


















Mohammad Hasan Payandeh, PhD candidate
David R. Cheriton School of Computer Science
Supervisor: Professor Jian Zhao
Exploratory data analysis (EDA) benefits from collaboration, yet distributed work often leaves analysts to make sense of data alone. We introduce Active Externalization Partners (AEPs), teams of conversational agents designed to scaffold solo analysts by turning externalized thoughts into opportunities for feedback, critique, and elaboration. Unlike prior AI-assisted EDA systems that rely on either a single agreeable assistant or opaque multi-agent pipelines, AEPs present stable, legible roles that analysts can engage with directly.
We instantiate this concept in DuckDuckTalk, a web-based conversational workspace centered on a computational notebook, where agents observe the analyst’s and other agents’ speech, code, and notebook actions. They respond to requests through dialogue, code suggestions, and annotations. We derived the system and agent design from a formative study of the collaborative roles analysts miss during solo work, and evaluated DuckDuckTalk in a controlled study using an EDA task. Results show positive improvement over the use of a mere chatbot for EDA, suggesting that AEPs can provide meaningful collaborative scaffolding for individual analysts.
To attend this PhD seminar in person, please go to DC 3317. You can also attend virtually on Zoom.
此内容由惯性聚合(RSS阅读器)自动聚合整理,仅供阅读参考。 原文来自 — 版权归原作者所有。