yılmaz kılıçaslan concept lattices as semantic models
TRANSCRIPT
![Page 1: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/1.jpg)
Yılmaz Kılıçaslan
Concept Lattices as Semantic Models
![Page 2: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/2.jpg)
Outline
· Thematic Roles– History– Thematic Hierarchies– Generalized Thematic Roles
· Formal Concept Analysis– Formal Concepts– Concept Lattices– Formal Contexts
· Concept Lattices as Models of Thematic Structures– Thematic Tier– Action Tier– Experience Tier
2
![Page 3: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/3.jpg)
Three Elements of Linguistic Meaning
3
WORDS REALITY
MINDS
LINGUISTIC MEANING
Words can be constituents of meanings.
Donald Davidson
The external world accommodates meanings.
Jon Barwise
Meaning’s natural home is the mind.
John locke
VIEWS OF REALITY:
1.ARISTOTELIAN
2. PLATONIC
3 NOMINALIST
4. CONCEPTUALIST
5. SOLIPSIST
NATURAL WORLD
REALM OF IDEAS
NO STRUCTURE
MIND
NO REALITY
SOURCE OF STRUCTURE:
![Page 4: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/4.jpg)
The Quoational TheoryA Nominalist View of Reality
4
Anti-realist with regard to universals
Example: John believes that Venus is spherical
Universals do not exist.
Quotations conceived of as objects Language is subsumed by reality.
A Version of First-Order Logic
Belief verbs are relations between individuals and the quotations of sentences.
B(j, + Venus is spherical , )
Belief verbs are relations between individuals and the quotations of sentences.
![Page 5: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/5.jpg)
Problems with the Quoational Theory
· Even though it seems to be sufficiently finely discriminating for belief contexts, the Quotational Theory is not free from problems.
· First of all, it is not intuitive to consider the object of a belief relation to be meaning-independent.
· Also, although it is not possible to quantify into quotation:
it is possible to quantify into belief contexts:
· It is a merit of possible-worlds semantics to handle quantification into a belief context:
5
x[‘x’ has eight letters]does not follow from‘bachelor’ has eight letters
It is possible for john to believe that Venus is spherical and this belief can be expressed as: John believes that the Morning Star is sphericaland for him not to know that the Morning Star is Venus.
x[B(j, ^s(x))](m)
![Page 6: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/6.jpg)
Relativizing a Proposition to a BelieverA Conceptualist View of Reality
6
Anti-realist like nominalists
Example: John believes that Venus is spherical might come out as true while John believes that the Morning Star is spherical
is false, because John’s understanding of Venus is not the same as his understanding of the Morning Star.
Universals do not exist.
Mind as the ground for predication Mind is subsumed by reality.
Semantics in terms of mental states
The semantics of an object of belief might be relativized to its believer.
![Page 7: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/7.jpg)
Problems with Conceptualism
· Relativization to believers’ mental states seem to provide us with meanings:– fine enough to distinguish between all pairs of belief sentences, – but, not coarse enough to make an inference like the following
valid:
John believes that clouds are alive.
Mary believes everything that John believes.----------------------------------------------------------Therefore, Mary believes that clouds are alive.
· We have no good way to classify the ideas that expressions stand for.
· The problem of external significance is simply pushed from expressions to ideas.
7
![Page 8: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/8.jpg)
Philosophical plausibility of non-Realist Views
· The non-realist views do not seem to be very plausible on philosphical grounds:– Nominalism: The world should have structure above
and beyond set membership.– Conceptualism: Mind and language would not have
evolved in a structureless world.– Solipsism: It is a form of madness to really believe
that the world is only a projection of one’s mind.
8
![Page 9: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/9.jpg)
The Aristotelean View of Reality
9
Realist ViewParticulars and universals exist independent of people’s minds and words.
Platonic ViewParticulars and universals exist in different realms and are not causally connected.
Scientific View1) There are properties and relations between things in this world, independent of language and mind.
2) These universals play a role in the causal order.
Aristotelian View Universals are real but their existence is dependent on particulars.
this table is wooden, brown, in the room, etc. It has all these properties; and it is not a thing without properties, behind them ...%
Saul Kripke
![Page 10: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/10.jpg)
Formal Concept Analysis
· Problem: How to formalize an ontology where both objects (particulars) and attributes (universals) are integrated in a systematic way.
· Solution: Formal Concept Analysis
10
A mathematical theory of concepts and concept hierarchies which aims to derive a formal ontology from a collection of objects and their attributes
![Page 11: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/11.jpg)
Formal Concepts
· A concept in FCA is a pair consisting of a set of objects, which is the ‘extent’, and a set of attributes, which is the ‘intent’, such that:– the extent consists of all objects that share the given
attributes
and– the intent consists of all attributes shared by the given
objects.
11
![Page 12: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/12.jpg)
Concept Lattices
· A formal ontology derived by FCA is a concept hierarchy where:– the set of all concepts is ordered by a subconcept-superconcept
relation, which is a particular order relation denoted by ≤.
· If (O1,A1) and (O2,A2) are concepts, the former is said to be
a subconcept of the latter, i.e. (O1,A1) ≤ (O2,A2) iff:– O1 ⊆ O2 ⇔ A1 ⊇ A2.
· A set ordered in this way is called a concept lattice.· A concept lattice can be drawn as a diagram in which
concepts are represented by nodes interconnected by lines going down from superconcepts to subconcepts.
12
![Page 13: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/13.jpg)
Deriving a Concept Lattice from a Formal Context
13
O = {John, Fido, Tweety}A = {animate, smart, two-legged, furry}
EXAMPLE: FORMAL CONTEXT:
animate smart two-legged furry
John ✓ ✓ ✓Tweety ✓ ✓ ✓Fido ✓ ✓ ✓
animate
j, f, t
animate,smart
j, f
animate, furry
f, t
animate,two-legged
j, t
animate,smart,two-legged
j
animate,two-legged,furry
tanimate,smart,furry
j
animate,smart,two-legged,furry
CONCEPT LATTICE:
![Page 14: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/14.jpg)
Concept Lattices as Domain Models
14
Gebäude
Haus
ev
bina
business
building
largeresidential or small
smallresidential
house
H O
A Concept Lattice for the building domain:
![Page 15: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/15.jpg)
FCA and Situation TypesLexical Aspect (Aktionsart)
15
S
dynamic,durative
S3
dynamic
S1
durative
S2
S: All existing situationsS1 ⊆ SS2 ⊆ SS3 ⊆ S1 S3 ⊆ S2
S3S1
S2
Semelfactives StatesActivities
dynamic,telic
S4
dynamic,durative,telic
S5
S4 ⊆ S1S5 ⊆ S4 S5 ⊆ S3S4 − S5: AchievementsS5: Accomplishments
SITUATION TYPES
![Page 16: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/16.jpg)
FCA and Relative TenseA Reichenbachian Account
16
Reference Time
E ≤ R, E ≥ R
Progressive
E ≤ R
Retrospective
Reference Time: A relation between event time (E) and reference time (R)
E ≤ R: Retrospective
E ≥ R: Prospective
(E ≤ R E ≥ R) ⇒ R = E: Progressive
ASSUMPTIONS ABOUT RELATIVE TENSE
Prospective
R ≥ R
![Page 17: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/17.jpg)
FCA and Thematic RolesCentral Roles
17
theme
j, f, m
actor,theme
j
actee, theme
f,m
actor,actee,theme
theme,exp
j,m
actor,theme,exp
j
actor,actee,theme,exp
actee,theme,exp
m
theme
j, f, m
actor,theme
j
actee,theme
m
actor,actee,theme
per
p,j,m
source,per
j, p
goal,per
m
source,goal,per
Example: John sent Mary Fido from Paris J
{j}
{source,...}
{m}
{goal,...} to Maryà MarieMeryem’e
Mary* Marie* Meryem
John gave Mary Fido. Jean a donné Marie Fido. Can Meryem’i Fido’yu verdi.
John gave Fido to Mary. Jean a donné Fido à Marie. Can Meryem’e Fido’yu verdi.
SUBJECT OBJECT
S-F
INA
LS-
INIT
IAL
actor + experiencer = agentactee + experiencer = patient
Bir çocuk camı kırdı.Çocuğu bir arı soktu.
EXPERIENCER SUBJ STIMULUS SUBJ
x like y y please xx fear y y frighten
![Page 18: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/18.jpg)
A Bilingual Concept Lattice Generator
18
Objects: tuples of synset numbers and sets of synonymous English words.
Attributes: the words of the hypernymic synsets.
![Page 19: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/19.jpg)
Thematic Lattice
19
LOCATIVE COGNITIVE_OR_IN_COGNITION IN_ACTION
PARTICIPANT
ACTION_IN_LOCATION_AND_COGNITION
LOCATION_IN_COGNITION LOCATION_IN_ACTION ACTION_IN_COGNITION
Mary is Parisian. J
She was in a park. J
Bees were flying around. J
She was happy. J
A bee hit her face. J
She got stung by another. J
She escaped the park. J
She saw her house broken into. J
![Page 20: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/20.jpg)
Thematic ‘Fractalization’ - I
20
![Page 21: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/21.jpg)
Thematic ‘Fractalization’ - II
21
![Page 22: Yılmaz Kılıçaslan Concept Lattices as Semantic Models](https://reader036.vdocuments.net/reader036/viewer/2022062321/56649e7c5503460f94b7edc2/html5/thumbnails/22.jpg)
Conclusion
22
Concept lattices can be formal and realist models of semantic domains for both lexical and grammatical forms of natural language.