Lexical Semantics
Read J & M Chapter 16.
The task of classifying all the words of language, or what's the same
thing, all the ideas that seek expression, is the most stupendous of
logical tasks. Anybody but the most accomplished logician must break
down in it utterly; and even for the strongest man, it is the severest
possible tax on the logical equipment and faculty.
Charles Sanders Peirce, letter to editor B. E. Smith of the Century
Dictionary
Relating Words and Concepts
Words
Surface properties
Morphological
Spelling
Pronunciation
Grammatical function
Part of speech
Subcategorization
Concepts
Some properties, e.g. number
Objects, actions, events,
properties
"
Meaning
Taxonomic relations
Inference rules
Register
Discourse conventions
One to Many Mappings
Homonyms
(same spelling,
same pronounciation,
different meanings)
spring
One to Many Mappings
Homographs
(same spelling,
different pronounciation,
same meaning)
bass
One to Many Mappings
Homophones
(different spelling,
same pronounciation,
different meanings)
night
knight
http://www.cooper.com/homophonezone/
One to Many Mappings
Polysemy
(multiple related
meanings)
knight
Many to One Mappings
Synonymy
food
nourishment
grub
Synsets
The largest synset in WordNet is:
buttocks, nates, arse, butt, backside, bum, buns, can, fundament, hi
ndquarters, hind end, keister, posterior, prat, rear, rear end, rump, s
tern, seat, tail, tail end, tooshie, tush, bottom, behind, derriere,
f
anny, ass
The next is:
dohickey, dojigger, doodad, doohickey, gimmick, hickey, gizmo,
gismo, gubbins, thingamabob, thingumabob, thingmabob,
thingamajig, thingumajig, thingmajig, thingummy
http://www.cogsci.princeton.edu/~wn/
Other Relations Among Words
Hyponymy
animal
mammal
horse
horse is a hyponym of mammal and animal.
mammal is a hypernym of horse.
The Same Thing for Verbs
Troponymy
go
walk
shuffle
amble
walk is a troponym of go.
swagger
march
Another Relation between Words
Meronymy
Nouns:
brim and crown are meronyms of hat
Verbs:
step is a meronym of walk
WordNet
http://www.cogsci.princeton.edu/~wn/
WordNet Sense Distribution
Maybe We Need to Represent Relationships
Among Concepts, not Words
Maybe We Need to Represent Relationships
Among Concepts, not Words
weightless
light
pale
Ontology
The subject of ontology is the study of the categories of things that exist or may exist in some
domain. The product of such a study, called an ontology, is a catalog of the types of things that
are assumed to exist in a domain of interest D from the perspective of a person who uses a
language L for the purpose of talking about D. The types in the ontology represent the predicates,
word senses, or concept and relation types of the language L when used to discuss topics in the
domain D. An uninterpreted logic, such as predicate calculus, conceptual graphs, or KIF, is
ontologically neutral. It imposes no constraints on the subject matter or the way the subject may
be characterized. By itself, logic says nothing about anything, but the combination of logic with
an ontology provides a language that can express relationships about the entities in the domain of
interest.
An informal ontology may be specified by a catalog of types that are either undefined or defined
only by statements in a natural language. A formal ontology is specified by a collection of names
for concept and relation types organized in a partial ordering by the type-subtype relation. Formal
ontologies are further distinguished by the way the subtypes are distinguished from their
supertypes: an axiomatized ontology distinguishes subtypes by axioms and definitions stated in a
formal language, such as logic or some computer-oriented notation that can be translated to logic;
a prototype-based ontology distinguishes subtypes by a comparison with a typical member or
prototype for each subtype. Large ontologies often use a mixture of definitional methods: formal
axioms and definitions are used for the terms in mathematics, physics, and engineering; and
prototypes are used for plants, animals, and common household items.
- John Sowa (http://www.jfsowa.com/ontology/)
An Example of an Ontology
Penman (Generalized) Upper Model:
http://www.darmstadt.gmd.de/publish/komet/gen-um/node9.html
What are the Real Differences between
Words and Concepts?
Concepts without words (e.g., schadenfreude, or <the gook that builds
up around the top of a ketchup bottle>)
Many to many mappings
Surface linguistic facts, such as subcategorization frames:
I gave the book to John.
I gave John the book.
I was mad at John.
I was angry at John.
* I was sore at John.
* I was livid at John.
I donated the book to John.
* I donated John the book.
Linguistic Facts are More Arbitrary than
World Knowledge
John gave/sent/read Bill the book.
* John donated/returned/transferred Bill the book.
One possible explanation:
Give, send, and read come to English through German.
Donate, return, and transfer come to English from Latin.
What Classes Should an Ontology Contain?
•Use words as the concepts. WordNet synsets do this.
•Use words plus some general concepts for which we don’t have words:
•SAYING&SENSING in the Upper Model
•Psychological Feature in WordNet
•Create new primitives and break words down into them:
•Conceptual Dependency
Representing Events and Relationships
Many nouns have a straightforward semantics: The noun corresponds
to a set of objects in the world. Examples: cat, apple, car.
Many adjectives can also be represented as sets of things that possess
some property: red, fuzzy, sharp.
But most verbs, as well as many nouns, adjectives, and adverbs,
represent events and relationships that have internal structure:
John gave Bill the book before he left for school.
How Close to the Surface Should We Stay?
John kicked the ball.
(1) e,x isa(e,kicking)  kicker(e,John)  kicked-obj(e,x)  isa(x,ball)
(2) e,x isa(e,kicking)  agent(e,John)  AE(e,x)  isa(x,ball)
(3) John  PROPEL  ball
John  MOVE  foot 
Loc(ball)
When is Deep Meaning Worthwhile?
Much of the early work on CD was for story comprehension. We
need deep meaning if we want to be able to answer questions like,
“What moved?”
But suppose we’re interested in MT?
Thematic Roles
A middle ground:
Mapping Surface Forms to Thematic Roles
Priority for subject assignment: AGENT, INSTRUMENT, THEME
Sue cooked the potatoes.
The steam cooked the potatoes.
The potatoes cooked.
Mary cooked.
How to assign roles? Selectional restrictions.
Mapping Words to Meanings
FrameNet (http://www.icsi.berkeley.edu/~framenet/)
Conceptual Dependency
One Important Issue – Lexicons Change all
the Time
As new concepts emerge:
http://www.ananova.com/news/story/sm_801230.html?menu=news.technology.email
As new expressions for old concepts emerge: 24/7