Due: Friday, April 5
Overview
Now that you have designed a scoring model, your next task is to design the user interface that allows a curator to interact with it.
The system will:
- Generate many possible gallery layouts
- Score them using your scoring model
- Present one or more layouts to the user
- Allow the user to provide feedback
- Use that feedback to guide the next iteration
Your job is to design how the human interacts with this system.
Objective
Design a user interface that:
- exposes your scoring model to the user
- allows meaningful feedback after each iteration
- supports iterative improvement (human-guided hill climbing)
Part 1: Interface Design
Describe and/or sketch your interface.
Include:
- how layouts are displayed (single layout, multiple options, etc.)
- how artworks are shown (position, size, metadata, etc.)
- what controls the user sees
You may use:
- diagrams
- wireframes
- labeled sketches
- simple mockups
Part 2: User Controls
How does the curator provide input?
Be specific. For example:
- sliders for preferred values
- controls for importance (weights)
- buttons like “more contrast” / “less contrast”
- ability to lock or remove artworks
- selecting between two candidate layouts
- marking parts of a layout as good or bad
Explain:
- what inputs exist
- what each input means
- how it connects to your scoring model
Part 3: Interaction Loop
Describe the full cycle:
- system generates layouts
- user sees layout(s)
- user gives feedback
- system updates scoring
- system generates new layouts
Be clear about:
- what happens after each user action
- how the system uses that input
Part 4: Types of Feedback
Identify at least two types of feedback your system supports:
Parameter-level feedback
Example:
- adjusting sliders
- changing weights
Layout-level feedback
Example:
- “these two should not be adjacent”
- “keep this artwork”
- “this side works better”
Explain how each type influences the next iteration.
Part 5: Algorithm Interaction
Describe how your interface connects to the algorithm:
- What changes in the scoring model when the user gives input?
- What kinds of “moves” should the algorithm try next?
- How does the system avoid getting stuck?
Part 6: Reflection
Briefly answer:
- What makes a good interface for this problem?
- What is difficult about translating human judgment into inputs?
- What could go wrong in your design?
Deliverable
Submit a document that includes:
- interface description (with diagrams or sketches)
- explanation of controls and interaction
- description of the feedback loop
- reflection
In-Class (April 5)
Each group will:
- present their interface design
- walk through a sample interaction
- explain how their system supports human-guided improvement
Goal
You are designing how humans and algorithms work together.
A good solution will:
- make it easy for the human to express intent
- make it clear how input affects the system
- support iterative refinement