SCOOP - System For Collaborative Open Ontology
Production
“A Knowledge Management tool that understands the content it manages.”
AAAI-SSS AMKM26 March 2003
Adam Pease, John LiTeknowledge[apease | jli]@teknowledge.com
http://projects.teknowledge.com/RKFhttp://ontology.teknowledge.com
System Overview
Formal Ontology
Theorem Prover
Simple Workflow
Language Interface
GUI
Problem Addressed
• Developers need to know whether they– Share content with other developers– Conflict with other developers’ content
What’s New
• Using theorem-proving to maintain consistency among KEs/SMEs– Vertical Consistency – consistency
among theories used by a single developer
– Horizontal Consistency – consistency among theories created by several developers
• Building on existing SCOOP workflow, coordination and voting support
Consistency Support
• Redundancy results in warnings– Redundacy among developers triggers
suggestion for statement lifting– Lifting results in redundant statement being
moved to a common file/theory
• Contradiction results in errors– Users must retract contradiction among his own
files/theories– Users may keep contradictions with respect to
other developer’s files, at the price of prohibiting further diagnostics with respect to those files
Voting
• Majority vote• Author’s vote is a tie-breaker• Other weighted voting schemes
easily adopted
Dimensions of Knowledge
FOL
CandidateInferences/Statements
StructuredDialogs
StructuredSources
EnglishStatements
TextDocuments
Form
ality
Tim
e
Person/Source/Domain
Knowledge Compartments
Knowledge Modules
Knowledge Versions
Benefits of history
Regression to earlier versions problematic content can be removed
and the system returned to an earlier state
Change tracking track the evolution of system
components understand why things were done understand how system must be modified to
support other changes
System Overview
Formal OntologyTheorem Prover
Simple Workflow
Language Interface
GUI
SUMO
• Incorporates over 50 publicly available sources of high-level ontological content
• Available in KIF (first order logic), DAML and LOOM
• May be used without fee for any purpose (including for profit)
• Refined extensively on the basis of input from SUO mailing list participants
• Mapped by hand to all 100,000 WordNet synsets
• 47 publicly released versions created over two years (approximately 1,000 concepts, 4000 assertions, and 750 rules so far)
What’s in the SUMO?
• Structural Ontology • Set/Class Theory• Number Hierarchy• Quantities and Units of Measure• Biological Taxonomy• Temporal Concepts• Mereotopology
SUMO Structure
Structural Ontology
Base Ontology
Set/Class Theory Numeric Temporal Mereotopology
Graph Measure Processes Objects
Qualities
SUMO Sources• PSL• Enterprise Ontology• ITBM-CNR ontologies
– Unrestricted-Time– Representation– Anatomy– Biologic-Functions– Biologic-Substances– topics– meronymy– topology– topo-morphology– localization– assessment– structuring-concepts– physical-concepts– top-level– social-objects– Quantities– Actors– Positions– Natural-Kinds
• Ontolingua ontologies– component-assemblies -
Gruber and Olsen– product-ontology - Fikes– physical-quantities - Brauch– scalar-quantities - Gruber
and Olsen– unary-scalar-functions -
Gruber and Olsen– abstract-algebra authored -
Genesereth, Gruber and Olsen
– kif-extensions– kif-relations– kif-sets– frame-ontology - Mribiere– okbc-ontology - Xpetard– Standard Dimensions - U
Madrid, Spain– Simple-Time - Mribiere – Standard-Units - Loeser +
Pinto– Agents - Arnaoudova– KIF-Numbers
• Allen - temporal ontology
• Pease - Core Plan Representation
• Borgo, Guarino, and Masolo's - Formal Theory of Physical Objects
• Casati and Varzi - Theory of Holes
• Level – verb taxonomy
• Russell and Norvig - upper ontology
• Smith - Formal Theory of Fiat/Bona Fide Boundaries/Objects
• Sowa - upper ontology
• Whitten - Starter KB
Domain Specific Ontologies
• Finance and investment• Real Estate• Terrain features• Computers and Networks (Quality of Service)• Periodic table of elements• North American Industrial Coding System• ECommerce services• Terrorism and Weapons of Mass Destruction• Air force planning and operations• Army planning• Ontologies developed outside Teknowledge
– Biological viruses– Intellectual property– Linguistic elements
System Overview
Formal Ontology
Theorem Prover
Simple Workflow
Language Interface
GUI
Simple Language Example
• "The dog bites the man" • Shows both simple noun mappings
and a simple verb mapping
(exists (?D ?M ?E)
(and
(instance ?E Biting)
(instance ?D Canine)
(instance ?M MalePerson)
(agent ?E ?D)
(patient ?E ?M)))
There exists a biting event, a canine and a male person such that the canine is the agentof the biting and the male person is thepatient of the biting.
d,m,e : Biting(e) Canine(d) MalePerson(m) agent(e,d) patient (e,m)
System Overview
Formal Ontology
Theorem Prover
Simple Workflow
Language Interface
GUI
Simple Language Generation
• Term translation• Relation templates
– Use C-like printf statements
• Result is awkward but usually grammatical
• Preserves deep meaning• English (Sevcenko), German (Wulf),
Czech (Sevcenko), Italian (Ulivieri & Molino)– Hindi, Telugu, Tagalog, Chinese, Russian in
progress
Plans
• Problem resolution heuristics - Knowing which statement to retract– Metrics for statement use– Statement assertion recency– User weighting
• Proof summarization– Tree-based visibility selection, hide
transformations with one premise• Hardening
– Improved control panel– Testing– Installation packaging
Plans
• Scaleup– How to handle 10 users, 100 users
•Coordination techniques change•Refine the agenda•Avoid deadlock
– Hierarchical conflict detection• Misc
– Better voting tabulation– Pluggable voting policy– Inference-based relevance searching