USC/ISI Automatic Data Extraction Demo

Step 5: Clustering author names

We use clustering to help identify correct extracts by finding subsets of extracts that share common features. The extracts are clustered according to the set of  positional features, including adjacent landmarks, position on the page and whether the extract is visible. Next, each cluster is given a score according to how similar it is to the set of training examples. The top scoring cluster should contain the correct extracts.
 
Next: Score clusters

AUTHOR : 208 clusters identified among 1950 possible extracts
Printing 72th cluster of 8 objects
The Bluest Eye
Cold Mountain
Abducting Diana
The Diviners
The Abridged Edition
The Selfish Gene
Assembling California
Call It Sleep

Printing 81th cluster of 12 objects
A Dry White Season
The Conservationist
Tar Baby
Charming Billy
The Third Chimpanzee
The Poisonwood Bible
The Insect Societies
On Human Nature
The Meme Machine
The Century
In Suspect Terrain
Statistical Learning Theory

Printing 88th cluster of 15 objects
Nadine Gordimer
Toni Morrison
Charles Frazier
Edward Ball
Barry Unsworth
Jared Diamond
Margaret Laurence
Fannie Flagg
Edward Osborne Wilson
Richard Dawkins
Tom Brokaw
John H . Holland
Henry Roth
Tom M . Mitchell
Stuart Kauffman

Printing 89th cluster of 13 objects
by Nadine Gordimer
by Toni Morrison
by Charles Frazier
by Edward Ball
by Barry Unsworth
by Jared Diamond
by Margaret Laurence
by Fannie Flagg
by Edward Osborne Wilson
by Richard Dawkins
by Tom Brokaw
by Henry Roth
by Stuart Kauffman

Printing 122th cluster of 20 objects *** correct cluster ***
Toni Morrison
Charles Frazier
Edward Ball
Rupert Lowe
Dario Fo
Fo Dario
Barry Unsworth
Jared Diamond
Margaret Laurence
Fannie Flagg
Sarah Landry
Edward Osborne Wilson
Richard Dawkins
Tom Brokaw
John H . Holland
Henry Roth
Thomas M . Mitchell
Tom M . Mitchell
Stuart Kauffman
Jeffrey Y . Tsao

Printing 123th cluster of 20 objects
Toni Morrison
Charles Frazier
Edward Ball
Rupert Lowe
Dario Fo
Fo Dario
Barry Unsworth
Jared Diamond
Margaret Laurence
Fannie Flagg
Sarah Landry
Edward Osborne Wilson
Richard Dawkins
Tom Brokaw
John H . Holland
Henry Roth
Thomas M . Mitchell
Tom M . Mitchell
Stuart Kauffman
Jeffrey Y . Tsao

Printing 132th cluster of 18 objects
Toni Morrison
Alice Munroe
Edward O . Wilson
Susan J . Blackmore
Richard Dawkins
Richard Dawkins
Peter Jennings
Sherry Sontag
John H . Holland
John H . Holland
Melanie Michell
Abraham Cahan
Vladimir S . Cherkassky
Richard S . Sutton
John H . Holland
Peter Coveney
Isabelle Stengers
Charles A . Lindbergh

Printing 151th cluster of 20 objects
South Africa
American Historical Fiction
American First Novelists
United States
Charleston Region
South Carolina
Continental European
Human Geography
Social Science
World History
Human Evolution
Popular American Fiction
Artificial Intelligence
Earth Sciences
Regional Geology
Artificial Intelligence
United States
Humorous Stories
Juvenile Fiction
Short Stories

Printing 156th cluster of 20 objects
Christopher D . Manning
Richard S . Sutton
Janet Kolodner
Vladimir Naumovich Vapnik
Steve Bucuvalas
John H . Holland
Peter Coveney
Isabelle Stengers
James Gleick
Gunnar Odhner
Charles A . Lindbergh
Brendan Gill
Matthew Joseph Bruccoli
Michael Korda
Julie Edwards
Carson Levine
Tom Wolfe
Diane Duane
Louis Sachar
Johanna Solomon

Printing 191th cluster of 19 objects
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews
Editorial Reviews

Printing 198th cluster of 20 objects
Jared Diamond
Margaret Eleanor Atwood
Margaret Laurence
Fannie Flagg
Barbara Kingsolver
Fannie Flagg
Edward O . Wilson
Edward O . Wilson
Edward Osborne Wilson
Richard Dawkins
Tom Wolfe
Tom Brokaw
David E . Goldberg
Saul Bellow
Saul Bellow
Henry Roth
Vladimir Naumovich Vapnik
Ron Chernow
Reeve Lindbergh
Louis Sachar

Printing 206th cluster of 14 objects
List Price
List Price
List Price
List Price
List Price
List Price
List Price
List Price
List Price
List Price
List Price
List Price
List Price
List Price
 
 


Next: Score clusters



Copyright: USC Information Sciences Institute 2000