My CHI 2022 Reading List
32 papers on data science, health, bias, and more
It’s time for CHI 2022, so here’s my personal reading list, based on titles, abstracts, and a quick skim of all the papers published in the ACM DL. (I’m interested in data science, health, and bias in modeling and annotation, so that’s mostly what you’ll find here.) I intentionally tried to cultivate a bias toward systems and interfaces, and I list the 32 selected papers in a randomized order.
Here are my personal bonus awards:
Best session title: “Mouth-based interaction”
Best paper title: “The Voight-Kampff Machine for Automatic Custom Gesture Rejection Threshold Selection”
Coolest concept: “O&O: A DIY toolkit for designing and rapid prototyping olfactory interfaces”
[Tensions and opportunities for] participatory design collaborations with *communities* rather than individuals, based on a lit survey
Designing for the Bittersweet: Improving Sensitive Experiences with Recommender Systems
Interviews with people experiencing bittersweet content through Facebook’s “Memories” recommendation feature
Iterative task model for data science workflows
Tisane: Authoring Statistical Models via Formal Reasoning from Conceptual and Data Relationships
Python package to facilitate building mixed-effects models from experimental assumptions
Variable trust in recommendation systems, when labeled as based on content (“your unique taste”), collaborative filtering (“based on like-minded people”), and demographics (“similar to you in age, ethnicity, and gender”)
Quantitative study of Twitter’s “Birdwatch” program
Review of empirical research on digital “nudges”
Not Just a Preference: Reducing Biased Decision-making on Dating Websites
Study on how presentation order of profile information affects racially biased decision-making on online dating platforms
What’s the Appeal? Perceptions of Review Processes for Algorithmic Decisions
Contestability of AI decisions, and what aspects of review processes makes them seem fair
“Mindsets”: explicit intent categories for people to more productively engage with recommendation systems. I like the qual -> quant -> validate approach (and the great process figure), echoes of my experiences (at ICWSM). Sruthi has another paper on POI recommendation at CHI 2022 as well.
Privacy-switching behavior on Twitter, a quantitative analysis
Jury Learning: Integrating Dissenting Voices into Machine Learning Models
An interesting extension of Gordon’s disagreement deconvolution that explicitly uses group/identity information about annotators to influence model predictions.
Model Positionality and Computational Reflexivity: Promoting Reflexivity in Data Science
Subjectivity in data science. The “annotator fingerprinting” proposal is really interesting, and I intend to mull on it for a bit.
Interesting idea for sharing YouTube recommendations (I had seen this already on Twitter).
Use of online mental health tests/self-screeners by young adults.
Whose AI Dream? In search of the aspiration in data annotation.
“Data annotation is a systematic exercise of power through organizational structure and practice”. This work adds to findings from Miceli et al.’s CSCW paper “Between subjectivity and imposition: Power dynamics in data annotation for computer vision”.
Pretty Princess vs. Successful Leader: Gender Roles in Greeting Card Messages
Interesting dataset: greeting card messages scraped from online templates.
Comparative qualitative study of 3 Wikipedia language editions, explaining differences in contribution rates despite similar contexts.
AI-Moderated Decision-Making: Capturing and Balancing Anchoring Bias in Sequential Decision Tasks
Cool quantitative study on the anchoring effect, where sequential labeling results in biased labeling.
Putting scientific results in perspective: Improving the communication of standardized effect sizes
On reporting of effect size.
OneLabeler: A Flexible System for Building Data Labeling Tools
Visual workflow editor for building up labeling/annotation interfaces. I have made a few of these in the last few years, and it looks like OneLabeler could have made that easier!
Diff in the Loop: Supporting Data Comparison in Exploratory Data Analysis
Automatic snapshotting of Pandas dataframes during exploratory analysis, to enable cross-version comparisons
Community prayer support as a component of digital health interventions
Semantic Gap in Predicting Mental Wellbeing through Passive Sensing
Inferring mental well-being, raising hard problems about which signals and “ground truth”
Human-AI Collaboration via Conditional Delegation: A Case Study of Content Moderation
Conditional delegation: a framework for formalizing the conditions where human and machine classifiers take particular actions (content moderation case study).
Two-handed Design: Development of Food Personality Framework Using Mixed Method Needfinding
Interesting mixed-method design, including validating unsupervised clusters via qual themes.
Taxonomy of shared control in human + AI systems, based on a lit/system survey.
Strategies for Fostering a Genuine Feeling of Connection in Technologically Mediated Systems
Design strategies for fostering human connection.
What Pronouns for Pepper? A Critical Review of Gender/ing in Research
Researcher and participant gendering of a gender-ambiguous humanoid robot. (Also, weird ACM DL page on this one for some reason.)
Designing Word Filter Tools for Creator-led Comment Moderation
Tool for managing word filters for content moderation.
Templates and Trust-o-meters: Towards a widely deployable indicator of trust in Wikipedia
Explicit trust indicators on Wikipedia articles, based on their edit history.
“Data silences in data work”: intentional and unintentional erasure of history and processes in data science practice. This paper taught me the term “epistemicide”.
That’s all, folks! Thanks to all the researchers for an exciting batch of work at CHI 2022!
