powerpoint file - Drexel University

Download Report

Transcript powerpoint file - Drexel University

An Interactive System for
CO-Citation Visualization
Xia Lin
Jan Buzydlowski
Howard D. White
Drexel University
Philadelphia, PA, USA
Co-Citation

a method for measuring the common
intellectual interest between a pair of
documents
Document 1
cites
?
Document 2

cites
Small & Griffith (in 70’s)
Later documents
A, B, C, …
Author Co-Citation

Insights into the intellectual structure of
science and scholarship through citations
over time.
Author 1
cites
?
Author 2

cites
White & Griffith (in 80’s)
Authors
A, B, C, …
Author as an Icon

An author
– represents a person;
– represents a body of writings.

A group of related authors
– comes to stand for a body of ideas
– represents subject relationships of
documents.
A Map of Information Scientists
Map Structures
Retrieval
(IR theories)
(Online retrieval)
(General)
User
(Communication)
Citation
Document
(Bibliometrics)
Data for Information Scientists Map
120 highly cited authors in
Information Science
 Co-citation count of every pair of the
120 authors
 A matrix of 120 by 120 of their cocitation counts, converted to Pearson
r’s

Data Collection

Labor-intensive process
– Decided who are highest-cited authors
in the field
– Conducted thousands of DIALOG
searches on paired authors
• C(120, 2) =7140 searches
– Processed data in a spreadsheet

White and McCain (JASIS, April 1998)
New Interactive System

The challenges
– To process data and generate the map
instantly.
– To provide interactive functions for the
viewer to explore the map and the
underlying data (with search engines).
– To provide different maps (with different
mapping algorithms).
The Database
– Institute for Scientific Information
– Arts and Humanities Database (AHCI)
• 1988 - 1997
• 1.26 million records
– BRS search engines
The Old Interface
System Structure
BRS Search
Engines
cgi
Web Interface
Web Server
Java Servlets
Java Applet
Mapping
Procedures
The New Interface
Authors co-cited most often with PLATO
Kruskal, Joseph B.
Kruskal, Clyde P.
Future Development

Adding different maps
– Pathfinders
– Multidimensional scaling
– Hierarchical clustering

Adding more interactions with search
engines
– Mining and exploration tools
– Subject labels