Download Doctoral Research Agenda Information Science Peter A. Hook Information Visualization Laboratory

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Doctoral Research Agenda
Peter A. Hook
Information Visualization Laboratory
March 22, 2006
Information Science
Information Visualization,
Knowledge Organization
Systems,
Bibliometrics
Legal Informatics,
Organization of Legal
Information,
Access to Legal
Information
Semantic Network
Theory of Learning,
Schematic Maps,
Pedagogy of the Online
Environment
Law
Educational Psychology /
Cognitive Science
1
Select Productivity to Date
Hook, Peter A. and Börner,
Katy. (2005) Educational
Knowledge Domain
Visualizations: Tools to
Navigate, Understand, and
Internalize the Structure of
Scholarly Knowledge and
Expertise. In Amanda Spink
and Charles Cole (eds.) New
Directions in Cognitive
Information Retrieval.
Springer-Verlag.
2004 IV Lab Open House
2005 SLIS – Doctoral Forum
2005 IV Lab Open House
2006 ALISE – San Antonio
2006 L578: User Interface Design
2006 Law & Networks UIUC
Law / IS / Pedagogy
General Research Question:
How can network graphing and
information visualization techniques
improve the understanding of the work of
the United States Supreme Court?
2
Peter A. Hook,
Network Derived Educational Visualizations
of the Work of the United States Supreme Court
JD MSLIS
Doctoral Student
Indiana University--Bloomington
School of Library and Information Science
http://ella.slis.indiana.edu/~pahook
Ideological Landscape of the Justices (1994 – 2003)
Visualizing Complex Joining Patterns
Opinion:
Part 1
Dissent
Stevens
Base Map Creation
• GOAL: Create a topical Base Map to serve as
a common reference point on which to layer
additional information
United States v. Booker, 125 S.Ct. 738 (2005).
July 2, 2005 New York Times
SLIS
• Technique 1 – Use the co-occurrence of top
level West topics (Key Numbers) to render the
topical adjacencies of American case law.
Breyer
• There are three types of West top level topics:
Opinion:
Part 2
ƒ Doctrinal (blue – Torts, Contracts)
ƒ Factual (red – Automobiles, Aviation)
Breyer
Appointed by a Democrat
Stevens I
Except
Footnote
17,
Part IV
Appointed by a Republican
Stevens II
ƒ Procedural (Federal Civil Procedure, Courts)
Stevens III
• PROBLEM – The three types can co-occur in a
wide variety of cases. For instance, procedural
topics may co-occur with just about any factual
or substantive topic.
Stevens IV
Voting frequencies represented as the edge weight between nodes and presented
visually as a graph. Scalia and Thomas vote most frequently together and are joined
least frequently by Stevens. O’connor, and to a lesser extent Kennedy, are the judges
most likely to join the liberal members of the Court. (Rendered with Pajek using a
stochastic, spring force algorithm.)
Dissent
Dissent
Scalia
Thomas
Customary, Non-Visual, Explanation of
Joining Relationships of the Justices
• This creates an interconnected mess when
visualized with spring force algorithms.
Voting Together > 50% (Non-Unanimous Cases)
Main Opinion
Dissent
Swing Vote Status of O’Connor and Kennedy
Network Graphic Approach to
Booker
Frequency of Voting Blocks in 5-4 Cases
(1994 -2003 Supreme Court Terms)
82
47%
17 5
Dissen
t
Opinio
n: Part
1
28
Breyer
Steve
ns
Voting Together > 49% (Non-Unanimous Cases)
16%
14
8%
Stevens I
Stevens
II
Stevens
IV
Except
Footnot
e 17,
Part IV
Opinio
n: Part
2
51
(Highest repetition – 3 times)
• All top level West topics
assigned to United States
Supreme Court cases from the
1944 term through the end of the
2004 term and their cooccurrence.
• 7,948 unique cases to which
19,789 topic assignments have
been made.
17 5
17 5
ALL OTHER GROUPINGS OF 5
(34 different groupings)
• DATASET - The dataset consists
of:
29%
Breyer
Stevens
III
• Of the 405 topics in the West
taxonomy, 291 appear in
opinions issued by the Supreme
Court for this time period.
• This results in 22,345 edges with
3743 unique topic pairings.
17 5
Total 5 to 4 Cases = 175
Source: Statistics harvested from the Harvard Law Review
West’s Topics by Specialty as a Network
• The domain map to the left is
the network relationship of all
West Topics by Specialty
minus the 20% most tenuous
topics that pulled everything to
the center.
Dissen
t
Disse
nt
Scalia
Thom
as
The above image represents all topics identified as
doctrinal and assigned to law School class subjects.
These were then subjected to a double treatment. (1)
Each of the 55 classes was paired with its most
frequently occurring subjects. (2) The graph was
reduced to an edge weight exceeding 10 case cooccurrences.
2004 Supreme Court Term West Topic Space (Procedural Topics Removed)
• The topic assignments are
more from the perspective of
a practitioner than a law
student.
• Note the lack of treatment of
Constitutional Law.
Visualizing the Harvard Law Review
Supreme Court Statistics
July 2, 2005 New York Times
3
Ideological Landscape of the Justices (1994 – 2003)
Appointed by a Democrat
Appointed by a Republican
Voting frequencies represented as the edge weight between nodes and presented
visually as a graph. Scalia and Thomas vote most frequently together and are joined
least frequently by Stevens. O’connor, and to a lesser extent Kennedy, are the judges
most likely to join the liberal members of the Court. (Rendered with Pajek using a
stochastic, spring force algorithm.)
Frequency of Voting Blocks in 5-4 Cases
(1994 -2003 Supreme Court Terms)
47%
82
17 5
28
16%
17 5
14
8%
17 5
ALL OTHER GROUPINGS OF 5
51
29%
(34 different groupings)
(Highest repetition – 3 times)
17 5
Total 5 to 4 Cases = 175
Source: Statistics harvested from the Harvard Law Review
4
Thresholding (Voting Together > 50%)
Reveals Ideological Cliques
Thresholding (Voting Together > 49%)
Reveals Ideological Cliques
5
Additional
Work
(Less Well Known
Stories)
Use the taxonomy of case types in
the Table of Contents to aggregate
voting associations by individual
topics.
Criminal Law and Procedure
Freedom of Speech & Expression
2nd Expansion – Expand Voting
Associations back to 1967
Richard M. Nixon
Gerald R. Ford
James E. Carter
Ronald Reagan
George H.W. Bush
William J. Clinton
1969-1974
Republican
1974-1977
Republican
1977-1981
Democrat
1981-1989
Republican
1989-1993
Republican
1993-2001
Democrat
1967-1968
1972
1975
1981
1986
1988
1990
1991
1993
1994
NIXON
FORD
REAGAN
REAGAN
REAGAN
H.W. BUSH
H.W. BUSH
CLINTON
CLINTON
1994-1995
2003-2004
Current Visualizations / Data
6
Base Map/Overlay Pedagological
Visualizations of the Work of the United
States Supreme Court
• Create visualizations of the work of the
United States Supreme Court to be used
for teaching.
• Create a topical Base Map to serve as a
common reference point on which to layer
additional information.
Base Map Creation
• Technique 1– Use the co-occurrence of West topics (keynumbers)
to render the topical adjacencies of American caselaw.
•
DATASET - The dataset
consists of:
•
All top level West topics
assigned to United States
Supreme Court cases
from the 1944 term
through the end of the
2004 term and their cooccurrence.
•
7,948 unique cases to
which 19,789 topic
assignments have been
made.
•
Of the 405 topics in the
West taxonomy, 291
appear in opinions issued
by the Supreme Court for
this time period.
•
This results in 22,345
edges with 3743 unique
topic pairings.
7
Problem:
• There are three types of West top level topics:
– Procedural (green)
– Factual (red)
– Doctrinal (blue)
• The three types can co-occur in a wide variety of
cases.
• For instance, procedural topics may co-occur
with just about any factual or substantive topic.
8
Technique 2
•
Use the “Topics By Specialty”
assignments in West’s Analysis of
American Law to create a topic map.
•
Problem: Some topics are still
assigned to too many top level
substantive categories.
•
This creates more unmanageable
spaghetti.
SKETCH 1
9
Remedy
• Remove the most tenuous, multiple,
subject assignments that pull everything to
the center:
10
Technique 3:
The above image represents all topics identified
as doctrinal and assigned to law School class
subjects. These were then subjected to a double
treatment. (1) Each of the 55 classes was paired
with its most frequently occurring subjects. (2)
The graph was reduced to an edge weight
exceeding 10 case co-occurrences.
2004 Supreme Court Term West Topic Space
(Procedural Topics Removed)
11
Research Goal
Spatial navigation / visualization of bibliographic data in
which the underlying structural organization of the domain
is conveyed to the user.
12
The End
13