ConceptNet - a pratical commonsense reasoning tool-kit

1
ConceptNet - a pratical commonsense reasoning
tool-kit

• H Liu and P Singh
• MIT Media Lab
• Speaker Yi-Ching(Janet) Huang

2
Introduction

• ConceptNet
• Freely available commonsense knowledge base
• Natual-language-processing tool-kit
• It supports many practical textual-reasoning
  tasks over real-world documents

3
Outline

• Comparison of ConceptNet, Cyc, and WordNet
• History, Construction and Structure
• Various contextual reasoning tasks
• Quantitative and Qualitative Analysis
• Conclusion

4
Comparison
Database content Resource Capabilities
ConceptNet (2002) Commonsense OMCS (from the public) (automatic) Contextual inference
WordNet (1985) Semantic Lexicon Expert (manual) Lexical categorisation word-similarity
Cyc (1984) Commonsense Expert (manual) Formalized logical reasoning
5
History of ConceptNet
1984
2000
2002
2004
6
Building ConceptNet

• 3 phases
• Extraction phase
• Extract from OMCS corpus
• English sentence -gt binary-relation assertion
• Normalization phase
• Relaxation phase
• Produce inferred assertion
• Improve the connectivity of the network

7
Structure of the ConceptNet knowledge base

• 1.6 million assertions (1.25 million are k-lines)
• twenty relation-types

8
(No Transcript)
9
Practical commonsense reasoning

• An integrated natural-language-processing engine
• MontyLingua
• Text document --gt VSOO frames
• Reasoning capabilities
• Node-level reasoning
• Document-level reasoning

10
Node-level reasoning

• Contextual neighborhoods
• Spreading activation
• Analogy-making
• Projection

11
Document-level reasoning

• Topic-gisting
• Disambiguation and classification
• Novel-concept identification
• Affect sensing

12
Characteristics and quality

• ConceptNets reasoning abilities hinge largely on
  the quality of its knowledge

13
Characteristics of the KB

• The histogram of nodal word-lengths

70
14
Characteristic of the KB

• Average frequency an assertion is uttered of
  inferred

90 uttered
15
Characteristics of the KB

• The connectivity of nodes in ConceptNet by
  measuring nodal edge-density

16
Quality of the knowledge

• Two dimensions of quality of ConceptNet, rated by
  human judges

17
Applications of ConceptNet
Commonsense Investing
ARIA
OMAdventure
AAA
GOOSE
Commonsense Predictive Text Entry
LifeNet
GloBuddy
Metafor
Emotus Ponens
MAKEBELIEVE
Overhear
SAM
Bubble Lexicon
What Would They Think?
18
Commonsense ARIA

• Analyize E-mails content and suggest the related
  photos

19
Emotus Ponens
20
MakeBelieve
21
Conclusion

• ConceptNet is presently the largest freely
  commonsense database
• Support many practical textual-reasoning tasks
• Goodness
• Easy to use
• Simple structure of WordNet
• Good for practical commonsense reasoning