Description
Poster - IEEE VIS: Visualization & Visual Analytics 2024, Florida
This poster is a work in progress of a greater research project for which we
seek to answer the following two questions:
Our guiding research questions are as follows:
1. How do different platforms (glanceable AR, non-AR mobile, and non-AR desktop) impact the social dynamics and visual presentation of real-time conversation analysis and data interpretation, and how do these factors shape user interaction with the information?
2. What types of data analysis and visual encodings of user’s real-time conversation data generally provide the most value?
'
This project explores how glanceable, real-time conversation timelines can enhance discussions in Augmented Reality (AR) while remaining lightweight and non-intrusive. AR is increasingly being used for conversational support, but most existing approaches focus on persistent or highly interactive visualizations. Instead, I wanted to design a system that augments conversations without distracting from them, allowing users to quickly glance at topic summaries and dismiss them as needed. Using Large Language Models (LLMs) and Natural Language Processing (NLP), I developed a method to recognize and classify topic shifts in real time. For visualization, I used XReal smart glasses, which offer an unobtrusive, everyday wearable AR experience. This work identifies key challenges and opportunities in minimalist AR conversation visualization, and through prototyping and testing, I reflect on different design techniques to make AR-enhanced conversations more seamless and intuitive.
