Preserving Privacy in Small Communities : Tailored Anonymization Techniques for Icelandic Conversational Data

Abstract

We examine the challenges and methodologies of anonymizing a dataset of Icelandic conversations, emphasizing the need for language-specific strategies due to Iceland’s small, interconnected population and the morphological richness of the language. We discuss the importance of preserving grammatical elements such as case and gender to maintain data utility for linguistic research. The study proposes an anonymization technique that balances data utility with privacy, resorting to pseudonyms that match the original phrase’s linguistic properties to protect individual identities while preserving the structural integrity of the Icelandic language.