containing the semantic explosion
DESCRIPTION
Presentation by Jaimie Murdock, Cameron Buckner and Colin Allen at PhiloWeb 2012 (WWW 2012), Lyon, France.TRANSCRIPT
Containing the Semantic Explosion
Jaimie MurdockCameron Buckner
Colin Allen
Indiana Philosophy Ontology (InPhO) ProjectIndiana University – Bloomington, IN, USA
17 April 2012
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 1 / 33
Introduction
Table of Contents
1 Introduction
2 Building Ontologies
3 Dynamic Ontology
4 Future Directions
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 2 / 33
Introduction
The Information Explosion
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 3 / 33
Introduction
The Information Explosion
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 4 / 33
Introduction
Guiding Principles
“Augmented intelligence”Let humans and computers each doonly what they do best
“Guided serendipity”Provide useful tools for scholars:
cross-referencing
semantic search
document classification
visualizations
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 5 / 33
Introduction
Guiding Principles
“Augmented intelligence”Let humans and computers each doonly what they do best
“Guided serendipity”Provide useful tools for scholars:
cross-referencing
semantic search
document classification
visualizations
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 5 / 33
Introduction
What Philosophers Know
Stanford Encyclopedia ofPhilosophyhttp://plato.stanford.edu
1522 authors
127 subject editors
1315 articles since 14 Sep 1995
1 million weekly articledownloads
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 6 / 33
Introduction
What Philosophers DON’T Know
What’s in the SEP!http://plato.stanford.edu
16.3 million words — equivalent to27,250 pages at 600 words/page
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 7 / 33
Introduction
The Indiana Philosophy Ontology (InPhO) Project
https://inpho.cogs.indiana.edu/
Pragmatic attempt to organize the discipline of philosophy through a threestep process of:
data mining
expert verification
machine reasoning
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 8 / 33
Introduction
The Indiana Philosophy Ontology (InPhO) Project
https://inpho.cogs.indiana.edu/
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 9 / 33
Introduction
The Indiana Philosophy Ontology (InPhO) Project
https://inpho.cogs.indiana.edu/
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 10 / 33
Building Ontologies
Table of Contents
1 Introduction
2 Building Ontologies
3 Dynamic Ontology
4 Future Directions
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 11 / 33
Building Ontologies
Computational Ontology
“A body of formally represented knowledge is based on aconceptualization: the objects, concepts, and other entities that arepresumed to exist in some area of interest and the relationships that holdamong them . . . A conceptualization is an abstract, simplified view of theworld that we wish to represent for some purpose . . . An ontology is anexplicit specification of a conceptualization.” — Gruber (1993)
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 12 / 33
Building Ontologies
Domain Ontologies
GOLD: General Ontology for Linguistic Description
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 13 / 33
Building Ontologies
Domain Ontologies
Gene Ontology
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 14 / 33
Building Ontologies
Formal Ontology
Goals: interoperability, permanence, precision
ULOs represent the most basic, enduring features of reality — notjust a “conceptualization”
Upper-level ontologies (ULOs) unite domain-level ontologies
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 15 / 33
Building Ontologies
Formal Ontology
Concerns
“Double experts” required
Manual design and change
Bias, controversy
Battlefield cluttered with ULOs . . .
BFO, Dolce, DnS, SUMO, Cyc, GFO, IDEAS, GOL, BWW, COSMO,OCHRE, Gist, OBO, PROTON, IFF, MSO, Sowas, . . .In civil war, population suffers most . . .
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 16 / 33
Building Ontologies
Formal Ontology
Concerns
“Double experts” required
Manual design and change
Bias, controversy
Battlefield cluttered with ULOs . . .
BFO, Dolce, DnS, SUMO, Cyc, GFO, IDEAS, GOL, BWW, COSMO,OCHRE, Gist, OBO, PROTON, IFF, MSO, Sowas, . . .In civil war, population suffers most . . .
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 16 / 33
Building Ontologies
Folksonomy
Can we induce an ontology from tagging behavior?
“Wisdom of the crowds”
Del.ico.us, Flickr, Wikipedia
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 17 / 33
Building Ontologies
Folksonomy Concerns
Uncontrolled vocabularyAmbiguity, idiosyncrasy, no definitions
Tagging behavior changes over time
Unstructured or training required
The world is not flat — expertise matters!
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 18 / 33
Building Ontologies
Formal vs. Dynamic Ontology
Formal Ontology
Large projects in business,medicine, and natural sciences
Stable consensus
Manual construction
Monolithic perspective
Dynamic Ontology
Smaller, open-access projects influid domains
Emergent consensus
Semi-automated construction
Multiple perspectives
Wish: Top-down quality for a bottom-up price!
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 19 / 33
Building Ontologies
Formal vs. Dynamic Ontology
Formal Ontology
Large projects in business,medicine, and natural sciences
Stable consensus
Manual construction
Monolithic perspective
Dynamic Ontology
Smaller, open-access projects influid domains
Emergent consensus
Semi-automated construction
Multiple perspectives
Wish: Top-down quality for a bottom-up price!
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 19 / 33
Dynamic Ontology
Table of Contents
1 Introduction
2 Building Ontologies
3 Dynamic Ontology
4 Future Directions
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 20 / 33
Dynamic Ontology
The Indiana Philosophy Ontology (InPhO) Project
https://inpho.cogs.indiana.edu/
Pragmatic attempt to organize the discipline of philosophy through a threestep process of:
data mining
expert verification
machine reasoning
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 21 / 33
Dynamic Ontology
1. Data Mining
Use statistical techniques to make predictions about content
J-measure (n-grams)
Measures similarity usingco-occurrences
Entropy (TF-IDF)
Measures generality based onprevalence
Further details: Niepert et al. 2007, Murdock et al. 2012
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 22 / 33
Dynamic Ontology
1. Data Mining
Use statistical techniques to make predictions about content
J-measure (n-grams)
Measures similarity usingco-occurrences
Entropy (TF-IDF)
Measures generality based onprevalence
Further details: Niepert et al. 2007, Murdock et al. 2012
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 22 / 33
Dynamic Ontology
2. Expert Verification
Present hypothetical relations to users.
Allows for untrained ontologists.
Amazon Mechanical Turk
Further details: Allen et al. 2008, Niepert et al. 2009, Buckner et al. 2010, Eckart et al. 2010
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 23 / 33
Dynamic Ontology
3. Machine Reasoning
Input – Expert feedback combined with statistical dataOutput – Populated ontology with taxonomic projection
Further details: Niepert et al. 2008
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 24 / 33
Dynamic Ontology
Applications
Automated cross-referencing
Context-aware search
Citation management
Document classification
Visualizations
Change and controversydiscovery
“Macroscopes” (Borner)
Metaphilosophy
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 25 / 33
Dynamic Ontology
Applications
Automated cross-referencing
Context-aware search
Citation management
Document classification
Visualizations
Change and controversydiscovery
“Macroscopes” (Borner)
Metaphilosophy
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 25 / 33
Future Directions
Table of Contents
1 Introduction
2 Building Ontologies
3 Dynamic Ontology
4 Future Directions
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 26 / 33
Future Directions
CodEx Workbench
Dynamic ontology toolkit for arbitrary corpora
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 27 / 33
Future Directions
CodEx Workbench
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 28 / 33
Future Directions
CodEx Workbench
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 29 / 33
Future Directions
CodEx Workbench
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 30 / 33
Future Directions
Guiding Principles
“Augmented intelligence”Let humans and computers each doonly what they do best.
“Guided serendipity”Provide useful tools for scholars:
cross-referencing
semantic search
document classification
visualizations
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 31 / 33
Future Directions
Guiding Principles
“Augmented intelligence”Let humans and computers each doonly what they do best.
“Guided serendipity”Provide useful tools for scholars:
cross-referencing
semantic search
document classification
visualizations
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 31 / 33
Future Directions
Acknowledgements
IUB New Frontiers in the Humanities
NEH Digital Humanities Initiative
NEH-DFG Bilateral Digital Humanities Program
Digging into Data Challenge
Alexander von Humboldt Foundation
Views expressed are the authors, and not necessarily the views of any funding angencies
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 32 / 33
Future Directions
Acknowledgements
Project Director
Colin Allen
Project Founders
Cameron Buckner, Mathias Niepert
Graduate Students
Jaimie Murdock, Robert Rose
Former Staff
Evan Boggs, Tarun Gangwani, ScottWeingart
Project Staff
Alex Frost, Wesley Pettyjohn, SamWaggoner
CogSci Support
Ruth Eberle, Nubli Kasa
Application Consultants
Ed Zalta & Uri Nodelman (SEP),Tony Beavers (Noesis)
Murdock, Buckner, Allen (IU) Containing the Semantic Explosion 17 April 2012 33 / 33