1 Ontology as Master Discipline of Information Science Barry Smith .
-
Upload
charles-bruce -
Category
Documents
-
view
220 -
download
0
Transcript of 1 Ontology as Master Discipline of Information Science Barry Smith .
1
Ontology as Master Discipline of Information Science
Barry Smith
http://ifomis.de
2
Ontology as Master Discipline of Information Science
Barry Smith
http://ifomis.de
Real
3
Real Ontology is a branch of philosophy
the science of what is
the science of the kinds and structures of objects, properties, events, processes and relations in reality
4
Real ontology seeks to provide a definitive and exhaustive classification of entities in all spheres of being.
5
It seeks to answer questions like this:
What categories of entities are needed for a complete description and explanation of all the goings-on in the universe?
6
Ontology is in many respects comparable to the theories produced by science
… but it is radically more general than these
7
It can be regarded as a kind of generalized chemistry or zoology (Aristotle’s ontology grew out of biological classification)
(Russell: Logic is a zoology of facts)
8
Aristotle
First ontologist
9
First ontology
(from Porphyry’s Commentary on Aristotle’s Categories)
10
Linnaean Ontology
11
Ontologies are
(very roughly)
taxonomical trees
12
Ontology is distinguished from the special sciences
it seeks to study all of the various types of entities existing at all levels of granularity
13
and to establish how they hang together to form a single whole (‘reality’ or ‘being’)
14
Unity achieved
via a good theory of relations
and via taxonomies
at different levels of granularity (atomic, molecular, cellular, organismic, etc.)
15
Sources for ontological theorizing:
thought experiments
the study of ancient texts
development of formal theories
the results of natural science
now also: working with computers
16
The existence of computers
and of large databases
allows us to express old philosophical problems in a new light
17
The problem of the unity of science
The logical positivist solution to this problem addressed a world in which sciences are identified with
printed textsWhat if sciences are identified with
Large Databases ?
18
Each family of databases
has its own idiosyncratic terms and concepts
by means of which it represents the information it receives
How to resolve the incompatibilities which result when databases need to be merged?
19
The Database Tower of Babel Problem
20
21
The term ‘ontology’came to be used by information scientists to describe the construction of standardized taxonomies designed to make databases mutually compatible
and thus to make data transportable from one environment to another
22
An ‘ontology’is a dictionary of terms formulated in a canonical syntax and with commonly accepted definitions and axioms
designed to yield a shared framework
for use by different information systems communities.
23
An ontology
is a concise and unambiguous description of the principal, relevant entities of an application domain and of their potential relations to each other
24
SO FAR
SO GOOD
25
But how was this idea in fact realized?
How did information systems engineers proceed to build ontologies?
By looking at the world, surely
Well, No
They built ontologies by looking at what people think about the world
26
27
Quine
28
For Quineans
Ontology studies, not reality,
but scientific theories
From ontology
… to ontological commitment
29
Quine:
each natural science has its own preferred repertoire of types of objects to the existence of which it is committed
30
Quineanism:
ontology is the study of the ontological commitments or presuppositions embodied in the different natural sciences
31
Quine:
only natural sciences can be taken ontologically seriously
The way to do ontology is exclusively through the investigation of scientific theories
32
Thus it is reasonable to identify ontology
– the search for answers to the question: what exists? –
with the study of the ontological commitments of natural scientists
All natural sciences are compatible with each other
33
PROBLEM
The Quinean view of ontology becomes strikingly less defensible
when the ontological commitments of various non-scientists are allowed into the mix
34
How, ontologically, are we to treat the commitments of
astrologists,
clairvoyants,
believers in voodoo?
35
How, ontologically, are we to treat the commitments of
patients who believe that their illness is caused by evil spirits or magic spells?
36
Growth of Quinean ontology outside philosophy:
Psychologists and cognitive anthropologists have sought to elicit the ontological commitments
(‘ontologies’, in the plural)
of different cultures and groups.
37
This is not ontology
Not all the things that people believe in are genuine objects of ontological investigation
Only what exists is a genuine object of ontological investigation
38
39
Why, then,
do information systems ontologists study peoples’ beliefs, thoughts, concepts
rather than the objects themselves?
40
Arguments for Ontology as Conceptual Modeling
Ontology is hard.
Life is short.
Let’s do conceptual modeling instead
41
programming real ontology into computers is hard
therefore:
we will simplify ontology
and not care about reality at all
42
Painting the Emperor´s Palace is
h a r d
43
therefore
we will not try to paint the Palace at all
... we will be satisfied instead with a grainy snapshot of some other building
44
45
Ontological engineers
neglect the standard of truth to reality
in favor of other, putatively more practical, standards:
above all programmability
46
They turn to substitutes:
to models,
to conceptualizations
because these are easier to handle
(… they move from messy noumenal reality to neatly packaged “phenomena” …)
47
For an information system ontology
there is no reality other than the one created through the system itself,
so that the system is, by definition, correct
48
Only those objects exist which are represented in the system
(constructivism)
49
Tom Gruber (1995):
‘For AI systems what “exists” is
what can be represented’
50
Ontological engineering
concerns itself with conceptualizations
It does not care whether these are true of some independently existing reality.
51
In the world of information systems
there are many surrogate world models
and thus many ontologies
52
… and all ontologies,
are equalboth good and bad,
53
ATTEMPTS TO SOLVE THETOWER OF BABEL PROBLEM
VIA ONTOLOGIES AS“CONCEPTUAL MODELS”
HAVEFAILED
54
HALF WAYTHROUGH
55
Can Real Ontology do Better?
Test Domain:
Medical Terminology
56
Example 1: UMLS
Universal Medical Language System
Taxonomy system maintained by National Library of Medicine in Washington DC
134 semantic types
800,000 concepts
10 million inter-concept relationships
57
Example 2: SNOMED
Systematized Nomenclature of Medicine
Taxonomy system maintained by the College of American Pathologists
121,000 concepts
340,000 relationships
58
SNOMED
designed to foster interoperability
to serve as a
“common reference point for comparison and aggregation of data throughout the entire healthcare process”
59
Problems with UMLS and SNOMED
Each is a fusion of several source vocabularies
They were fused without an ontological system being established first
They contain circularities, taxonomic gaps, unnatural ad hoc determinations
… several billion dollars still being wasted in the making of retrospective fixes
60
Example 3: GALEN
Two levels:
ontologically powerful model of clinical information inside the computer
plus
a range of terminological services for clinical tasks involving different coding systems, including natural language
61
Problems with GALEN
Ontology is ramshackle and has been subject to repeated fixes
Its unnaturalness makes coding slow and expensive
Coding thus far limited in extent
(surgical processes)
62
Blood
63
Representation of Blood in UMLS
Blood
Tissue
EntityPhysical Object
Anatomical StructureFully Formed Anatomical Structure
An aggregation of similarly specialized cells and the associated intercellular substance.
Tissues are relatively non-localized in comparison to body parts, organs or organ components
Body SubstanceBody Fluid Soft Tissue
Blood as tissue
64
Representation of Blood in SNOMED
Blood
Liquid Substance
Substance categorized by physical state
Body fluid
Body Substance
Substance
Blood as fluid
65
Representation of Blood in GALEN
Blood
SoftTissue
DomainCategoryPhenomenon
Blood as SoftTissue with two states:LiquidBlood and CoagulatedBlood
SubstanceTissue
GeneralisedSubstance SubstanceorPhysicalStructure
66
Plus attempts at Patients’ Ontology
based on WordNet = online lexical reference system for the English language
67
Representation of Blood in WordNet
Blood
Humor
the four fluids in the body whose balance was believed to determine our emotional and physical state
along with phlegm, yellow and black bile
EntityPhysical Object
SubstanceBody Substance
Body Fluid
Blood as humor
68
So what is the ontology of blood?
69
We cannot solve this problem just by looking at concepts
70
concept systems may be simply incommensurable
71
the problem can only be solved
by taking the world itself into account
72
This implies a view of ontology
not as a theory of concepts
but as a theory of reality
But how is this possible?
How can we get beyond our concepts?
answer: ontology must be maximally opportunistic
73
Maximally opportunistic
means:
look at concepts and beliefs critically
and always in the context of a wider view which includes independent ways to access the objects themselves
at different levels of granularity
74
Ontology must be maximally opportunistic
This means:
don’t just look at beliefs
look at the objects themselves
from every possible direction,
formal and informal
scientific and non-scientific …
75
Maximally opportunistic
means:
look at the same objects at different levels of granularity:
76
“skritek”
objects are in the worldnot all concepts correspond to objects
not all concepts are relevant to ontology
concepts are in the head
77
problem of ‘merging’ ontologies
“skritek”
“blaznivy”
78
How to solve the Tower of Babel Problem?
How to fuse these mutually incompatible ‘conceptual models’ of blood ?
By drawing on the results of philosophical work in ontology carried out over the last 2000 years
79
First Step:
A good medical domain ontology
presupposes a good formal or top-level ontology
80
Formal
part
hole
connected (spatially, causally)
substance
system
state
81
Material
organism
tissue
symptom
circulatory system
organ
is the nose an organ?
is the circulatory system an organ
82
Second step: select out the good conceptualizations
these have a reasonable chance of being integrated together into a single ontological system
• based on tested principles
• robust
• conform to natural science
83
IFOMIS
Institute for Formal Ontology and Medical Information Science
University of Leipzig
84
PARTNERS
Ontology Group, LADSEB-CNR, Padua/Trento
Institute of Cognitive Sciences and Technologies (ISTC-CNR), Rome
ONTEK Corporation
Language and Computing, Belgium www.landc.be
85
Strategy: A Network of Domain Ontologies
Material (Regional) Ontologies
Basic Formal Ontology (BFO)
86
A Network of Domain Ontologies
BFO
87
A Network of Domain Ontologies
B(Chem)O
BFO
88
A Network of Domain Ontologies
B(Med)O
B(Chem)O
BFO
89
A Network of Domain Ontologies
B(Cell)O
B(Med)O
B(Chem)O
BFO
90
A Network of Domain Ontologies
B(Gen)O B(Cell)O
B(Med)O
B(Chem)O
BFO
91
A Network of Domain Ontologies
B(Epidem)O
B(Gen)O B(Cell)O
B(Med)O
B(Chem)O
BFO
92
Ontology
like cartography
must work with maps at different scales and with maps picking out different dimensions of invariants
93
94
Thus ontology needs
95
Ontological ZoomingOntological Zooming
96
Universe/Periodic Table
animal
bird
canary
ostrich
fishontology of
biological species
ontology of DNA space
97
Universe/Periodic Table
animal
bird
canary
ostrich
fish
both are transparent partitions of one and the same reality
98
There are many compatible map-like partitions
many maps at different scales,
all transparent to the reality beyond
the mistake arises when one supposes
that only one of these partitions is a true map of what exists
99
Medical ontologies
at different levels of granularity:
cell ontology
drug ontology *
protein ontology
gene ontology *
anatomical ontology *
epidemiological ontology
100
Medical ontologies
disease ontology
therapy ontology
pathology ontology *
and also
physician’s ontology
patient’s ontology
101
Medical ontologies
and even
hospital management (billing) ontology *
* = already exists (but in a variety of mutually incompatible forms)
102
Partitions should be cuts through reality
a good medical ontology should NOT be compatible with the conceptualization of disease as:
caused by evil spirits and demons and cured by skritek
103
IFOMIS TestsUniform top-level ontology for medicine
applicable at distinct granularities
Test-case development of partial medical domain ontologies applied to:
• Standardization of clinical trial protocols
• Clinical trial Merkmal-dictionary
• Processing of unstructured patient records (www.landc.be)
104
Uniform top-level ontology for medicine
ONE YEAR
Applicable at distinct granularities (e.g. gene ontology)
FOUR YEARS
Standardization of clinical trial protocol
TWO YEARS
Clinical trial Merkmal-dictionary
TWO YEARS
Processing of unstructured patient records (www.landc.be)
THREE YEARS
105
Uniform top-level ontology for medicine
NO COMPETITOR
applicable at distinct granularities
NO COMPETITOR
Standardization of clinical trial protocol
NO SERIOUS COMPETITOR
Clinical trial Merkmal-dictionary
NO COMPETITOR
Processing of unstructured patient records
MANY COMPETITORS, BUT GOOD MEASURES OF EFFECTIVENESS
106
The End