Ontology Development

1

• Ontology Development
• To Support
• AQUA Question-Answering
• Richard Fikes
• Jessica Jenkins Bill MacCartney Rob McCool
  Deborah McGuinness
• Knowledge Systems Laboratory
• Stanford University
• www.ksl.stanford.edu

2
KSL and the WMD Coalition

• Tools for ontology creation, evolution, and
  maintenance
• Coalition teams have adopted KSLs Ontolingua and
  Chimaera as a standard for ontology development,
  maintenance, and analysis
• (Additionally, we use other internal tools
  JTP, DQL, IW)
• Initial evaluation and ongoing support
• KSL evaluated initial stage of WMD ontology using
  our Chimaera tools, reviewed findings with other
  teams, and taught others how to use the tools
  themselves

3
KSL and the WMD Coalition

• Tools for ontology creation, evolution, and
  maintenance
• Initial evaluation and ongoing support
• Knowledge representation and consulting work
• KSL providing some core new KB development for
  CNS core Russian naval facilities and Newly
  Independent States facilities and is augmenting
  this with extraction results
• Providing consultation on sources / merging
  opportunities -Counter-terrorism KBs built for
  DARPA (HPKB, ISX-Cyladian- HORUS, SAIC-Cycorp,
  ) semantic web for the military (SWMU ontology
  tutorial), ontologies for information fusion, etc.

4
KSL and the WMD Coalition

• Tools for ontology creation, evolution, and
  maintenance
• Initial evaluation and ongoing support
• Knowledge representation and consulting work
• Knowledge extraction
• KSL knowledge extraction tools for RNF and NIS
• New focus on utilizing other important useful KB
  sources
• SUMO is a core ontology for ontology sharing
  3,900 axioms relations and sets processes and
  objects temporal, spatial, and mereological
  relations agents, etc.
• Domain ontologies WMDs, terrorism, biological
  viruses,
• Written in SUO-KIF, a proprietary dialect of KIF
• Published by Teknowledge under GNU public license
  as part of IEEE SUO working group
  (ontology.teknowledge.com/)

5
SUMO

• SUMO requires translation to be used with any
  reasoner
• KSL has successfully translated SUMO to
  plain-vanilla KIF
• Full translation complete semantic content
  retained
• Highly portable should be fully compatible with
  most FOL reasoners
• Accurate most test queries demonstrably
  answerable from translated axioms
• However, the result is not yet fully usable
• Most test queries not answered from full axiom
  set in reasonable time
• SUMO was not designed for efficient automated
  reasoning
• Solution a smarter translator, and some SUMO
  brain surgery
• Translation to DAML may provide another path
• The existing translation is quite lossy, but
  enables some query-answering
• Further work will enable a more complete and
  accurate translation

6
DAML Versions of SUMO, WMD, and Terrorism

• DAML translations of SUMO, WMD, and 5
  terrorism-related ontologies and knowledge bases
  (KBs) provided by Teknowledge

7
DAML Versions of SUMO, WMD, and Terrorism

• A few simple translations were used by
  Teknowledge to translate original KIF content to
  DAML
• Example subrelation

lObjectProperty
(subrelation father parent)

• Example KIF triples to RDF triples

rdfresourceSpain/
(part MadridSpain Spain)
8
DAML Versions of SUMO, WMD, and Terrorism

• Teknowledges DAML files provide a great starting
  point, but there are a few problems
• Syntactic errors and issues with resolving
  references across files -- these problems are
  easy to fix.
• A large amount of the original KIF content was
  dropped in the translation to DAML.
  Reincorporating some of this content is trivial,
  but it is generally nontrivial.

Example (trival) TransitiveRelation


(instance part TransitiveRelation)
9
KIF - DAML Example (nontrivial)
Original KIF
( (and (instance ?SUBSTANCE BiochemicalAgent) (
possesses ?AGENT ?SUBSTANCE)) (capability
BiochemicalAttack agent ?AGENT))
Translation of capability ternary to binary
relation
rdfIDcapabilityRole rdfresourceCapability/ rdfresourceCaseRole/ operty rdfIDcapabilityProcess


Property
perty
10
KIF - DAML Example (nontrivial) cntd.
Original KIF
( (and (instance ?SUBSTANCE BiochemicalAgent) (
possesses ?AGENT ?SUBSTANCE)) (capability
BiochemicalAttack agent ?AGENT))
Translation of ?AGENT has BiochemicalAttack
capability if it possesses a BiochemicalAgent
ability rdfresourcesumoagent/
Attack/ rdfIDAgentsWithBiochemicalAttackCapability

y/ micalAttackAgentCapability/


rdfresourcesumopossesses/


11
Query-Answering Example 1

• What has the capability of being the agent of a
  biochemical attack?
• Query pattern (capability ?agt
  Biochemical-Attack-Agent-Capability)
• Knowledge in the ontology
• A thing is an Agent-With-Biochemical-Attack-Capab
  ility if and only if it
• Has a capability Biochemical-Attack-Agent-Capabil
  ity or
• Possesses a Biochemical-Agent
• An Agent-With-Biochemical-Attack-Capability has
  capability Biochemical-Attack-Agent-Capability
• A Nerve-Agent is a Biochemical-Agent
• If AGT has capability Biochemical-Attack-Agent-Ca
  pability, then AGT is capable of being an
  agent in a Biochemical-Attack
• If C is the capability of playing role R in
  processes of type PT, and AGT is known to have
  played role R in a process of type PT, then AGT
  has capability C
• Knowledge from documents
• Al-Qaida is a Foreign-Terrorist-Organization
  that possesses a Nerve-Agent
• Aum-Supreme-Truth-Chemical-Attack-27-Jun-94 is a
  Chemical-Attack whose agent is
  Aum-Supreme-Truth

12
Query-Answering Example 1

• What has the capability of being the agent of a
  biochemical attack?
• Query pattern (capability ?agt
  Biochemical-Attack-Agent-Capability)
• Answer Al-Qaida
• Al-Qaida is a Foreign-Terrorist-Organization
  that possesses a Nerve-Agent from
  documents
• A thing is an Agent-With-Biochemical-Attack-Capab
  ility if and only if it
• Has a capability Biochemical-Attack-Agent-Capabil
  ity or
• Possesses a Biochemical-Agent from
  ontology
• An Agent-With-Biochemical-Attack-Capability has
  capability Biochemical-Attack-Agent-Capability
  from ontology
• If AGT has capability Biochemical-Attack-Agent-Ca
  pability, then AGT is capable of being an
  agent in a Biochemical-Attack from ontology
• A Nerve-Agent is a Biochemical-Agent
  from ontology

13
Query-Answering Example 1

• What has the capability of being the agent of a
  biochemical attack?
• Query pattern (capability ?agt
  Biochemical-Attack-Agent-Capability)
• Answer Aum-Supreme-Truth
• Aum-Supreme-Truth-Chemical-Attack-27-Jun-94 is a
  Chemical-Attack whose agent is
  Aum-Supreme-Truth from documents
• Playing the role agent in a Biochemical-Attack
  requires the capability Biochemical-Attack-Agent
  -Capability from ontology
• If playing role R in a process of type PT
  requires capability C, and Agt plays role R in a
  process of type PT, then Agt has capability C
  from ontology
• Aum-Supreme-Truth has capability
  Biochemical-Attack-Agent-Capability

14
Query-Answering Example 2

• Who are the agents of attacks that used the same
  type of weapons as Recent-Attack-001?
• Query pattern (type Recent-Attack-001 ?res)
  (onProperty ?res instrument) (hasClass ?res
  ?inst-type) (type ?attack ?res) (agent ?attack
  ?agt)
• Must-bind variables ?agt ?attack
• Knowledge in the ontology
• A Mortar-Attack has an instrument of type
  Mortar
• Knowledge from documents
• Recent-Attack-001 is a Thing that has an
  instrument of type Mortar
• Revolutionary-Armed-Forces-Of-Colombia-Mortar-Att
  ack-1-Jul-00 is a Mortar-Attack that has agent
  Revolutionary-Armed-Forces-Of-Colombia.
• Answer Revolutionary-Armed-Forces-Of-Colombia

15
AQUA Program Plan

• Overview of the project
• Goal is to create a system that can answer
  complex questions
• With plus up funding, we now have an end-to-end
  system. Makes use of KSLs Ontolingua Knowledge
  Server and Java Theorem Prover (JTP) to develop
  answers to queries
• Uses SAIC and other technology to automatically
  populate KBs with information from new text
  sources
• Uses multiple extractors from multiple sources to
  answer queries
• KSL extractor
• UMBC/NMSU extractor
• IBM extractor

16
AQUA Current Plans
NL QUESTION
MOQA Text ? TMR Translator
A
SAIC TMR ? KIF Mapper/Translator
C
MOQA NL ?TMR Query Processor
B
CNS TEST DATA
IBM Text ? KIF Translator
KIF-Formatted Question
Ontolingua Knowledge Server ---------------- JAVA
Theorem Prover
KIF Answer/ Proof tree
KSL Extractor
D
E
SAIC KIF ? TMR Mapper/Translator
KSL generated explanation
MOQA TMR ? NL Answer Processor
NL ANSWER
17
AQUA Initial Concept
SAIC Interlingua? KIF Translator
NMSU Query Processor
QUESTION
Interlingua Query
NL Query
KIF Query
Interlingua Answer
KIF Answer
NL Answer
SAIC KIF ? Interlingua Translator
NMSU NL Generator
ANSWER
KSL Java Theorem Prover
18
Key Tasks - SAIC

• Perform translation of Onyx/UMBC extracted TMRs
  to KIF (Item A)
• Align two disparate ontologies
• Translate terms once aligned
• Both formalized queries and extracted text need
  to be translated
• Develop CNS WMD ontology
• Co-ordinate subcontractors and develop system
  interfaces

19
Key Tasks - Onyx

• Provide formalized translation of NL queries
  (MOCA item B)
• Perform extraction of CNS data into text (MOCA
  item A)

20
Key Tasks - IBM

• Assist in relations extraction from text into WMB
  ontology

21
KSLs Current Activities

• JTP Hybrid reasoning for query answering
• Includes a temporal reasoner
• Is a DQL (DAML Query Language) server
• Knowledge Base Partitioning Enabling Q-A from
  large scale KBs using parallel heterogeneous
  reasoners
• Inference Web Providing understandable
  explanations for derived query answers
• Knowledge extraction from semi-structured
  documents
• Tables, lists, outlines, property-value pairs,
  etc.

22
SAIC Current Activities

• SAIC
• In-house Ontolingua server with JTP now installed
  and in use in development efforts
• Ontology is available as part of demonstration in
  the demo rooms
• Please visit the SAIC/KSL demo stand

23
SAIC Current Activities (cont.)

• SAIC spearheading a federation of a WMD ontology
  development effort, assisted by Stanford KSL
• Begun development of CNS ontology. Ontology is
  currently 700 terms and viewable in our in-house
  version of Ontolingua. (Demo available)

24
SAIC Current Activities (cont.)

• Discussions underway with Sergei to put Onyx
  under subcontract to SAIC. Subcontract to go out
  as soon as possible.
• Labor division is defined and agreed to
• Major issue Due to subcontract issues Onyx is
  still not under subcontract. This affects Q?A
  ayatem development rates as this task is on the
  critical path for system development.
• Distributed ontology to KSL and IBM.
• Development of the ontology is critical in order
  to allow the extractors to function appropriately

25
WMD Ontology Creation Initial -Confederation
Assignments

• Stanford/KSL NIS-Facilities (439 terms) and
  Russian-Naval-Facilities (365 terms)
• IBM MPT-Topic (771 terms)
• Xerox-Parc Missiles-Topic (765 terms)
• Tecknowledge NIS-Nuclear-Weapons-Aggregate (219
  terms)
• Battelle Nuclear-Safety-Assistance (36 terms)

26
Year Two Project Goals

• Complete CNS ontology development
• Participate in TREC
• System is still immature
• Novel appoach
• Significant potential for further development
• Refine interfaces and determine system metrics
  to ensure maximum performance in future system
  iterations

27
TREC participation

• SAIC is signed up for TREC participation this
  year.
• A multi-pronged approach is possible with the
  current architecture
• With the SAIC/Onyx route and NL interface, gives
  the initial capability for an end-to-end system
  with restricted domain and range
• Formatted queries possible for IBM extraction
• System will be very immature in year 1 and likely
  achieve poor TREC scores, but will mature in
  multiple and novel directions over time

28
Future Plans

• Continue multi-pronged approach (running multiple
  extractors over a uniform KB)
• Plan further enhancements (Possibly add more
  extractors or reasoners
• Leverage multiple KB approach to optimize
  research in multi-partition reasoning
• Develop effective metrics to determine efficacy
  of this approach and which pathways are optimal

29
Future plans (Cont)

• Work on implementing latter Proof tree to NL
  mocha interface in the future (Reverse TMR to
  KIF)
• Transition from KIF to DAML format where possible
• Extend range and capabilities of question
  answering. Initial participation will be limited
  in terms of domain and range of questions.