Ontologies

1
Ontologies

• Brian Matthews,
• Business Information Technology Dept, CLRC
• b.m.matthews_at_rl.ac.uk

2
Who am I?

• Group Leader, Information Science Engineering,
  Business Information Technology Department
• Advanced development of Business Information
  Systems
• Research programme in information technology and
  distributed systems
• Metadata, Trust, DIS, Workflow
• Deputy Manager, W3C Office for the UK and Ireland

3
Overview

• What are Ontologies?
• Types, examples, tools
• Semantic Web
• How we might use Ontologies.

4
What are Ontologies?

• Currently a hot topic
• Cover the gap between
• Storage of data
• Meaning of data
• Allow a shared understanding of information
• Between people and people
• Between people and machines
• Between machines and machines
• Arising out of many years of research
• Knowledge engineering, expert systems, natural
  language processing, information retrieval,
  digital libraries
• Now causing much interest again
• Data and Knowledge Management
• The Semantic Web

5
Origins

• The term Ontology comes from Philosophy
• The Science of Being
• Aristotle, Metaphysics.
• Immanuel Kant, Charles Sanders Pierce
• the science of being in general embracing such
  issues as the nature of existence and the
  categorical structures of reality
• But also an ontology is
• a set of things whose existence is acknowledged
  by a particular theory or system of thought
• - Oxford Companion to Philosophy, 1995

6
The Meaning Triangle

• Map symbols to things via concepts

refers to
evokes
Field
Symbol
Thing
stands for
Ogden Richards 1927
7
Ontologies in Computer Science

• Similarly in Computer Science an Ontology is an
  engineering artefact describing what exists in a
  particular domain.
• An ontology is a formal, explicit specification
  of a shared conceptualisation.
• Conceptualisation a model of some phenomenon in
  the world which identifies the relevant concepts
  of that phenomenon.
• Explicit the concepts used and the constraints
  on their use are explicitly defined.
• Formal the ontology should be machine
  understandable, i.e. the machine should be able
  to interpret the semantics of the information
  provided.
• Shared an ontology captures consensual
  knowledge, that is, it is not restricted to some
  individual, but accepted by a group.

8
Communication via Ontologies
Ontology description
symbol
00101100
field
formal models
ontology
Internal models

concept
thing
9
Syntax (Markup)is not enough

• ltinvestigationgt
• ltexperimentorgtCharles Darwin
• lt/experimentorgt
• lttitlegtBeagle Voyage
• lt/titlegt
• lt/investigationgt

• ltinvestigator nameCharles Darwin/gt
• ltstudy name Beagle Voyage /gt
• lt/investigatorgt

Essentially mean the same thing. - need to map
them together on a conceptual level.
10
Components of Ontologies

• An Ontology is a formal object
• Requires a formal language to describe them
• The Ontology Definition Language gives
• A VOCABULARY of shared common concepts
• A set of RELATIONSHIPS between concepts
• A set of CONSTRAINTS on the concepts
• These provide a common semantics for the area of
  interest.
• Instances of concepts then have values for the
  relations and should obey the constraints.

11
Types of Ontology

• Top-level Ontology
• Cyc (www.cyc.com), WordNet (www.cogsci.princeton.e
  du/wn)
• Domain Ontology
• UMLS (www.nlm.nih.gov/research.umls)
• Materials Microcharacterization Collaboratory
  Line Pouchard
• CLRC Metadata Format
• CERIF
• Dublin Core
• Thesauri
• A thesauri can be considered a cut down ontology
• E.g. Agrovoc (www.fao.org/agrovoc)

12
Thesauri

• A Thesaurus has
• A set of concepts arranged in a hierarchy.
• Each concept has a preferred term (a word or
  phrase) and a set of alternative terms which can
  be used for the concept.
• Each concept can refer to its broader concept and
  narrower concept in its hierarchy. Other concepts
  can be also be related to a particular concept.
• A concept which has no broader term is a top
  concept .
• A concept can have associated with it a set of
  scope notes providing information about the
  concept.
• A simplified Ontology with a broader/narrower
  concept relation.

13
Example Thesaurus

• Can use used-for term to provide alternate
  keywords
• Can refine the search by searching for narrower
  terms
• Can provide more results by using broader terms.
• Can provide related results by using siblings or
  other related terms.

animals
domestic animals
Vetinary science
pets
vets schools
hound
dog
cat
14
Further relationships

• Ontologies in general provide a richer language
  of relationships between concepts.
• Typical ones
• isa relationship - Subclass/superclass
• Thermometer is an Instrument
• part-of Meronymy
• Wheel is part of a Car
• Attribute provides a particular value
• E.g. Temperature
• Can define your own relationships
• Can define concepts via their relationships
• ColdObjects are those with a temperature less
  than 5C.
• Can reason over the relationships.

15
Example Metadata Model
16
Study Description

• The Study is the basic unit for a scientific
  activity.
• Can be further divided into
• Programmes for connected studies.
• Investigations for a single measurement,
  experiment or simulation.
• Note using UML as an Ontology Description
  Language.

17
Hierarchy of Data Holdings

• With investigations, there are associated data
  holdings.
• These are themselves arranged in a hierarchy
  data sets, and files, with links between them
• Logical organisation identity separated from
  location.

Investigation
Data Holding
Data Holding
Data Holding
Data-Set 1 (Raw)
Data-Set 2 (Inter)
Data-Set 3 (Final)
File 1 name date
File 1 name date
File 1 name date
18
OIL

• OIL (Ontology Inference Layer) , is
  representation and inference layer for
  ontologies, which unifies three important aspects
  
• formal semantics and efficient reasoning support
  as provided by Description Logics
• rich modelling primitives as provided by the
  Frame community
• syntactical exchange notations as provided by the
  Web community.
• Developed by the European Project
  On-To-Knowledge
• Tool OilEd

19
Example OIL Ontology
class-def tree subclass-of plant class-def
branch slot-constraint is-part-of
has-value tree class-def defined carnivore
subclass-of animal slot-constraint eats
value-type animal class-def defined
herbivore subclass-of animal
slot-constraint eats value-type (plant or (
slot-constraint is-part-of has-value plant))

• ontology-container
• title African Animals
• ontology-definitions
• slot-def eats
• slot-def is-part-of
• properties transitive
• slot-def weight
• range (min 0)
• properties functional
• slot-def colour
• range string
• properties functional
• class-def animal
• class-def plant
• disjoint animal plant

class-def mammal subclass-of animal
class-def elephant subclass-of herbivore
mammal slot-constraint colour
has-filler grey class-def defined
african-elephant subclass-of elephant
slot-constraint comes-from has-filler Africa
20
OilEd
21
The Semantic Web

• The Web is one HUGE data structure
• However, when you build a data structure, you
  know meaning behind it
• The Web is chaotic - why are resources are
  linked?
• Imagine a library where all the books have the
  same text on the cover, and the only catalogues
  are compiled by photocopying the books, cutting
  up the copies, and arranging the words in the
  order of frequency. Johan Hjelm
• Google is great at returning all the pages on the
  web that mention "Tim Berners-Lee
• But what about returning those pages written by
  Tim Berners-Lee?
• The Semantic Web adds well-defined meaning to
  describe the Web (Metadata).

22
Add Meaning to Resources
23
A Layered Architecture
24
Resource Description Framework (RDF)

• Knowledge representation
• Designed to make statements about web resources.
• Statements in form of triples
• (Subject , Predicate, object)
• For metadata descriptions
• Has an XML Syntax

25
RDF Schemas

• Allow simple Ontologies to be constructed
• Define new classes of concepts
• Define new properties
• Define sub-classes and sub-properties
• Define source and target of properties.

26
RDF(S) Example
27
Annotation a Semantic Web Application

• Allows user to add comments to other web sites
• And make comments on the comments
• Uses RDF Metadata

28
RSS a Semantic Web Application

• Allows site syndication - parts of websites to be
  published and distributed across different sites.
• Uses RDF Metadata

29
DAMLOIL

• DAMLOIL is an Ontology language for Web
  resources.
• Uses RDF and RDF Schema.
• Extends these languages with richer modelling
  primitives.
• Adds primitives of frame-based languages.
• Also uses XML Schema Datatypes
• Uses many of the language components of OIL.
• Clean and well defined semantics.
• DARPA Agent Markup Language (DAML) - US project,
  to develop a language and tools for the Semantic
  Web.
• www.daml.org
• DAMLOIL defines an RDF based vocabulary for
  defining ontologies on the Web.
• Now forming major input to the Web Ontology -
  OWL.

30
DAMLOIL Example

• Ontology for Weather reports
• ltdamlClass rdfID"WindEvent"gt
• ltrdfscommentgtSuperclass for all events
  dealing with windlt/rdfscommentgt
• ltrdfslabelgtWind eventlt/rdfslabelgt
• ltrdfssubClassOf rdfresource"WeatherEvent"/gt
• lt/damlClassgt
• ltdamlProperty rdfID"windDirection"gt
• ltrdfslabelgtWind directionlt/rdfslabelgt
• ltrdfsdomain rdfresource"WindEvent"/gt
• lt/damlPropertygt
• ltdamlProperty rdfID"windSpeed"gt
• ltrdfslabelgtWind speedlt/rdfslabelgt
• ltrdfsdomain rdfresource"WindEvent"/gt
• lt/damlPropertygt
• http//mnemosyne.umd.edu/aelkiss/weather-ont.daml
  

31
More on DAMLOIL

• Can add constraints on classes as in OIL.
• Web Accessible Ontology
• Resources can then refer to Ontology (anywhere on
  the Web) to declare meaning.
• Can then reason over the ontology using web
  tools

ltstudygt ltinvestigatorgtCharles Darwin
lt/investigatorgt ltstudynamegtBeagle Voyage
lt/studynamegt lt/studygt
32
Semantic Web current status

• The Semantic Web has been around several years
• Base technologies well-established
• Gone through several iterations
• Lots of academic interest
• Convincing applications are still missing
• However, many demonstrators and interesting
  applications
• Dublin Core, Thesauri, Annotations.
• CC/PP, P3P, PICS,
• Need to demonstrate the benefit of a common
  framework.

33
Standardisation

• Ontology on the Web grew in the 1990s
• 1995 - SHOE (Simple HTML Ontology Extensions),
  Univ of Maryland.
• 1996/7 - Ontobroker, Univ. of Karlsruhe
• 1997-1999 - OIL (Ontology Interchange Level),
  Amsterdam led EU project
• Spin-off from Govt Investment in SWeb Technology
  
• 1999 - The DARPA Agent Markup Language Program
  (DAML).
• 2000 - EU IST Project (Framework 5, 6)
• 2000 some US National Science Foundation funding
• proposed - govt "jumpstart" activities in Japan
  and Australia
• Standardization Efforts
• 1996- Meta-Content Format (Note)
• 1997 - W3C Metadata Activity (RDF Recommendation
  1999)
• 2000-03 - DAML 0.5 released
• 2001-03 - DAMLOIL 1.0 spec developed by "US/EU
  ad hoc Joint Committee on Agent Markup Language"
• 2001-11 - Web Ontology Working Group

34
How can Ontologies be used?

• Shared conceptualisation between people and
  communities
• Shared meaning between machines
• Community portals
• More precise searching
• Sharing concepts across domains

35
Shared Conceptualisation

• Gives a thinking tool to design the metadata
  model
• Independent of representation in e.g. XML
• Independent of representation in e.g. DB Schemas
• Allows people to consider what concepts and
  relationships are needed
• Allow people to share between them the concepts
  they are using.

36
Shared meaning between machines
ltinvestigationgt ltexperimentorgtCharles Darwin
lt/experimentorgt lttitlegtBeagle Voyage
lt/titlegt lt/investigationgt
ltinvestigator nameCharles Darwin/gt ltstudy
name Beagle Voyage /gt lt/investigatorgt
37
Community portals

• E.g. SEmantic portAL (SEAL)
• Uses Ontology to organise and search resources
• Ontology will provide associated information.

http//www.aifb.uni-karlsruhe.de/WBS
38
More Powerful Searching

• Use the relationships in the Ontology to guide
  the search
• Return authors who knows about a topic from the
  information that authors write papers about a
  topic.
• Return values with properties in

author
knows
topic
wrote
about
paper
39
Information Sharing across Domains

• Ontology Mapping

CERIF Metadata
Data Portal Metadata
40
Finally

• Ontologies form a basis for the good design of
  metadata
• Ontologies can be used as a tools for
  interchanging and querying metadata.
• An important component of the emerging Semantic
  Web.
• The Web and the Grid from e-Science to
  e-Business , EuroWeb 2002 Conference
• Oxford, 17-18 December 2002 http//www.w3c.rl.ac
  .uk/EuroWeb/