GOL A General Ontological Language

1
GOLA General Ontological Language
Heinrich Herre Inst. of Medical
Informatics University of Leipzig
Barbara Heller Inst. of Medical
Informatics University of Leipzig
Wolfgang Degen Inst. of Theoretical
Informatics University of Erlangen
Barry Smith Department of Philosophy University
of Buffalo
2
Contents

• Aims and Motivation
• Application Scenario
• Sets, Individuals, Universals
• Basic Types of Individuals
• Basic Types of Relations
• Comparison to Upper-Level Ontologies
• Future Research

3
Aims of the Project GOL

• Development of a well-founded upper-level
  ontology
• Construction of a unified framework for modelling
  ontological structures
• Applications to the medical domain

Application Scenario
4
Application scenario Competence Network for
Malignant Lymphomas

• About 10,000 new diseases a year
• Great therapeutic progress
• Different established clinical trial groups
• Hodgkin-Lymphomas
• High-malignant Non-Hodgkin-Lymphomas
• Low-malignant Non-Hodgkin-Lymphomas
• with over 30 clinical trial protocols
• with up to 300 clinics/practitioners
• with different reference centres for diagnosis
  and therapy

5
Application oriented Goal of GOL

• Definition of ontologically based biometrical and
  medical data dicitionaries in the field of
  protocol- and guideline- based medicine
• is essential for
• Realisation of a computer-based quality
  management in clinical trial execution
• based on the harmonization of documentation
  criteria
• predefinition of processes for controlling and
  securing data and process quality

6
Information and Communication Services based on
Data Dictionaries
Participants
Clinics, Oncol. Specialists Primary
Pathologists Radio Therapists Self-help
Groups Patients Relatives
Material
Data Input Requests
Acknowledgement
I N T E R N E T
Depositions Documents
Participants Specific Query
Notification Documentation
Judgement
Notification
Information Services
Clinical Trial Centres
Reference Centres
Morbus Hodgkin Hm-NHL Lm-NHL Heterogenous Data
Bases
Data Dictionaries
Reference-RX Reference-Pathology Reference-Laborat
ories Heterogenous Data Bases
Communication Services
Data Bases
Patient Data
Patient Data
Electonical dispatch Material dispatch
(conventional)
7
Motivation I

• Every domain-specific ontology must use some
  upper-level ontology
• Standard modelling languages such as KIF, CycL,
  F-logic are confined to set-theoretical
  construction principles
• Standard classification systems in medicine such
  as GALEN, UMLS, SNOMED are not strong enough

8
Motivation II

• ClaimThere are ontological relations between
  urelements (objects, things, events ...) which
  exist independently of set-theoretical
  structures.
• We want to work with the real things
  directly not with set- theoretical substitutes

9
Ontology versus Set Theory

• The facile translation of ontological relations
  into sets removes the possibility of our gaining
  insight into reality

10
Hierarchy of Categories
Top-Category
Relation
Entity
Set
Urelement
formal
material
Universal
Individual
Topoid
Substance
Moment
Chronoid
Situoid
11
Hierarchy of Universals
Universals
Colour
Substrate
Space
Time
Shape . . .
solid
gas
fluid
12
Sets and Urelements I

• Sets
• abstract entities
• independent of space and time
• determined by their extensions
• Urelements
• not sets
• have internal structure which the membership
  relation cannot unfold

13
Sets and Urelements II

• Basic Axiom
• For every finite collection of entities there
  exists a set containing them as elements

14
Individuals and Universals I

• Individuals
• belong to the realm of concrete things
• are confined by space and time
• Universals
• abstract entities
• independent of space and time
• determined by their intensions
• are patterns of features realized by their
  instances

15
Individuals and Universals II

• Basic Axiom
• For every universal U there exists a set S which
  is the extension Ext(U) of U
• Ext(U) a a is instance of U

16
Substances

• exists in and of itself
• possesses material bulk
• occupies space
• bears qualities
• Examples
• you and me, the moon, a tennis ball, a house, a
  desk

17
Moments

• can exist only in a substance
• are dynamic
• can be lost over time
• Examplesactions, passions, a blush, a
  handshake, a thought

18
Situoids I

• are parts of the world that can be comprehended
  as a whole and do not need other entities in
  order to exist
• always imply a certain cut through reality, which
  means a certain granularity and point of view

19
Situoids II

• each situoid has associated with it a finite
  number of universals, which are (roughly) those
  universals which we need in order to grasp the
  situoid itself
• the universals associated with a situoid
  determine which material relations and
  individuals occur in it and thus which
  granularity and viewpoint it presupposes

20
Situoids III

• have a location in space and time
• frame a certain spatial region (called a topoid)
  and a certain temporal interval (called a
  chronoid)

21
Situoids IV

• Examples 1
• Johns kissing of Mary in a certain environment
• This situoid contains the substances John and
  Mary and a relational moment kiss which
  connects them. BUT we have to add a certain
  environment and further activities.
• Falling apple

22
Situoids V

• Example 2
• A part of the world capturing the life of tree
  in a certain environment. If a tree is considered
  as an organism, then the universals imply the
  viewpoint of a biologist and the granularity of
  branches, leaves, etc. (rather than electrons,
  atoms, etc.).

23
Chronoids, Topoids

• Chronoids are temporal durations
• Topoids are spatial regions having a certain
  mereotopological structure
• AssumptionChronoids and topoids have no
  independent existence, they depend on the
  situoids which they frame

24
Processes I

• are constituents of situoidsA configuration C
  in the situoid S is defined as some result of
  taking a collection of substances and other
  individuals occurring in S and adding moments and
  material relations from S which serve to glue
  them together

25
Processes II

• are sequences of configurations
• Example 1 Football match
• Every football match is a sequence of
  configurations of 22 players and 1 ball within a
  suitable situoid and during a time interval of
  about 120 minutes (including the break)

26
Processes III

• Example 2
• An individual case of malaria is a concrete
  process realized by a sequence of configurations
  containing a person (a substance) within a
  situoid and certain changing moments associated
  with the disease.

27
Material Relations

• are individuals with the power of connecting
  entities
• Exampleskisses, contracts, conversations

28
Refined Theory of Relations I

• A relator is an individual connecting entities. A
  relator which has substances as arguments is of
  1st order (these are exactly moments) A relator
  is of (n1)st order if the heighest of the
  relators it relates es equal n
• AxiomAt least one of the arguments of a relator
  is an individual

29
Refined Theory of Relations II

• Let Rel be the class of all relators, and r,s be
  relators. r lt s (s is stronger than r) iff r is
  among the arguments of s
• AxiomThe ordering 'lt ' does not contain an
  infinite chain r1 lt r2 lt ...lt rn lt...

30
Refined Theory of Relations III

• Relations ( Formal relations, mediated relations)
• Examples.

31
Refined Theory of Relations IV

• Examples

32
Refined Theory of Relations V

• Hierarchy of relations
• Relations(formal, mediated)
• Universals Relator-Universals.

33
Basic Relations I

• x ? y (membership relation)
• x lt y (part-of relation)
• lt (x,y,z) (relativized part-of)
• x y (instantiation)
• i (x,y) (inherence)
• x y (framing)
• x y (containment)

34
Basic Relations II

• x y (framing)
• x c y (containment)
• o (x,y) (location)
• h (x,y1, ..., yn) (holding)
• a (x,y) (association)

35
Comparison to KIF

• Basic Ontology of KIF
• Most general category is a object
• A set is a collection of objects
• An individual is any object which is not a set
• Functions and relations are finite sets of lists

36
Comparison to Russell-Norvig
37
Comparison to Sowas UO
38
Comparison to LADSEB UO
39
For more information about GOL please contact

• www.ontology.uni-leipzig.de
• Secretary Birgit Binder
• Tel. [phone]
• Fax [phone]
• E-mail binder_at_IMISE.uni-leipzig.de
• Postal adress Institute of Medical Informatics
• University of Leipzig
• Liebigstr. 27
• 04103 Leipzig, Germany