COMMUNICATION
Outline
|
|
|
Communication as Action |
|
Formal Grammar |
|
Syntactic Analysis (Parsing) |
|
Augmented Grammars |
|
Semantic Interpretation |
|
Ambiguity and Disambiguation |
|
Discourse Understanding |
|
|
Communication
|
|
|
|
Communication |
|
Intentional exchange of information
brought about by the production and perception of signs drawn from a shared
system of conventional signs |
|
Humans use language to communicate most
of what is known about the world |
|
The Turing test is based on language |
Communication as Action
|
|
|
|
Speech
act |
|
Language production viewed as an action |
|
Speaker, hearer, utterance |
|
Examples: |
|
Query: “Have you smelled the wumpus
anywhere?” |
|
Inform: “There’s a breeze here in 3 4.” |
|
Request: “Please help me carry the
gold.” “I could use some help carrying this.” |
|
Acknowledge: “OK” |
|
Promise: “I’ll shoot the wumpus.” |
Fundamentals of Language
|
|
|
|
|
|
|
Formal language: A (possibly infinite)
set of strings |
|
|
|
Grammar: A finite set of rules that
specifies a language |
|
|
|
Rewrite rules |
|
nonterminal symbols (S, NP, etc) |
|
terminal symbols (he) |
|
S ® NP VP |
|
NP ® Pronoun |
|
Pronoun ® he |
|
|
Chomsky Hierarchy
|
|
|
|
Four classes of grammatical formalisms: |
|
Recursively enumerable grammars |
|
Unrestricted rules: both sides of the
rewrite rules can have any number of terminal and nonterminal symbols |
|
AB ® C |
|
Context-sensitive grammars |
|
The RHS must contain at least as many
symbols as the LHS |
|
ASB ® AXB |
|
Context-free grammars (CFG) |
|
LHS is a single nonterminal symbol |
|
S ® XYa |
|
Regular grammars |
|
X ® a X ® aY |
|
|
Component Steps of
Communication
|
|
|
|
SPEAKER: |
|
Intention |
|
Know(H,ØAlive(Wumpus,S3)) |
|
Generation |
|
“The wumpus is dead” |
|
Synthesis |
|
[thaxwahmpaxsihzdehd] |
|
|
Component Steps of
Communication
|
|
|
|
HEARER: |
|
Perception: |
|
“The wumpus is dead” |
|
Analysis |
|
(Parsing): |
|
|
|
|
|
(Semantic Interpretation): ØAlive(Wumpus, Now) |
|
Tired(Wumpus, Now) |
|
(Pragmatic Interpretation): ØAlive(Wumpus1, S3) |
|
Tired(Wumpus1, S3) |
|
|
Component Steps of
Communication
|
|
|
|
|
|
|
HEARER: |
|
Disambiguation: |
|
ØAlive(Wumpus1,S3) |
|
|
|
Incorporation: |
|
TELL( KB, ØAlive(Wumpus1,S3)
) |
|
|
Formal Grammar
|
|
|
|
|
|
|
The lexicon for eo: |
|
|
|
Noun ® stench | breeze | glitter | wumpus | pit | pits | gold | … |
|
Verb ® is | see | smell | shoot | stinks | go | grab | turn | … |
|
Adjective ® right | left | east | dead | back |
smelly | … |
|
Adverb ® here | there | nearby | ahead | right | left | east | … |
|
Pronoun ® me | you | I | it | … |
|
Name ® John | Mary | Boston | Aristotle | … |
|
Article ® the | a | an | … |
|
Preposition ® to | in | on | near | … |
|
Conjunction ® and | or | but | … |
|
Digit ® 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
|
|
Formal Grammar
|
|
|
|
|
|
|
The grammar for eo: |
|
|
|
S ® NP VP I + feel a breeze |
|
| S Conjunction S I feel a breeze + and + I smell a wumpus |
|
NP ® Pronoun I |
|
| Name John |
|
| Noun pits |
|
| Article Noun the + wumpus |
|
| Digit Digit 3 4 |
|
| NP PP the wumpus + to the east |
|
| NP RelClause the wumpus + that is smelly |
Formal Grammar
|
|
|
|
|
|
|
The grammar for eo (continued): |
|
|
|
VP® Verb stinks |
|
| VP NP feel + a breeze |
|
| VP Adjective is + smelly |
|
| VP PP turn + to the east |
|
| VP Adverb go + ahead |
|
PP ® Preposition NP to + the east |
|
RelClause® that VP that + is smelly |
|
|
Formal Grammar
|
|
|
|
|
|
|
Parts of speech |
|
Open class: noun, verb, adjective,
adverb |
|
Closed class: pronoun, article,
preposition, conjunction, … |
|
|
|
Grammar |
|
Overgenerate: “Me go Boston” |
|
Undergenerate: “I think the wumpus is
smelly” |
|
|
Parse Tree
Syntactic Analysis
(Parsing)
|
|
|
|
Parsing: The process of finding a parse
tree for a given input string |
|
Top-down parsing |
|
Start with the S symbol and search for
a tree that has the words as its leaves |
|
Bottom-up parsing |
|
Start with the words and search for a
tree with root S |
|
|
Trace of Bottom-up
Parsing
|
|
|
List of nodes Subsequence Rule |
|
the wumpus is dead the Article ® the |
|
Article wumpus is dead wumpus Noun ® wumpus |
|
Article Noun is dead Article Noun NP ® Article Noun |
|
NP is dead is Verb ® is |
|
NP Verb dead dead Adjective ® dead |
|
NP Verb Adjective Verb VP ® Verb |
|
NP VP Adjective VP Adjective VP ® VP Adjective |
|
NP VP NP VP S ® NP VP |
|
S |
Subjective &
Objective Cases
|
|
|
|
Overgeneration: |
|
S ® NP VP ® NP VP NP ® NP Verb NP |
|
Pronoun Verb NP ® Pronoun Verb Pronoun |
|
|
|
She loves him |
|
*her loves he |
|
She ran towards him |
|
*She ran towards he |
Handling Subjective &
Objective Cases
|
|
|
|
|
|
|
S ® NPs VP | … |
|
NPs ® Pronouns
| Name | Noun | … |
|
NPo ® Pronouno
| Name | Noun | … |
|
VP ® VP NPo | … |
|
PP ® Preposition NPo |
|
Pronouns ® I | you | he | she | it | … |
|
Pronouno ® me | you | him | her | it | … |
|
|
|
Disadvantage: Grammar size grows
exponentially |
Augmented Grammars
|
|
|
|
Handling case, agreement, etc |
|
Augment grammar rules to allow
parameters on nonterminal categories |
|
NP(Subjective) |
|
NP(Objective) |
|
NP(case) |
|
|
Definite Clause Grammar
(DCG)
|
|
|
|
|
|
|
The grammar for e1: |
|
|
|
S ® NP(Subjective) VP | … |
|
NP(case) ®
Pronoun(case) | Name | Noun | … |
|
VP ® VP NP(Objective) |
… |
|
PP ® Preposition NP(Objective) |
|
Pronoun(Subjective) ® I | you | he | she | it | … |
|
Pronoun(Objective) ® me | you | him | her | it | … |
|
|
Definite Clause Grammar
(DCG)
|
|
|
|
|
|
|
Each grammar rule is a definite clause
in logic: |
|
|
|
S ® NP VP |
|
NP(s1) Ù VP(s2) Þ S(s1 + s2) |
|
NP(case) ®
Pronoun(case) |
|
Pronoun(case, s1) Þ NP(case, s1) |
|
|
|
DCG enables parsing as logical
inference: |
|
Top-down parsing is backward chaining |
|
Bottom-up parsing is forward chaining |
Verb Subcategorization
Verb Subcategorization
|
|
|
S ® NP(Subjective) VP([ ]) |
|
VP(subcat) ®
Verb(subcat) |
|
| VP(subcat + [NP])
NP(Objective) |
|
| VP(subcat + [Adjective]) Adjective |
|
| VP(subcat + [PP]) PP |
|
VP(subcat) ®
VP(subcat) PP |
|
| VP(subcat) Adverb |
|
Verb([NP,NP]) ® give | hand | … |
|
|
Parsing Using Verb
Subcategorization
Semantic Interpretation
|
|
|
Semantics: meaning of utterances |
|
First-order logic as the representation
language |
|
Compositional semantics: meaning of a
phrase is composed of meaning of the constituent parts of the phrase |
|
|
Semantic Interpretation
|
|
|
Exp(x) ® Exp(x1)
Operator(op) Exp(x2) |
|
{ x = Apply(op, x1, x2) } |
|
Exp(x) ® ( Exp(x) ) |
|
Exp(x) ® Number(x) |
|
Number(x) ® Digit(x) |
|
Number(x) ® Number(x1) Digit(x2) { x = 10 ´ x1
+ x2 } |
|
Digit(x) ® x { 0 ≤ x ≤ 9 } |
|
Operator(x) ® x { x Î { +, -, ´, ¸ }} |
|
|
Semantic Interpretation
Semantic Interpretation
|
|
|
John loves Mary |
|
Loves(John, Mary) |
|
|
|
(ly lx Loves(x,y)) (Mary) º lx Loves(x, Mary) |
|
(lx Loves(x, Mary)) (John) º Loves(John, Mary) |
|
|
|
S(rel(obj)) ® NP(obj) VP(rel) |
|
VP(rel(obj)) ® Verb(rel) NP(obj) |
|
NP(obj) ® Name(obj) |
|
Name(John) ® John |
|
Name(Mary) ® Mary |
|
Verb(ly lx Loves(x,y) ) ® loves |
Semantic Interpretation
Pragmatic Interpretation
|
|
|
|
|
|
|
Adding context-dependent information
about the current situation to each candidate semantic interpretation |
|
|
|
Indexicals: phrases that refer directly
to the current situation |
|
“I am in Boston today” |
|
(“I” refers to speaker and “today”
refers to now) |
|
|
Language Generation
|
|
|
|
The same DCG can be used for parsing
and generation |
|
Parsing: |
|
Given: S(sem, [John, loves, Mary]) |
|
Return: sem = Loves(John, Mary) |
|
Generation: |
|
Given: S(Loves(John, Mary), words) |
|
Return: words = [John, loves, Mary] |
Ambiguity
|
|
|
|
Lexical ambiguity |
|
“the back of the room” vs. “back up
your files” |
|
“In the interest of stimulating the
economy, the government lowered the interest rate.” |
|
Syntactic ambiguity (structural
ambiguity) |
|
“I smelled a wumpus in 2,2” |
|
Semantic ambiguity |
|
“the IBM lecture” |
|
Pragmatic ambiguity |
|
“I’ll meet you next Friday” |
Metonymy
|
|
|
|
Denotes a concept by naming some other
concept closely related to it |
|
Examples: |
|
Company for company’s spokesperson
(“IBM announced a new model”) |
|
Author for author’s works (“I read
Shakespeare”) |
|
Producer for producer’s product (“I
drive a Honda”) |
|
|
Metonymy
|
|
|
Representation of “IBM announced” |
Metaphor
|
|
|
|
Refer to concepts using words whose
meanings are appropriate to other completely different kinds of concepts |
|
Example: corporation-as-person
metaphor: |
|
Speak of a corporation as if it is a
person and can experience emotions, has a mind, etc. |
|
“That doesn’t scare Digital, which has
grown to be the world’s second-largest computer maker.” |
|
“But if the company changed its mind,
however, it would do so for investment reasons, the filing said.” |
|
|
Disambiguation
Discourse Understanding
|
|
|
|
Discourse: multiple sentences |
|
Reference resolution: The
interpretation of a pronoun or a definite noun phrase that refers to an
object in the world |
|
“John flagged down the waiter. He
ordered a ham sandwich.” |
|
“He” refers to “John” |
|
“After John proposed to Mary, they
found a preacher and got married. For the honeymoon, they went to Hawaii.” |
|
“they”? “the honeymoon”? |
Discourse Understanding
|
|
|
|
|
Structure of coherent discourse:
Sentences are joined by coherence relations |
|
Examples of coherence relations between
S1 and S2: |
|
Enable or cause: S1 brings about a change of state that
causes or enables S2 |
|
“I went outside. I drove to school.” |
|
Explanation: the reverse of enablement,
S2 causes or enables S1 and is an explanation for S1 |
|
“I was late for school. I overslept.” |
|
Exemplification: S2 is an example of
the general principle in S1 |
|
“This algorithm reverses a list. The
input [A,B,C] is mapped to [C,B,A].” |
|
Etc. |
|
|