Midterm 2 Flashcards
What are the primary data types?
1-dimensional
* 2-dimensional
* 3-dimensional
* Multi-dimensional
* Temporal
* Tree
* Network
* Text
* Categorical
What is 1D Data?
Usually a list
What are the concerns with 1D Data?
Space and cognition concerns
What are some ways to lessen the concerns of 1D Data?
- Grouping
- Hierarchical arrangement
- Fish eye design
What are the pros of 1D design solutions?
- Reduce the required space for
displaying the whole list - Make the search more efficient
What are the cons of 1D design solutions?
- Require additional knowledge/information to organize
the list - Where to put the divider, how to build a hierarchy
- Navigation could be an issue
- Fisheye display is a non-linear mapping between information
space (e.g., spatial presentation) and interaction control space
(e.g., mouse movement)
What is 2D data?
Documents, tables, maps, etc.
What is the problem with 2D data?
Balance between detail and context
Focused Content
What users are intreseted in
Context information
What users may need to know about the contexts of focused contents
What are the basic techniques?
- What the size of something is according to users interests
- Context on demand
What is an example of distortion?
Fisheye views
What is an example of multiple scales?
Overviews
What are two examples of context on demand?
Switch between two discrete states and continuous zooming
What are the pros of context on demand?
Allowing the access to both context and content information
What are the cons of context on demand?
- Increasing cognitive burdens
- Interactive control over views needs to be intuitive
enough.
What is real 3D spatial data?
Physically mapped to real-world phenomena
What is artificial 3D spatial data?
Mapping is defined by users/desginers
What are the challenges of displaying 3D data?
- How to leverage the depth information to help people?
- Type 1
- Type 2
What is a type 1 challenge when displaying 3D data?
- Natural spatial mapping
- Focus is on other attributes: color, shape, etc.
What is a type 2 challenge when displaying 3D data?
- What attributes to use for three axes?
- How to represent other attributes?
What are the pros of 3D design?
- More useful space for richer information
- Making information more interesting and fun
- Easy connection between real-world phenomena and
visualized objects
What are the cons of 3D design?
- Navigation could be challenging
- Interpreting certain 3D relationships could be cognitively
expensive - Slow down the machine dramatically.
What is an example of multi-dimensional data?
- A user on facebook
– A possible design for
airplane wing
What is the most common multi-dimensional design
Parallel coordinates
What are the pros of multi-dimensional design?
- Can accommodated many dimensions
– Can deal with both numerical and categorical data
– Support the observation of relationship among
dimensions (correlation)
What are the cons of multi-dimensional design?
- Required trained eyes to read the chart
– Location of dimensional axes affects the
perception of the result.
What are two types of tree design?
- Hyperbolic tree
- Space tree
Different levels in analysis of Text
Corpora ➔ Corpus ➔ Document cluster➔ Document ➔ Word
Corpora
All emails of DNC web server
Corpus
all emails by a person
Document cluster
messages on one specific issue
Document
a particular email message
Word
individual words
What to display?
Raw text
* Relationship among entities
* Summarization of text
* Measures
* Criteria
Measures
Words and frequency
Criteria
temporal evolution, group
comparison, etc.
Raw text
The content of documents, codes,
books, etc.
What are the issues with raw text
- Focus + Context
- Type of text
- Comparing text
- Topics of documents
Topics of A Corpus
Using natural linguistic processing (NLP) tools
Topic
bag-of-word with high frequencies of occurrence
Large scale level
document collections, not a specific
document
What are the limitations of a word cloud?
- Size of word collection
- Control of stop words
Geospatial data
Data about objects, events, activities, etc.
Data with attributes on locations
Geospatial data visualization
Visualizing certain data attributes by mapping location
attributes to a map
Types of Geospatial Visualization
Vector-based
Raster-Based
What are the two types of vector-based geospatial visualization?
Shapes
Scalable
What are the two types of Rasetr-based geospatial visualization?
Pixels (bitmap)
Non-scalable
What is a popular raster-based visualization method?
Heatmaps
What are the pros of heat maps?
- Showing complex data distribution in one view
- Using different mapping functions
- Data prepared offline (good to very large datasets)
What are the cons of heat maps?
- Static, usually
- Limited user interactions
- Non-scalable (poor visual results at certain scales)
- Poor transparency of mapping functions.
What is a popular vector-based geo-visualization method?
Choropleth Map
What is a Choropleth Map?
- Geographic areas filled with color based on aggregated
data at a scale level - Geographic objects (e.g., areas, lines, points) are drawn
based on data from shape files.
What are the pros of a Choropleth Map?
- Very interactive: color, area,
- Flexible on visualized attributes
- Availability of shape data
What are the cons of a Choropleth Map?
Matching between attribute and shape data
Cartogram
GeoVisualization Design
size of areas are proportional to an attribute value).
What are the two general approaches for analysis?
Top-down
Bottom-up
Top-down
- Template matching, model fitting,
- Filling the blank with provided information
Bottom-up
- Building a model, an explanation,
- Extracting and connecting useful concepts based on
provided information
Top-down Steps
- Have a hypothesis based on certain guidance, theory, experience, etc.
- Know what types of information to search
- Find and analyze information
- Confirm the hypothesis
- Or refine the hypothesis and start the process again
Pros of the Top-down approach
- Efficient: search can be more purposeful
- Reliable: solid theories (usually)
- Logical: theories, again
Cons of the Top-down approach
- Restrict data requirement
- Availability, quality, quantity
- Overreliance on the hypothesis (prone to errors or
biases
Bottom-up Steps
- Have data
- Search through data, find patterns
- build a
hypothesis
Bottom-up Pros
May lead to new understandings
Bottom-up Cons
- Misled by data (e.g., untruthful data)
- Hard to be generalized: may not be theory at all.
- Uncertain or incomplete results
Implications for Visual Analytics Design
- Understanding analytical processes
- Sensemaking, information foraging
- Supporting analytical activities
- Visualization and organization of data
- Supporting processes
Sensemaking
a new or unfamiliar situation where
people need to understand what is going on
What is an example of sensemaking
Getting lost in a new city
Outcomes of sensemaking
- New knowledge, new facts, …
- Declarative knowledge
- New problem-solving techniques
Essence of Sensemaking
- A process that involves learning, doing, and understanding.
- A process in which user activities are often situated.
- A process that consists of information gathering, interpretation,
and synthesis.
How can we use Visualization to Support Sensemaking Processes
- Help people collect information
- Help people understand and analyze information
- Help people organize information
- Help people understand their activities
What are the different aspects of the challenges facing visual analytics?
Process, data, model, knowledge, AI
What is the process aspect?
- How does an analyst work?
- “Don’t make me think” ➔ “Help me think quickly and
effectively” - How do a group of analysts work?
- Group dynamics + visual analytics
What is the data aspect?
- Scalability of data
- Tools for small data sets may work for large data sets.
- Quality of data
- Data is often messy.
What is the model aspect?
- Only considering data attributes
- Distributions, types, etc.
what is the knowledge aspect?
- What new knowledge to provide
- What do they know already?
What is the theory of the visual analytics process?
The visual analytics process always involves complex
activities like problem-solving, and decision-making.
Collaborative Visual Analytics
Analyzing big data often requires
team work.
What are the two aspects when working effectively as a team?
Social aspects
Task aspects
How to support collaboration?
Information artifacts
Visualization tools
Distributed expertise
- Divide tasks into subtasks
- Integrate results from individual experts
Biases
People favor information they are familiar with and down-play
information they don’t know much.
what do models not do?
consider interactivity with users
What role can AI play?
data processing, datapattern discovery, graph selection, graph design, UI design, system evaluation
Well-defined problems
Clear information about initial state, goal state, operators,
and operator restrictions
Ill-defined problems
Little or no information about any components
