ontosem2owl. plan of the talk ● ontosem overview ● features of ontosem ontology ● mapping...
TRANSCRIPT
Plan of the talk
● OntoSem Overview● Features of OntoSem Ontology● Mapping OntoSem2OWL● Motivation● Possible Application Scenarios
About OntoSem
● Ontological Semantics (OntoSem) is a theory of meaning in natural language. [Sergei Nirenburg and Victor Raskin,
Ontological Semantics, Formal Ontology and Ambiguity]
● Aims to extract and represent the meaning in text in a language independent form.
● It supports practical, large scale NLP applications such as MT, QA, Information Extraction, NLG.
● Supported by a constructed world model encoded in a rich Ontology. [Sergei Nirenburg and Victor Raskin, Ontological
Semantics, MIT Press, Forthcoming]
Basic Components
● Preprocessor– Converts the natural language text to Text Meaning
Representation (TMR)
● Static Knowledge Source– Ontology (language independent)– Lexicon (for each language)– Ontomasticon (to store proper names)– Fact repository (stores learnt instances of concepts
and TMRs)
Architecture of the Analyzer
Preprocessor
InputText
SyntacticAnalyzer
Text MeaningRepresentation(TMR)
Grammar:Ecology
MorphologySyntax
Lexicon andOnomasticon
Static Knowledge Resources
SemanticAnalyzer
Ontology andFact Repository
Static Knowledge Sources
● Ontology 6000 concepts
● English Lexicon 45000 entries
● Spanish Lexicon 40000 entries
● Chinese Lexicon 3000 entries
● Fact repository 20000 facts
[Sergei Nirenburg, Ontological Semantics: Overview, Presentation CLSP JHU, Spring 2003]
Text Meaning RepresentationsHe asked the UN to authorize the war.
REQUEST-ACTION-69 AGENT HUMAN-72 THEME ACCEPT-70 BENEFICIARY ORGANIZATION-71 SOURCE-ROOT-WORD ask TIME (< (FIND-ANCHOR-TIME))
ACCEPT-70 THEME WAR-73 THEME-OF REQUEST-ACTION-69 SOURCE-ROOT-WORD authorize
ORGANIZATION-71 HAS-NAME United-Nations BENEFICIARY-OF REQUEST-ACTION-69 SOURCE-ROOT-WORD UN
HUMAN-72 HAS-NAME Colin Powell AGENT-OF REQUEST-ACTION-69 SOURCE-ROOT-WORD he ; reference resolution has been carried out
WAR-73 THEME-OF ACCEPT-70 SOURCE-ROOT-WORD war
Example from[Marjorie McShane, Sergei Nirenburg, Stephen Beale, Margalit Zabludowski, The Cross Lingual Reuse and Extension of knowledge Resources in Ontological Semantics]
The OntoSem Ontology
Concept ::= root | object-or-event | property
property ::= relation | attribute | ontology-slot
Slot = PROPERTY + FACET + FILLER
Example frame from the Ontology
Concept Slot Facet Filler(s)REPLACEMENT-FOR DEFINITION VALUE “when x is a replacemt for y”
IS-A VALUE PHYSICAL-OBJECT-RELATION, EVENT-RELATIONINVERSE VALUE REPLACED-BYDOMAIN SEM EVENT, OBJECTRANGE SEM EVENT, OBJECT
Example from [P.J Beltran-Ferruz, P.A Gonzalez-Calero, P. Gervas Converting Mikrokosmos frames into Description Logics.]
Types of Slots
SLOTs are essentially PROPERTIES– ATTRIBUTE
Maps a concept or a set of concepts to values (numerical/ literals)
– RELATION Property that connects two or more concepts.
– ONTOLOGY-SLOT
Describe the ontology.
Types of FacetsVALUE
● FACET is used to restrict the values that may be stored.
● filler is the actual value
● May be instance, a Concept, literal, number
● Example:
earth
.............
number-of-moons VALUE 1
..............
[Sergei Nirenburg, Ontology Tutorial, ILIT UMBC]
Types of FacetsSEM
● Filler may be violated in certain cases.● Most commonly used Facet.● Example:
CONCEPT: EVENT
AGENT SEM ANIMAL
NATION
ORGANIZATION
PLANT
Types of FacetsRELAXABLE-TO
● Indicates “Typical violations” of the constraints listed in SEM Facets.
● Example:
CONCEPT: EVENT
AGENT SEM ANIMAL
NATION
ORGANIZATION
PLANT
RELAXABLE-TO DEITY
Types of FACETSDEFAULT
● Refers to the most frequent or expected constraint on the property
● ExamplePAY
THEME DEFAULT MONEY
TYPES OF FACETSOther FACETS...
● NOT: specifies that the given filler(s) must be excluded from the set of acceptable fillers.
● DEFAULT-MEASURE: specifies measuring unit for the numerical range that fills VALUE, DEFAULT or SEM.
● INV: Indicates that there exists an inverse property.
Fact Repository
● Stores instances of real-world facts
● Represents instances of ontological concepts.
OntoSem2OWL Motivation
● This project is investigating the feasibility of developing a system to translate ontologies and data between ontosem and OWL.
● Will facillitate sharing a rich, extensive language independent ontology with other Semantic Web applications.
● Additionally, if an OWL2OntoSem equevalent mapping can be made the OntoSem Ontology and Fact repository can be augmented by reusing existing ontologies on the Semantic Web.
Related WorkConverting Mikrokosmos frames into Description Logic
● Microkosmos Ontology: – A precursor to OntoSem
– Originally used for MT
[ Kavi Mahesh and Sergei Nirenburg, Meaning Representation for Knowledge Sharing in Practical Machine Translation
J.E Lonergan, Lexical Knowledge Engineering: Mikrokosmos Revisited]
● Propose a translation of frame based representation of Mikrokosmos to SHIQ and OWL.
[P.J Beltran-Ferruz, P.A Gonzalez-Calero, P. Gervas Converting Mikrokosmos frames into Description Logics.
P.J Beltran-Ferruz, P.A Gonzalez-Calero, P. Gervas Converting frames into OWL: Preparing Mikrokosmos for Linguistic Creativity]
Related WorkOOP, Frame Systems and DL vocabulary
OOP SYSTMES FRAME SYSTEMS DESCRIPTION LOGICS
Instance Frame instance, individual Instance, individualAttribute, instance variable Slot Role, attributeValue Filler FillerClass, type Frame, schema Class, concept
[Ora lassila, Deborah McGuiness The Role of Frame-Based Representation on Semantic Web]
Related WorkMapping Mikrokosmos to SHIQ
● Unary Predicates
Map into DL Classes
● Binary Predicate
Map into DL relation**
** check if its slot constraint??
● Special Case
Related WorkMapping Mikrokosmos concepts to DL Classes
<RECORD> <CONCEPT>CN</CONCEPT> <SLOT>IS-A</SLOT> <FACET>VALUE</FACET> <FILLER>Ci</FILLER></RECORD>(From Spencer notation of Mikrokosmos)
Class-def(primitive | defined CN
subclass-of Ci,......Cn slot-constraint1
slot-constraint2........................slot-constraintn
Information about classes and subclasses is stored in RECORDs using IS-A Slots
Related WorkMapping Mikrokosmos slot constraints to DL
<RECORD> <CONCEPT>CN</CONCEPT> <SLOT>SN</SLOT> <FACET>FACET</FACET> <FILLER>C</FILLER></RECORD>(From Spencer notation of Mikrokosmos)
Class-def(primitive | defined CN
subclass-of Ci,......Cn slot-constraint1
slot-constraint2........................slot-constraintn
Information about slot constraints is stored in RECORDs where slots are PROPERTIES
Related WorkMapping Mikrokosmos FACETs to DL
FACET DL Mapping
DEFAULT-MEASURE CN slot-constraint SN value-type CDEFAULT This information is stored in annotation.INV Handled automatically in class where the INV of this property is defined.NOT No need to handle this case. (DL only uses affirmative information)???RELAXABLE-TO This information is stored in annotation.SEM CN slot-constraint SN value-type C
Related WorkMapping Mikrokosmos ONTOLOGY-SLOTs to DL
ONTOLOGY-SLOT DL Mapping
DEFINITION This information is stored in annotation.SPANISH1 or ENGLISH1 Slot-constraint ENGLISH1 has-filler d.
Related WorkBuilding DL relations
<RECORD> <CONCEPT>SN</CONCEPT> <SLOT>SLOT</SLOT> <FACET>FACET</FACET> <FILLER>X</FILLER></RECORD>(From Spencer notation of Mikrokosmos)
Information requred for DL relations is encoded in records with ONTOLOGY-SLOTs in their SLOT field:
INVERSEslot-def SN inverses X
DOMAIN, RANGEslot-def SN domain disjoint X1.....Xnslot-def SN range disjoint X1.....Xn
MEASURED-INslot-def SN range X (treated like range)
Addional information in PROPERTYs that cannot be mapped easily is stored in CLASS-<PROPERTY-NAME>.
Application ScenariosAugmenting OntoSem FR with Semantic Web data
<foaf:Person> <foaf:name>Tim Finin</foaf:name> <foaf:firstName>Tim</foaf:firstName> <foaf:surname>Finin</foaf:surname> <foaf:nick>Tim</foaf:nick>
…………………………………………
<foaf:birthDate>1949-08-04</foaf:birthDate> <foaf:myersBriggs>ENTP</foaf:myersBriggs> <foaf:plan>http://www.cs.umbc.edu/~finin/schedule.html</foaf:plan> <foaf:publications>http://www.cs.umbc.edu/%7Efinin/cv/index.shtml#publications</foaf:publications> <foaf:weblog rdf:resource="http://ebiquity.umbc.edu/v2.1/blogger/" /> <foaf:aimChatID>timFinin</foaf:aimChatID> <foaf:mbox_sha1sum>49953f47b9c33484a753eaf14102af56c0148d37</foaf:mbox_sha1sum> <foaf:homepage rdf:resource="http://umbc.edu/~finin/"/> <foaf:depiction rdf:resource="http://umbc.edu/~finin/passport.gif"/> <foaf:phone rdf:resource="tel:+1-410-455-3522"/> <foaf:workplaceHomepage rdf:resource="http://umbc.edu/"/> <foaf:workInfoHomepage rdf:resource="http://umbc.edu/~finin/"/> <foaf:schoolHomepage rdf:resource="http://web.mit.edu/"/>
……………………………………………………
OntoSem
Fact Rep
Store FOAF data as Facts in OntoSem’s Fact Repository.
Application ScenariosReference Resolution
● Ontological-Semantics reference resolution
Not only deals with relating differnet references to the same individual in text but also mapping them to the real-world model.
● Augment OntoSem with FOAF data to resolve ambiguity in reference resolution.
[Beale S., M Mc. Shane, S.Nirenburg, Ontological Semantics Reference Resolution: Setting the Stage]
Application ScenariosReference Resolution
FOAF file Anupam Joshi
FOAF file A Joshi
OntoSem
A Joshi is an Associate Professor in the Computer Science department at UMBC.
A Joshi is a Philosophy student at RandomUniversity.
A Joshi, UMBC => Anupam Joshi
A Joshi, Random => A….. Joshi