Faculty Recruiting Support CICS

Human-centered Technologies for inclusive collection and analysis of public-generated data

07 Jun
Tuesday, 06/07/2022 3:00pm to 5:00pm
Zoom
PhD Dissertation Proposal Defense
Speaker: Mahmood Jasim

Abstract: The meteoric rise in the popularity of public engagement platforms -- social media, customer review websites, and public input solicitation efforts by local and national governments -- promises an inclusive environment for public data generation and analysis at a local and global scale. However, people often struggle to engage, participate, and share their opinions due to inaccessibility, the rigidity of traditional public engagement methods, and the lack of options to provide opinions while avoiding potential confrontations. At the same time, data analysts and decision-makers grapple with the challenges of analyzing, sensemaking, and making informed decisions based on public-generated data, which includes high dimensionality, ambiguity present in human language, and a lack of tools and techniques that can enable them to peruse and sublimate public-generated data into concrete and actionable insights. In this dissertation, I explore the design and development of technologies that can facilitate inclusive public data collection and analysis and investigate their impact on public engagement and decision-making. 

First, I investigated traditional public consultation methods including town halls that remain the modus operandi for community consultation. Built on a formative study with 66 town hall attendees and 20 organizers, I designed and developed CommunityClick, a communitysourcing system that captures reticent attendees' feedback anonymously in real-time by using modified iClickers during town halls. It also enables organizers to author more comprehensive reports by providing them with a feedback-weighted summary of the transcript augmented with attendees' feedback. From a field experiment at a town hall meeting, I demonstrated how CommunityClick can improve inclusivity by providing multiple avenues for attendees to share opinions. Additionally, interviews with eight expert organizers demonstrated CommunityClick's utility in creating more comprehensive and accurate reports to inform critical civic decision-making. As an extension to the in-situ CommunityClick, I propose an improved online version --- CommunityClick-Virtual, that will allow attendees to share their opinions during online public meetings by providing free-form text opinions alongside customizable feedback. 

Next, I explored social media --- arguably the most prominent platform for public-generated data --- that has become an indispensable part of modern society and many of our daily lives. Popular media coverage suggests that the recent shift towards algorithmic content curation to increase user retention has diverted social media towards marketing, advertisement, and personal advocacy. Other factors might include the relatively stagnant interface design maintained to support user retention. In this work, I aim to take a step back and engage with social media users directly by conducting an interview study to explore their social media experience. An initial interview with 46 participants revealed that people are aware of diminished user agency, are actively looking for deeper and more meaningful interactions, and feel increasingly disconnected and unable to create meaningful and personal connections. Based on these responses, I will probe to identify future design prospects that could impact people's social media usage experience and modify the social media interface design space for connectivity and interactions.
 
Although increased access to online engagement platforms has allowed analysts and decision-makers -- especially in the civic domain -- to broaden their outreach to collect a larger number of public-generated data such as community input, comments, and ideas, sensemaking of such input remains a challenge due to the unstructured nature of text comments and the ambiguity present in human language. As such, community input is often left unanalyzed and unutilized in policymaking. To address this problem, I interviewed 14 civic leaders to understand their practices and requirements. I identified challenges around organizing the unstructured community input and surfacing the community's reflections beyond binary sentiments. Based on these insights, I built CommunityPulse, an interactive visual analytics system that scaffolds different facets of community input. I demonstrated the efficacy of CommunityPulse by evaluating the system with another 15 experts that suggest CommunityPulse is effective in surfacing multiple facets such as reflections, priorities, and hidden insights while reducing the required time, effort, and expertise for community input analysis.

Systems such as CommunityPulse provides decision-makers with an aggregated summary of public-generated data, which is predominantly free-form text in nature. However, summarization often leads to the suppression of marginalized opinions, which adds to the challenge of analyzing free-form text data. I explored ways to address this issue by investigating how supporting serendipitous discovery and analysis of short free-form texts can encourage readers to explore texts more comprehensively and allow them to make more confident decisions. To do so, I studied product reviews as one of the most prominent examples of public-generated data that have an impact on decision-making. I propose two interventions --- Exploration Metrics that can help readers understand and track their exploration patterns through visual indicators and a Bias Mitigation Model that intends to maximize knowledge discovery by suggesting sentiment and semantically diverse reviews. I designed, developed, and evaluated a visual text analytics system called Serendyze, where I integrated these interventions. I recruited 100 crowd workers to use Serendyze to make purchase decisions based on product reviews. The evaluation suggests that exploration metrics enabled readers to efficiently cover more reviews in a balanced way, and suggestions from the bias mitigating model influenced readers to make confident data-driven decisions. I propose to extend this work by investigating the utility of Serendyze in exploring civic input or social media posts to make informed data-driven decisions. 

Advisor: Narges Mayhar

Join via Zoom