Learnable and Expressive Visualization Authoring through Blended Interfaces
1. Learnable and Expressive
Visualization Authoring through
Blended Interfaces
Sehi L’Yi, Astrid van den Brandt, Etowah Adams, Huyen N. Nguyen, Nils Gehlenborg
2. How can we design a system that enables users to
easily create highly customized visualization?
3. How can we design a system that enables users to
easily create highly customized visualization?
Learnability Expressivity
12. Our Ideas Multimodal User Interaction
:
InChorus (Srinivasan et al. CHI 2020); DataBreeze (Srinivasan et al. TVCG 2021); TouchPivot (Jo et al. CHI 2017); Data@Hand (Kim et al. CHI 2021)
13. Our Ideas Multimodal User Interaction
:
InChorus (Srinivasan et al. CHI 2020); DataBreeze (Srinivasan et al. TVCG 2021); TouchPivot (Jo et al. CHI 2017); Data@Hand (Kim et al. CHI 2021)
14. Our Ideas Multimodal User Interaction
:
InChorus (Srinivasan et al. CHI 2020); DataBreeze (Srinivasan et al. TVCG 2021); TouchPivot (Jo et al. CHI 2017); Data@Hand (Kim et al. CHI 2021)
15. Our Ideas Multimodal User Interaction
:
InChorus (Srinivasan et al. CHI 2020); DataBreeze (Srinivasan et al. TVCG 2021); TouchPivot (Jo et al. CHI 2017); Data@Hand (Kim et al. CHI 2021)
16. Our Ideas:
InChorus (Srinivasan et al. CHI 2020); DataBreeze (Srinivasan et al. TVCG 2021); TouchPivot (Jo et al. CHI 2017); Data@Hand (Kim et al. CHI 2021)
+
Multimodal User Interaction
Complementarity Interaction Martin, 1998
17. Our Ideas Multimodal User Interaction
:
InChorus (Srinivasan et al. CHI 2020); DataBreeze (Srinivasan et al. TVCG 2021); TouchPivot (Jo et al. CHI 2017); Data@Hand (Kim et al. CHI 2021)
+
Complementarity Interaction Martin, 1998
18. Our Ideas Multimodal User Interaction
:
InChorus (Srinivasan et al. CHI 2020); DataBreeze (Srinivasan et al. TVCG 2021); TouchPivot (Jo et al. CHI 2017); Data@Hand (Kim et al. CHI 2021)
+ +
Martin, 1998
Complementarity Interaction
21. Our Ideas Learnability Studies
:
Visualization Literacy, Visualization Onboarding, etc
Kwon and Lee, CHI 2016; Ruchikachorn and Mueller, TVCG 2015; Kim et al. CHI 2023; Wang et al. TVCG 2022; Stoiber et al. Vis. Psychol. 2023
?
22. Our Ideas Learnability Studies
:
Kwon and Lee, CHI 2016; Ruchikachorn and Mueller, TVCG 2015; Kim et al. CHI 2023; Wang et al. TVCG 2022; Stoiber et al. Vis. Psychol. 2023
?
!
• Brushing & Linking
• Difference
• Morphing
Linking
!
Visualization Literacy, Visualization Onboarding, etc
35. User Study Design
Hands-on Sessions
15 mins
Introduction
20 mins
Self-Learning
15 mins
Reproduction Interview
10 mins
Task 1 Task 2 Task 3
Pre-Study Survey Post-Study Survey
36. Interaction Patterns
“It is very nice to choose between templates, and then, in shelves, I can
change x, y, and color. And then, if I really want to fine-tune the visualization, I
could go into the code or ask AI to do it for me. It was very useful to have all
four, and it wasn’t anything cluttered or something like that.” P6
37. Interaction Patterns
“I didn’t know what to put in the code editor, so I just asked to see what the
natural language interface changes. And, based on that, I can continue
myself” P1
39. “I love that [changes on the shelf configuration interface] reflected in the code
because, particularly, I am not familiar with Gosling. [...] It was kind of very cool
to see how they are talking to each other. That kind of allows me to
understand what is going on. I feel that that kind of teaches me Gosling.”
P3
40. “The different kinds of scatterplots
confused me sometimes. I was somehow
expecting only one type of each plot
instead of slight variations.”
P1
41. Main Takeaway
Combining and linking authoring interfaces can
help learn a unfamiliar authoring system to create
complex visualizations
42. • Screen scalability
• How to combine authoring techniques without output interfaces?
e.g., Visualization by Demonstration (Saket et al. 2017)
• Would users always start with the most familiar interfaces?
Open Questions
• Screen scalability
• How to combine authoring techniques without output interfaces?
e.g., Visualization by Demonstration (Saket et al. 2017)
• Would users always start with the most familiar interfaces?
43. Summary
• A conceptual framework of blended interfaces to
blend multiple authoring interfaces
• An implementation of Blace for genomics data
visualization
• A user evaluation with domain experts (n=12) to
demonstrate the effectiveness of blended interfaces sehilyi.com
[email protected]
SEHI L’YI
@sehi_lyi
@sehilyi.bsky.social
NIH K99/R00 Postdoc Fellow
Harvard Medical School
Formative
User Study
Tomorrow