More about HKUST
Turning the Physical World into a Data Interface: Augmenting Situated Visualization and Data Storytelling through Physical Affordances
PhD Thesis Proposal Defence
Title: "Turning the Physical World into a Data Interface: Augmenting Situated
Visualization and Data Storytelling through Physical Affordances"
by
Mr. Kentaro TAKAHIRA
Abstract:
In our daily lives, data is embedded in the physical environments we inhabit.
However, interacting with this data remains challenging because conventional
visualization paradigms are primarily designed for two-dimensional screens.
When applied to the physical world, these screen-based interactions often
fail to leverage users' natural embodied cognition and struggle to adapt to
diverse, opportunistic environments. This disconnect results in significant
cognitive and manipulative friction during everyday sensemaking and
synchronous data storytelling.
To address this pervasive challenge, this thesis proposes turning the
physical world itself into an active data interface. By systematically
mapping and leveraging a spectrum of physical affordances, ranging from
spatial configurations and tangible manipulations to the semantic states of
everyday objects, this approach transforms passive environments into dynamic
media for data interaction. Drawing on Augmented Reality (AR), Computer
Vision, and Generative AI, this research establishes comprehensive frameworks
to seamlessly synchronize users' physical actions with data visualizations.
I investigate this physical-digital integration through three interconnected
systems. The first study introduces VisTranslator, a multimodal AR system
that addresses everyday situated sensemaking. Driven by a novel Large
Language Model (LLM) pipeline, it interprets unstructured, physically
embedded data to dynamically generate in-situ visualizations, allowing users
to interpret and interact with static embedded data in a personalized,
context-aware way through object-based manipulations. The second study
explores tangible interaction for structuring complex data during synchronous
storytelling. I present TangibleNet, an AR prototype that employs physical
magnets as tangible proxies for network nodes. This system enables presenters
to intuitively filter and restructure node-link diagrams in real time,
creating a direct, low-cognitive-load link between physical manipulation and
digital control. The third study proposes semantic coupling to achieve
seamless integration between physical objects and their associated data
visualizations. InSituTale introduces the novel concept of augmented physical
data storytelling, leveraging the spatial configurations and semantic states
of everyday objects to trigger context-aware visualization changes.
Ultimately, this thesis provides empirical evidence and actionable design
implications demonstrating how utilizing the inherent affordances of physical
objects can fundamentally enhance intuitive, expressive data interactions
across varied physical contexts.
Date: Monday, 28 April 2026
Time: 10:00am - 12:00pm
Venue: Room 2132C
Lift 22
Committee Members: Prof. Huamin Qu (Supervisor)
Dr. Xiaojuan Ma (Chairperson)
Dr. Tristan Braud