[Colloquium] Emma Peterson Candidacy Exam/Dec 9, 2025

[via Colloquium]

This is an announcement of Emma Peterson's Candidacy Exam.
===============================================
Candidate: Emma Peterson

Date: Tuesday, December 09, 2025

Time:  2:30 pm CST

Location: JCL 298

Title: Human-centered privacy: Exploring cross-disciplinary methods from data redaction to artistic communication

Abstract: Human-centered privacy is an approach to research that empowers individuals to take control of their data. In computer science, this often overlaps the fields of human-computer interaction (HCI) and usable privacy and security; however, lessons from these fields are broadly applicable to cross-disciplinary research as well, and vice-versa. In this proposal, we outline a conversation between computer science and humanities and the arts in two projects: one that uses methods from HCI and language analysis technology to solve privacy problems in humanities research (among other research spaces), and another that uses lessons from art applied to HCI privacy and security disciplines.

In the first, we explore redaction, a method of removing or substituting information that might make human subjects identifiable. In particular, we examine redaction and definition of identifiable information in unstructured text data, such as conversation transcripts, often collected by social and behavioral science (SBS) researchers. Based on previous work, we propose a design for a workflow of semi-automated, human-in-the loop redaction of identifiers, drastically reducing overhead for typical redaction workflows. This also integrates results from our previous work on determining what identifiable information to redact, as well as how to leverage existing tools and large language models (LLMs) to automatically flag items for users. The second project draws lessons from artists, investigating privacy communication topics through a large-scale collection and study of gallery art and exhibitions. Typical user-facing communication about privacy (e.g., privacy policies or dashboards) is frequently ignored and arguably boring.  Thus, we turned to artworks, which are deliberately provocative and engaging. In previous work, we compiled over 800 artworks related to privacy topics, qualitatively analyzing a sample, to see what artists are saying about privacy, and how; we then examined how these techniques might be applied to privacy communications within an HCI context. The proposed work will extend this study by implementing and evaluating design suggestions for integrating artistic mechanisms of communicating privacy topics with common privacy communication strategies.


Advisor: Blase Ur

Committee: Blase Ur, Marshini Chetty, Alex Kale