1. historical account

sentencedebonairΚινητά – Ασύρματες Τεχνολογίες

10 Δεκ 2013 (πριν από 3 χρόνια και 6 μήνες)

59 εμφανίσεις

Joost Breuker

Leibniz Center for Law

University of
Amsterdam

Ontology,
ontologies

and ontological
reasoning

1. historical account

Overview


Philosophical roots (
-
500


now)


Ontology as metaphysics


Formal Ontology


Lingua
Univeralis

Philosophica

(17
th



18
th

century)


AI and knowledge engineering


80
-
ies: Naïve physics manifesto


90
-
ies: Knowledge acquisition


2
nd

millennium: Semantic Web

Ontology as part of philosophical meta
-
physics


”the theory or study of
being as such
;
ie

of the
basic characteristics of all
reality

(
Encyclopedia

Brittanica
)


Greek roots:


Heraclitus
--
> Aristotle
(categories;
universalia
)


physis
/logos: tangible/intangible (concrete/abstract)


Plato: ideas (forms) shape our understanding of the changing
reality


See:
John Sowa, Knowledge Representation, Brooks/Cole, 2000,
chapter 2


Aristotle,Kant
, Peirce, Husserl, Brentano, Heidegger, …


A synthesis

Aristotle’s categories by Brentano

Kant’s categories

Necessity

Community

Limitation

Totality

Existence

Causality

Negation

Plurality

Possibility

Inherence

Reality

Unity

MODALITY

RELATION

QUALITY

QUANTITY

Sowa’s synthesis as a lattice

Formal ontology: philosophical views in ontological engineering


Formal = axiomatically defined terms


Active community


Applied Ontology Journal (IOS
-
Press)


FOIS bi
-
annual conferences
(
www.formalontology.org
/)


Association
(
www.iaoa.org/
)


Focus on top
-
ontologies


(meta
-
)properties rather than concepts


Typical examples:


Sowa’s


SUMO (IEEE)


DOLCE (Nicola
Guarino
)


BFO (Barry Smith)

And SUMO

Top of DOLCE

BFO (Basic Formal Ontology)

Continuant

Occurrent

(Process)

Independent

Continuant


(
molecule,

cell, organ,

organism
)


Dependent

Continuant


(
quality,

function,

disease
)

Functioning


Side
-
Effect,

Stochastic

Process, ...

..... ..... .... .....

10

(Meta)
-
properties
vs

entities


Abstract
vs

Concrete (Sowa,
CyC
)


Is a thought, a process, a circle abstract? tangible?


Occurrent

vs

Continuant (DOLCE, BFO)


About
process

vs

object


Process: time
-
perspective (duration)


Object: space
-
perspective (position)


However:


Time: processes are as `continuant’ as objects (
eg

gravitation is inherent to matter (objects))


Both processes and objects have `life
-
cycles’/duration


Space: processes are localized in/by objects


Even `fields’ spread from some object (source)


Occurrent
/continuant cannot be a dichotomy


Paradox
-
1


…these top
ontologies

have been reused
extensively as the basic structure of many
domain
ontologies


Eg

BFO (Smith,
Ontobra

2008)








AstraZeneca
-

Clinical Information Science

BioPAX
-
OBO

BIRN Ontology Task Force (BIRN OTF)

Computer Task Group Inc.

Duke University Laboratory of Computational Immunology

Dumontier

Lab

INRIA Lorraine Research Unit

Kobe University Graduate School of Medicine

Language and Computing

National Center for Multi
-
Source Information Fusion

Ontology Works

Science Commons
-

Neurocommons


University of Texas Southwestern Medical Center

13

Groups and Organizations using BFO



BioTop
: A Biomedical Top
-
Domain Ontology

Common Anatomy Reference Ontology (CARO)

Foundational Model of Anatomy (FMA)


Gene Ontology (GO)

Infectious Disease Ontology

Ontology for Biomedical Investigations (OBI

Ontology for Clinical Investigations (OCI)

Phenotypic Quality Ontology (
PaTO
)

Protein Ontology (PRO)

RNA Ontology (
RnaO
)

Senselab

Ontology

Sequence Ontology (SO)

Subcellular

Anatomy Ontology (SAO)

Vaccine Ontology (VO)


14

Paradox
-
1 explained??


…these top
ontologies

have been reused
extensively as the basic structure of many
domain
ontologies

1.
They are not really used in an axiomatic way!

2.
The pre
-
structuring is very thin


This philosophical heritage is


confusing, and


based on false parenthood:

Ontology

is concerned with reality and existence;
ontologies

with knowledge…











However, there are far more interesting roots of
ontologies

in philosophy than in its
metaphysical reflections


Artificial Universal Language (
lingua
univeralis

philosophica
)


Middle age:
eg

Ockham


`Babel’: many languages as a problem


Back to paradise: a universal language


Babel is due to the fact the semantics do not
transpire in the word
-
image or syntax


This universal language should focus on the universal
semantics

of thought (
vs

Latin, Esperanto, …)


A revival in the 17
th

century



John Locke (1632
-
1704) on the language delusion

17
th

century philosophers


many proposals, but most noteworthy:


John Wilkins


Gottfried Leibniz


For more see: J.L. Borges, The analytic language of John Wilkins; Umberto Eco (1995)
The search for the perfect language; Steve Pinker (1994) Words and Rules


The first ontological engineer: John Wilkins (1614
-
1672)

The first ontological engineer: John Wilkins (1614
-
1672)


Basic assumptions:


`to repair the ruins of Babel’


There is a limited number of primitive categories (elements)


Other concepts are combinations of primitives (molecules)


--
> taxonomy/lattice


Classified about 2000 concepts


40 top categories and many subcategories


A word is constructed by assigning a fixed
syllable/letter to a string in descending this `graph’


Eg
:
Zita

= animal (Z) + beast (
i
) + canine (
t
) + dog (a)



NB: the basis for `Roget’s thesaurus’ (1852, now)


An example: on measurement

distinguishing `count/`mass’ nouns…

…even including a definitions of NIL and of concept


Nihil
:

“whatever can be named but cannot be thought”


Concept:

“thought in so far as it is a thought of something”





Gottfried Leibniz (1646
-
1716)


Lingua
Characteristica

Universalis


More formal: numbers instead of pronounceable
characters


eg

animal=2, rational=3, human


6


philosophy = 5, philosopher
-
> 30


Calculus Ratiocinator


calculus is binary (1679) and based on prime numbers


this calculation can be performed mechanically




Leibniz’ mechanical calculator


Portable


Inspired by
Pascal, but
also
multiplication


Gottfried Leibniz (1646
-
1716)


Lingua
Characteristica

Universalis


More formal: numbers instead of pronounceable
characters


Eg

animal=2, rational=3, human


6


Philosophy = 5, philosopher
-
> 30


Calculus Ratiocinator


calculus is binary (1679) and based on prime numbers


this calculation can be performed mechanically


Basic assumptions


All our ideas are compounded from a very small number of simple
ideas, which form the alphabet of human thought (
cf

Pinker, 2008)


Complex ideas proceed from these simple ideas by a uniform and
symmetrical combination, analogous to arithmetical multiplication.




Leibniz’ ontological engineering:


Methodology:

(
De Arte
Combinatoria
, 1666
)

1.
Systematic identification of all simple concepts

2.
Careful choice of `signs’
(
eg

prime numbers)

3.
Rules for combination


Reuse:

Wilkins’ top ontology as a starting point


Knowledge representation with inference
engine

(multiplication/resolution)


Prime numbers as signs


Binary calculus


(later: too simple…)


Mechanical rendering
was planned


Leibniz’ dream



“Once the characteristic numbers of most notions
are determined, the human race will have a
new kind of tool, a tool that will increase the
power of the mind much more than optical
lenses helped our eyes, a tool that will be as
far superior to microscopes or telescopes as
reason is to vision”



from: Leibniz, Philosophical Essays *)


…even in service of dispute and justice: “
Calculemus



"The only way to rectify our
reasonings

is to make them as
tangible as those of the Mathematicians, so that we can find our
error at a glance, and when there are disputes among persons, we
can simply say: Let us calculate [
calculemus
], without further ado,
to see who is right.[16]"

(The Art of Discovery 1685, Wiener 51, Wiener,
Philip, 1951. Leibniz: Selections. Scribner.)



"...the plan I have had for a long time to reduce all human
thinking to a calculation, such as we know it in algebra or the
ars

combinatoria
...so that many arguments could be solved, the
certain could be distinguished from the uncertain and even grades
of probability could be measured. Then if two were arguing, they
could say to each other `Let us calculate’

(letter to Protestant
Pietist

Jacob
Spener
, July 1687)

Dream or self delusion?


Medal on calculator:


Motto



“superior to man”



Inscription:


“The model of creation discovered by
G.W.L”


“one is enough for deriving everything
from nothing”

…publish or perish?


Known for…


“Theodicy” (


Voltaire
Candide
)


Invention of calculus (quarrel with Newton)


But wrote…


43 volumes published from 1923 on: still going on


15,000 letters, 1000 correspondents (e
-
mail…), 200K pages



Some justice two centuries later


Russell (1900)


inventor of formal logic


Physics: basis for relativity


Mechanical reasoning (Turing, Wiener, AI)


Ontologies

for reasoning


Mechanized normative (legal) reasoning


Mechanized dispute resolution….


However the Lingua
Universalis

movement is only


A (virtual) ancestor of ontological engineering
by hindsight


Wilkin’s

essay is reprinted in the 90ies (400


100 $ a
copy) but hardly available!


They had an interesting indirect but practical
effect in the development of thesauri:


Roget’s thesaurus (1852, …)


The direct, `real’ sources are in AI and in
knowledge engineering…

Artificial Intelligence


McCarty & Hayes (69)… Some philosophical
problems etc.


The frame problem


Ontology as set of concepts to be represented


Hayes

(79)
The second naive physics manifesto


A classic,
but

only

in
printed

form
!



axiomatization

of
common
-
sense



Clusters of concepts


Places & positions, spaces & objects,


Qualities, quantities & measurement


Change, time & history


Energy, effort & motion, ….

Where it really started


Approaches to knowledge engineering (80ies
-
90ies)


USA: Expert systems, rapid prototyping


EU: (
Common)KADS


Separating domain knowledge from problem solving method


PSM as control structure over domain level inferences


Software engineering life cycle


Specification: Conceptual Modeling
Language(CML
)


Design: system architecture & knowledge base


Meeting ground: EKAW; Banff/KCAP

Combining the two approaches


CommonKADS


Specifying domain knowledge = ontology


Is
-
a, part
-
of and dependency structures in CML


USA


Interoperability between knowledge representation
(KR) formalisms: KIF (Stanford)


Ontologies

can be formalized in KIF


Ontolingua


Ontologies

for creating (specifications of)
knowledge bases of knowledge systems

Defining an ontology as:


“… the specification of a conceptualization”
(Gruber, 92)





(… which applies to any model…)

Ontologies

taking off on higher grounds:


Business process support


Knowledge management:


Indexing and managing large amounts of information in
companies and institutions


Document management


Expertise management


The Semantic Web initiative


Ontologies

representing the semantics of the content
of web pages

Conclusions of the historical account


Ontology has a confusing
--

but also inspiring

-

influence on developing
ontologies


The
lingua
universalis

movement could have
been a more adequate source of inspiration


The potential roots of
ontologies



and
ontological reasoning


were rather in AI, in
particular in Qualitative and Model based
Reasoning (QR/MBR)


Ontologies

come from the Knowledge
Acquisition community for
specifying the
conceptualization of knowledge bases