14 April 2023 in Proceedings of the ACM on Human-Computer Interaction
In recent years, the popularity of AI-enabled conversational agents or chatbots has risen as an alternative to traditional online surveys to elicit information from people. However, there is a gap in using single-agent chatbots to converse and gather multi-faceted information across a wide variety of topics. Prior works suggest that single-agent chatbots struggle to understand user intentions and interpret human language during a multi-faceted conversation. In this work, we investigated how multi-agent chatbot systems can be utilized to conduct a multi-faceted conversation across multiple domains. To that end, we conducted a Wizard of Oz study to investigate the design of a multi-agent chatbot for gathering public input across multiple high-level domains and their associated topics. Next, we designed, developed, and evaluated CommunityBots - a multi-agent chatbot platform where each chatbot handles a different domain individually. To manage conversation across multiple topics and chatbots, we proposed a novel Conversation and Topic Management (CTM) mechanism that handles topic-switching and chatbot-switching based on user responses and intentions. We conducted a between-subject study comparing CommunityBots to a single-agent chatbot baseline with 96 crowd workers. The results from our evaluation demonstrate that CommunityBots participants were significantly more engaged, provided higher quality responses, and experienced fewer conversation interruptions while conversing with multiple different chatbots in the same session. We also found that the visual cues integrated with the interface helped the participants better understand the functionalities of the CTM mechanism, which enabled them to perceive changes in textual conversation, leading to better user satisfaction. Based on the empirical insights from our study, we discuss future research avenues for multi-agent chatbot design and its application for rich information elicitation.
1 June 2022 in Computer Graphics Forum
10 June 2024 in Computer Graphics Forum
16 October 2020 in IEEE Transactions on Visualization and Computer Graphics
Differential Privacy is an emerging privacy model with increasing popularity in many domains. It functions by adding carefully calibrated noise to data that blurs information about individuals while preserving overall statistics about the population. Theoretically, it is possible to produce robust privacy-preserving visualizations by plotting differentially private data. However, noise-induced data perturbations can alter visual patterns and impact the utility of a private visualization. We still know little about the challenges and opportunities for visual data exploration and analysis using private visualizations. As a first step towards filling this gap, we conducted a crowdsourced experiment, measuring participants' performance under three levels of privacy (high, low, non-private) for combinations of eight analysis tasks and four visualization types (bar chart, pie chart, line chart, scatter plot). Our findings show that for participants' accuracy for summary tasks (e.g., find clusters in data) was higher that value tasks (e.g., retrieve a certain value). We also found that under DP, pie chart and line chart offer similar or better accuracy than bar chart. In this work, we contribute the results of our empirical study, investigating the task-based effectiveness of basic private visualizations, a dichotomous model for defining and measuring user success in performing visual analysis tasks under DP, and a set of distribution metrics for tuning the injection to improve the utility of private visualizations.