with task non-essential sensitive disclosures, yet type based PII redaction is contex…
Reinforcement LearningBeyond PII: How Users Attempt to Estimate and Mitigate Implicit LLM Inference
Large Language Models (LLMs) such as ChatGPT can infer personal attributes from seemingly innocuous text, raising privacy risks beyond memorized data leakage. While prior work has demonstrated these risks, little is know…
DP-Rewrite: Towards Reproducibility and Transparency in Differentially Private Text Rewriting
Text rewriting with differential privacy (DP) provides concrete theoretical guarantees for protecting the privacy of individuals in textual documents. In practice, existing systems may lack the means to validate their pr…
Privacy PreservingDP-MLM: Differentially Private Text Rewriting Using Masked Language Models
The task of text privatization using Differential Privacy has recently taken the form of $\textit{text rewriting}$, in which an input text is obfuscated via the use of generative (large) language models. While these meth…
Decoder