Topicalizer: Reframing Core Concepts in Machine Learning Visualization by Co-designing for Interpretivist Scholarship

Status

Eric P. S. Baumer, Drew Siedel, Lena McDonnell, Jiayun Zhong, Patricia Sittikul, and Micki McGee. (2020). Topicalizer: Reframing Core Concepts in Machine Learning Visualization by Co-designing for Interpretivist Scholarship. Human-Computer Interaction, Special Issue on Unifying Human Computer Interaction and Artificial Intelligence 35(5-6): 452-480.

Abstract

Computational algorithms can provide novel, compelling functionality for interactive systems. However, designing such systems for users whose expertise lies outside computer science poses novel and complex challenges. This paper focuses specifically on the domain of designing interactive topic modeling visualizations to support interpretivist scholars. It describes a co-design process that involved working directly with two such scholars across two different corpora. The resultant visualization has both several similarities and key differences with other topic modeling visualizations, illustrated here using both the final design and discarded prototypes. The paper’s core contribution is an attention to how our emphasis on interpretation played out, both in the design process and in the final visualization design. The paper concludes by discussing the kinds of issues and tensions that emerged in the course of this work, as well as the ways that these issues and tensions can apply to much broader contexts of designing interactive algorithmic systems.

DOI | pdf

All Users are (Not) Created Equal: Predictors Vary for Different Forms of Facebook Non/use

Status

Eric P. S. Baumer, Patrick Skeba, Shion Guha, and Geri Gay. (2019). All Users are (Not) Created Equal: Predictors Vary for Different Forms of Facebook Non/use. Proceedings of the ACM Human-Computer Interaction 2, CSCW.

Abstract

Relatively little work has empirically examined use and non-use of social technologies as more than a dichotomous binary, despite increasing calls to do so. This paper compares three different forms of non/use that might otherwise fall under the single umbrella of Facebook “user”: (1) those who have a current active account; (2) those who have deactivated their account; and (3) those who have considered deactivating but not actually done so. A subset of respondents (N=256) from a larger, demographically representative sample of internet users completed measures for usage and perceptions of Facebook, Facebook addiction, privacy experiences and behaviors, and demographics. Multinomial logistic regression modeling shows four specific variables as most predictive of a respondent’s type: negative effects from “addictive” use, subjective intensity of Facebook usage, number of Facebook friends, and familiarity with or use of Facebook’s privacy settings. These findings both fill gaps left by, and help resolve conflicting expectations from, prior work. Furthermore, they demonstrate how valuable insights can be gained by disaggregating “users” based on different forms of engagement with a given technology.

DOI

Who is the “Human” in Human-Centered Machine Learning: The Case of Predicting Mental Health from Social Media

Status

Stevie Chancellor, Eric P. S. Baumer, and Munmun De Choudhury. (2019). Who is the “Human” in Human-Centered Machine Learning: The Case of Predicting Mental Health from Social Media. Proceedings of the ACM Human-Computer Interaction 2, CSCW. [Honorable Mention Award]

Abstract

“Human-centered machine learning” (HCML) combines human insights and domain expertise with data-driven predictions to answer societal questions. This area’s inherent interdisciplinarity causes tensions in the obligations researchers have to the humans whose data they use. This paper studies how scientific papers represent human research subjects in HCML. Using mental health status prediction on social media as a case study, we conduct thematic discourse analysis on 55 papers to examine these representations. We identify five discourses that weave a complex narrative of who the human subject is in this research: Disorder/Patient, Social Media, Scientific, Data/Machine Learning, and Person. We show how these five discourses create paradoxical subject and object representations of the human, which may inadvertently risk dehumanization. We also discuss the tensions and impacts of interdisciplinary research; the risks of this work to scientific rigor, online communities, and mental health; and guidelines for stronger HCML research in this nascent area.

DOI

Speaking on Behalf of: Representation, Authority, and Delegation in Computational Text Analysis

Status

Eric P. S. Baumer and Micki McGee. (2019). Speaking on Behalf of: Representation, Authority, and Delegation in Computational Text Analysis. in Proceedings of the AAAI/ACM Conference on Artificial Intelligence, Ethics, and Society (AIES). (Honolulu, HI).

Abstract

Computational tools can often facilitate human work by rapidly summarizing large amounts of data, especially text. Doing so delegates to such models some measure of author- ity to speak on behalf of those people whose data are being analyzed. This paper considers the consequences of such delegation. It draws on sociological accounts of representation and translation to examine one particular case: the application of topic modeling to blogs written by parents of children on the autism spectrum. In doing so, the paper illustrates the kinds of statements that topic models, and other computational techniques, can make on behalf of people. It also articulates some of the potential consequences of such statements. The paper concludes by offering several suggestions about how to address potential harms that can occur when computational models speak on behalf of someone.

DOI

Departing and Returning: Sense of Agency as an Organizing Concept for Understanding Social Media Non/use Transitions

Status

Eric P. S. Baumer, Rui Sun, and Peter Schaedler. (2018). Departing and Returning: Sense of Agency as an Organizing Concept for Understanding Social Media Non/use Transitions. Proceedings of the ACM: Human-Computer Interaction 1, CSCW: 23:1-23:19.

Abstract

Recent work has identified a variety of motivations for various forms of technology use and non-use. However, less work has closely examined relationships between those motivations and the experiences of transiting among these different forms of use and non-use. This paper fills that gap by conducting a qualitative interview- and diary-based study where participants were asked to deactivate their Facebook account. An abductive analysis suggests participants’ experiences can be organized under the conceptual umbrella of sense of agency, which refers to an individual’s perception that their actions are under their own control. The analysis shows how, across disparate motivations, all participants took actions that increased their own subjective sense of agency, regardless of whether they returned to Facebook or not. The discussion applies this conceptual lens to prior studies of technology use and non-use. Doing so shows how sense of agency may provide an organizing orientation for understanding subjective experiences of use and non-use.

DOI

Interpretive Impacts of Text Visualization: Mitigating Political Framing Effects

Status

Eric P. S. Baumer, Jaime Snyder, and Gery Gay. (2018). Interpretive Impacts of Text Visualization: Mitigating Political Framing Effects. ACM Transactions on Human-Computer Interaction (ToCHI), 25(4), 20:1–20:26.

Abstract

Information visualizations are often evaluated as a tool in terms of their ability to support performance of a specific task. This article argues that value can be gained by instead evaluating visualizations from a communicative perspective. Specifically, it explores how text visualization can influence the impacts that framing has on the perception of political issues. Using data from a controlled laboratory study, the results presented here demonstrate that exposure to a text visualization can mitigate framing effects. Furthermore, it also shows a transfer effect, where participants who saw the visualization remained uninfluenced by framing in subsequent texts, even when the visualization was absent. These results carry implications for the methods used to evaluate information visualization systems, for understanding the cognitive and interpretive mechanisms by which framing effects occur, and for exploring the design space of interactive text visualization.

DOI

Socioeconomic Inequalities in the Non/use of Facebook

Status

Eric P. S. Baumer (2018). Socioeconomic Inequalities in the Non/use of Facebook. in Proceedings of the ACM Conference on Human Factors in Computing Systems (CHI). (Montréal, QC).

Abstract

Use and non-use of technology can occur in a variety of forms. This paper analyzes data from a probabilistic sample of 1000 US households to identify predictors for four different types of use and non-use of the social media site Facebook. The results make three important contributions. First, they demonstrate that many demographic and socioeconomic predictors of social media use and non-use identified in prior studies hold with a larger, more diverse sample. Second, they show how going beyond a binary distinction between use and non-use reveals inequalities in social media use and non-use not identified in prior work. Third, they contribute to ongoing discussions about the representativeness of social media data by showing which populations are, and are not, represented in samples drawn from social media.

ACM  pre-print

A Hat Trick at GROUP

I recently had three submissions accepted to the ACM Conference on Supporting Group Work (GROUP). The first was a paper to which I contributed about online policy discussion, specifically in the context of MTurk. Among other things, this paper offers some nice empirical evidence about the importance of considering opinions with finer grained distinctions than agree vs. disagree. Second is a paper analyzing how different types of regretful experiences on Facebook can lead to different types of non-use. The main take away is that it matters less who feels the regret than who takes the action that ends up being regretted. Third is a curated collection of short design fiction pieces written by students in the class I taught this past spring. It demonstrates the efficacy of design fiction for thinking through ethical issues in computing.

What Would You Do? Design Fiction and Ethics

Status

Eric P. S. Baumer, Timothy Berrill, Sarah C. Botwinick, Jonathan L. Gonzales, Kevin Ho, Allison Kundrik, Luke Kwon, Tim LaRowe, Chan P. “Sam” Nguyen, Fredy Ramirez, Peter Schaedler, William Ulrich, Amber Wallace, Yuchen Wan, and Benjamin Weinfeld. (2018). What Would You Do? Design Fiction and Ethics. in Proceedings of the ACM Conference on Supporting Group Work (GROUP), Design Fiction Track. (Sanibel Island, FL).

Abstract

Design fiction can be highly effective at envisioning possible futures. That envisioning enables, among other things, considering ethical implications of possible technologies. This paper highlights that capacity through a cu- rated collection of five short design fiction pieces, each accompanied by its own author statement. Spanning multiple genres, each piece highlights ethical issues in its own way. After considering the unique strategies that each piece uses to highlight ethical issues, the paper concludes with considerations of how design fiction can advance broader discussions of ethics in computing.

Regrets, I’ve Had A Few: When Regretful Experiences Do (and Don’t) Compel Users to Leave Facebook

Status

Shion Guha, Eric P. S. Baumer, and Geri Gay. (2018). Regrets, I’ve Had A Few: When Regretful Experiences Do (and Don’t) Compel Users to Leave Facebook. in Proceedings of the ACM Conference on Supporting Group Work (GROUP). (Sanibel Island, FL).

Abstract

Previous work has explored regretful experiences on social media. In parallel, scholars have examined how people do not use social media. This paper aims to synthesize these two research areas and asks: Do regretful experiences on social media influence people to (consider) not using social media? How might this influence differ for different sorts of regretful experiences? We adopted a mixed methods approach, combining topic modeling, logistic regressions, and contingency analysis to analyze data from a web survey with a demographically representative sample of US internet users (n=515) focusing on their Facebook use. We found that experiences that arise because of users’ own actions influence actual deactivation of their Facebook account, while experiences that arise because of others’ actions lead to considerations of non-use. We discuss the implications of these findings for two theoretical areas of interest in HCI: individual agency in social media use and the networked dimensions of privacy.

DOI