A tale of two tools (NVSS and VisLink) Shreya Rawal 1
Dec 20, 2015
1
A tale of two tools (NVSS and VisLink)
Shreya Rawal
2
Network Visualization by Semantic Substrates
(Ben Shneiderman and Aleks
Aris)
3
Some problems in node-link diagram
Node Occlusion
Edge Crossing
Edge Tunneling under nodes
4
How to define semantic substrates?
Group nodes into regions According to one attribute
Categorical, ordinal, etc.
In each region: Place nodes according to the
remaining attribute(s)
Give users the control of link visibility
5
Dataset
Court Cases (Nodes)Supreme Court Cases
Circuit Court Cases
Citation of one case by another (Links)Within the same court (Circuit to Circuit and so on)
Within the different court (Circuit to Supreme and so on)
6
Force Directed Layout36 Supreme & 13 Circuit Court decisions
368 citations on Regulatory Takings 1978-2002
Arranged in chronological order from left to right
7
Force Directed Layout36 Supreme & 13 Circuit Court decisions
368 citations on Regulatory Takings 1978-2002
Arranged in chronological order from left to right
8
NVSS 1.0Supreme court cases
Circuit court cases
9
Filtering links by source-target
Check boxes for filtering the data
10
Filtering links by time (per year)
Slider for filtering by time
11
Three Regions
Links from District Courts
Indicates longevity of cases (short to long)
DistrictCircuitSupreme
12
NVSS Scalability
• 1280x1024• 1,122 nodes• 7,645 links
13
Analysis of Paraiso Manifesto using NVSS
Towers on the Island
Edges joining a call of specific duration to the corresponding tower
Calls arranged by duration anddays
14
Analysis of social network: Call Analysis for 1st Day
15
Analysis of social network: Comparison of long vs.short
duration calls
Short Duration Call Long Duration Call
16
VisLink: Revealing Relationships Amongst
Visualizations(Christopher Collins and Sheelagh
Carpendale)
17
VisLink
A method by which visualizations and the relationships between them can be interactively explored.
18
Compare multiple visualizations
Using different datasets and different visualization techniques
Same dataset but different visualizations
19
Theory behind VisLink (1)
VisLink: 3D space within which any number of 2D semi-transparent visualization planes are positioned.
20
Theory Behind VisLink (2)
Two individual visualizations
Required Visualization
21
Visualization of Lexical Data
Radial hyponymy graphhyponymy relationship: {lawyers, attorney} IS-A {job, occupation}
Force directed layout for similar clustering on words (alphabetically)Do some sets of synonyms contain high concentration of
orthographically (alphabetically) similar words?
22
Widget Interaction: VisLink
Book Pages Rotation
Center accordion
translation
Garage door rotation
Visualization planes are independently manipulated with three widgets.
23
3-D Navigation
24
Example
• Treemap showing occupations of the members of Congress before elections• Scatterplot of individual fundraising success• arranged alphabetically by states• party by the color of nodes• running for House or Senate by shape
• Geographic map of zipcode
25
Where did the most successful fundraising journalist get elected?
26
Semantic Substrate vs. VisLink
27
Comparison between NVSS and VisLink (1) Similarities:
Both dealt with multiple visualizations (on the same/multiple dataset or on the subsets of the same dataset).
Both focused on large amount of data.Both used user interaction for filtering data to extract maximum information.
28
Comparison between NVSS and VisLink (2)
Constraint:You cannot apply the concept of NVSS unless you have dataset which can be logically sub-divisible.
29
My Project:
Traces files generated by large-scale applications written in Erlang (CouchDB, Yaws, etc.).
Main components of dataProcesses
Process communication (message send and message receive)
Process spawning
Timestamp
30
Issues!!!
Is using multiple visualization solution to deal with more amount of data?
Big picture vs. Filtered Data?
2D vs. 3D?VisLink: a very effective use of 3rd dimension.
31
References:Aris, A. 2006. Network Visualization by Semantic Substrates. IEEE Transactions on Visualization and Computer Graphics 12, 5 (Sep. 2006), 733-740.
Carpendale, S. 2007. VisLink: Revealing Relationships Amongst Visualizations. IEEE Transactions on Visualization and Computer Graphics 13, 6 (Nov. 2007), 1192-1199.
Aris, A. and Shneiderman, B. 2007. Designing semantic substrates for visual network exploration. Information Visualization 6, 4 (Dec. 2007), 281-300.
Aris, A. and Vulleot, R. Visual Analysis of the Paraiso Manifesto by Symantic Substrate.
http://www.cs.umd.edu/hcil/nvss/#presentations(Last visited on March 2, 2009)
http://www.cs.toronto.edu/~ccollins/research/VisLink/flash/index.html(Last visited on March 2, 2009)