Embed Size (px)
Citation preview
CS344: Introduction to
Artificial Intelligence
Pushpak Bhattacharyya
CSE Dept.,
IIT Bombay
Lecture 18-19-20– Natural Language
Processing (ambiguities and parsing)
Importance of NLP
Text based computation needs NLP
Machine translation High Quality Information Retrieval
Linguistics+Computation
Perpectivising NLP: Areas of AI and
their inter-dependencies
Search
Vision
PlanningMachine
Learning
Knowledge
RepresentationLogic
Expert
SystemsRoboticsNLP
AI is the forcing function for Computer Science, and NLP of AI
Languages and the speaker population
Language Population (2001 census; rounded
to most significant digit)
Hindi 450, 000, 000
Marathi 72, 000, 000
Konkani 7, 000, 000
Sanskrit 6000
Nepali 13, 000, 000
Languages and the speaker population (contd.)
Language Population (2001 census;
rounded to most significant
digit)
Kashmiri 5, 000, 000
Assamese 13, 000, 000
Tamil 60, 000, 000
Malayalam 33, 000, 000
Bodo 1, 000, 000
Manipuri 1, 000, 000
Great Linguistic Diversity
Major streams
Indo European
Dravidian
Sino Tibetan
Austro-Asiatic
Some languages are ranked within 20 in the world in terms of the populations speaking them
Interesting “mixed-race” languages
Marathi and Oriya: confluence of Indo Aryan and Dravidian families
Urdu: structure from Indo Aryan (Hindi), vocabulary from Persian and Semitic (Arabic) आज मेरी परीक्षा है (aaj merii pariikshaa
hai) {today I have my examination}
आज मेरा इम्तहान है (aaj meraa imtahaan
hai)
3 Language Formula Every state has to
implement
Hindi
The state language (Marathi, Gujarathi, Bengali etc.)
English
Big time translation requirement, e.g.,during the financial year ends
Multilingual Information Access needed for large GoI sector
Legislature Judiciary Education Employment Agriculture Healthcare Cultural
Provide one-stop access and insight into information related to key Government bodies and execution areas
Enable citizens exercise their fundamental rights and duties
Science Housing TaxesTravel & Tourism
Banking & Insurance
International Sports
Need for NLP
Machine Translation
Information Retrieval and Extraction with NLP
Better precision and recall
Summarization
Question Answering
Cross Lingual Search (very relevant for India)
Intelligent interfaces (to Robots, Databases)
Combined image and text based search
Automatic Humour analysis and generation
Last but not the least, window into human mind; language and brain
Roles of Broca’s and Wernicke’s areas
Broadly, Broca’s area is concerned with Grammar while Wernick’s area is concerned with semantics
Damage to former interferes with grammar, e.g. role confusion with voice change: “Ram was seen by Shyam” interpreted as Ram is the seer
Damage to Wernick’s area: finds it difficult to put a name to an entity (which is a tough categorization task)
Evidence of difference between humans and apes in the complexity of language processing: Frontal lobe heavily used in humans ("The brain differentiates human and non-human grammars: Functional localization and structural connectivity" (Volume 103, Number 7, Pages 2458-2463, February 14, 2006)).
MT is needed: Internet Accessibility Pattern
User Type (script) % of World
Population
% access to the
Internet
Latin 39 84
Kanzi (CJK) 22 13
Arabic 9 1.2
Brahmi and Indic 22 0.3
Number of Potential users of Internet
0
50
100
150
200
250
300
350
400
450
Englis
h
Japan
ese
Chin
ese
French
Span
ish
Ger
man
Hin
di
India
n Lan
guages
Languages
Po
pu
lati
on
in
millio
n
Series1
Series2
No of Internet Users in the year 2001
No of Internet Users in the year 2010 (Projected)
Living Languages
Continent No of languages
Africa 2092
Americas 1002
Asia 2269
Europe 239
Pacific 1310
Total 6912
Stages and Challenges of NLP
NLP is concerned with Grounding
Ground the language into perceptual, motor and cognitive capacities.
Grounding
Chair
Computer
Grounding faces 3 challenges
Ambiguity.
Co-reference resolution (anaphora is a kind of it).
Elipsis.
Ambiguity
Chair
Co-reference Resolution
Sequence of commands to the robot:
Place the wrench on the table.
Then paint it.
What does it refer to?
Elipsis
Sequence of command to the Robot:
Move the table to the corner.
Also the chair.
Second command needs completing by using the first part of the previous command.
Stages of processing (traditional view)
Phonetics and phonology
Morphology
Lexical Analysis
Syntactic Analysis
Semantic Analysis
Pragmatics
Discourse
Phonetics
Processing of speech
Challenges
Homophones: bank (finance) vs. bank (river
bank)
Near Homophones: maatraa vs. maatra (hin)
Word Boundary
aajaayenge (aa jaayenge (will come) or aaj aayenge (will come today)
I got [ua]plate
Phrase boundary
Milind Sohoni’s mail announcing this seminar: mtech1 students are especially exhorted to attend as such seminars are integral to one's post-graduate education
Disfluency: ah, um, ahem etc.
Morphology
Word formation rules from root words
Nouns: Plural (boy-boys); Gender marking (czar-czarina)
Verbs: Tense (stretch-stretched); Aspect (e.g. perfective sit-had sat); Modality (e.g. request khaanaa khaaiie)
First crucial first step in NLP
Languages rich in morphology: e.g., Dravidian, Hungarian, Turkish
Languages poor in morphology: Chinese, English
Languages with rich morphology have the advantage of easier processing at higher stages of processing
A task of interest to computer science: Finite State Machines for Word Morphology
Lexical Analysis
Essentially refers to dictionary access and obtaining the properties of the word
e.g. dognoun (lexical property)take-’s’-in-plural (morph property)animate (semantic property)4-legged (-do-)carnivore (-do)
Challenge: Lexical or word sense disambiguation
Lexical Disambiguation
First step: part of Speech Disambiguation
Dog as a noun (animal)
Dog as a verb (to pursue)
Sense Disambiguation
Dog (as animal)
Dog (as a very detestable person)
Needs word relationships in a context
The chair emphasised the need for adult education
Very common in day to day communications and can occur in the form of single or multiword expressions
e.g., Ground breaking ceremony (Prof. Ranade’s email to faculty 14/9/07)
Technological developments bring in new terms, additional meanings/nuances for existing terms
Justify as in justify the right margin (word processing context)
Xeroxed: a new verb
Digital Trace: a new expression
Communifaking: pretending to talk on mobile when you are actually not
Discomgooglation: anxiety/discomfort at not being able to access internet
Helicopter Parenting: over parenting
Syntax
Structure Detection
S
NPVP
V NP
Ilike
mangoes
Parsing Strategy
Driven by grammar S-> NP VP
NP-> N | PRON
VP-> V NP | V PP
N-> Mangoes
PRON-> I
V-> like
Challenges: Structural Ambiguity
Scope The old men and women were taken to safe locations(old men and women) vs. ((old men) and women)Seen in Amman airport: No smoking areas will allow
Hookas inside Preposition Phrase Attachment
I saw the boy with a telescope(who has the telescope?)
I saw the mountain with a telescope(world knowledge: mountain cannot be an instrument of seeing)
I saw the boy with the pony-tail(world knowledge: pony-tail cannot be an instrument of seeing)
Very ubiquitous: today’s newspaper headline “20 years later, BMC pays father 20 lakhs for causing son’s death”
Structural Ambiguity…
Overheard
I did not know my PDA had a phone for 3 months
An actual sentence in the newspaper
The camera man shot the man with the gun when he was near Tendulkar
Headache for parsing: Garden Path sentences
Consider
The horse raced past the garden (sentence complete)
The old man (phrase complete)
Twin Bomb Strike in Baghdad (news paper heading: complete)
Headache for Parsing
Garden Pathing
The horse raced past the garden fell
The old man the boat
Twin Bomb Strike in Baghdad kill 25(Times of India 5/9/07)
Semantic Analysis
Representation in terms of
Predicate calculus/Semantic Nets/Frames/Conceptual Dependencies and Scripts
John gave a book to Mary
Give action: Agent: John, Object: Book, Recipient: Mary
Challenge: ambiguity in semantic role labeling
(Eng) Visiting aunts can be a nuisance
(Hin) aapko mujhe mithaai khilaanii padegii(ambiguous in Marathi and Bengali too; not in Dravidian languages)
Pragmatics
Very hard problem
Model user intention
Tourist (in a hurry, checking out of the hotel, motioning to the service boy): Boy, go upstairs and see if my sandals are under the divan. Do not be late. I just have 15 minutes to catch the train.
Boy (running upstairs and coming back panting): yes sir, they are there.
World knowledge
WHY INDIA NEEDS A SECOND OCTOBER (ToI, 2/10/07, yesterday)
Discourse
Processing of sequence of sentences Mother to John:
John go to school. It is open today. Should you bunk? Father will be very angry.
Ambiguity of openbunk what?Why will the father be angry?
Complex chain of reasoning and application of world knowledge (father will not be angry if somebody else’s son bunks the school)
Ambiguity of fatherfather as parent
orfather as headmaster
Complexity of Connected Text John was returning from school dejected
– today was the math test
He couldn’t control the class
Teacher shouldn’t have made him
responsible
After all he is just a janitor
ML-NLP
NLP as an ML task
France beat Brazil by 1 goal to 0 in the quarter-final of the world cup football tournament. (English)
braazil ne phraans ko vishwa kap phutbal spardhaa ke kwaartaar phaainal me 1-0 gol ke baraabarii se haraayaa. (Hindi)
Categories of the Words in the Sentence
France beat Brazil by 1 goal to 0 in the quarter final of the world cup football tournament
bytoin
theof
Brazilbeat
France10
goalquarter finalworld cupFootball
tournament
contentwords
functionwords
Further Classification 1/2
Brazilbeat
France1
goal0
quarter finalworld cupfootball
tournament
BrazilFrance
1goal
0quarter finalworld cupfootball
tournament
beat
BrazilFrance
1goal
0quarter finalworld cupFootball
tournament
noun
verb
propernoun
commonnoun
Further Classification 2/2
bytoIntheof
thebytoinof
determinerpreposition
Why all this?
Fundamental and ubiquitous information need
who did what
to whom
by what
when
where
in what manner
Semantic roles
beat
France
Brazil
worldcup
football
quarterfinals
1 goal to 0
agent
patient/theme
manner
time
modifier
Semantic Role Labeling: aclassification task
France beat Brazil by 1 goal to 0 in the quarter-final of the world cup football tournament
Brazil: agent or object?
Agent: Brazil or France or Quarter Final or World Cup?
Given an entity, what role does it play?
Given a role, it is played by which entity?
A lower level of classification: Part of Speech (POS) Tag Labeling
France beat Brazil by 1 goal to 0 in the quarter-final of the world cup football tournament
beat: verb of noun (heart beat, e.g.)?
Final: noun or adjective?
Uncertainty in classification: Ambiguity
Visiting aunts can be a nuisance
Visiting:
adjective or gerund (POS tag ambiguity)
Role of aunt: agent of visit (aunts are visitors)
object of visit (aunts are being visited)
Minimize uncertainty of classification with cues from the sentence
What cues?
Position with respect to the verb:
France to the left of beat and Brazil to the right: agent-object role marking (English)
Case marking:
France ne (Hindi); ne (Marathi): agent role
Brazil ko (Hindi); laa (Marathi): object role
Morphology: haraayaa (hindi); haravlaa (Marathi):
verb POS tag as indicated by the distinctive suffixes
Cues are like attribute-value pairs prompting machine learning from NL data
Constituent ML tasks
Goal: classification or clustering
Features/attributes (word position, morphology, word label etc.)
Values of features
Training data (corpus: annotated or un-annotated)
Test data (test corpus)
Accuracy of decision (precision, recall, F-value, MAP etc.)
Test of significance (sample space to generality)
What is the output of an ML-NLP System (1/2)
Option 1: A set of rules, e.g.,
If the word to the left of the verb is a noun and has animacy feature, then it is the likely agent of the action denoted by the verb.
The child broke the toy (child is the agent)
The window broke (window is not the agent; inanimate)
What is the output of an ML-NLP System (2/2)
Option 2: a set of probability values
P(agent|word is to the left of verb and has animacy) > P(object|word is to the left of verb and has animacy)> P(instrument|word is to the left of verb and has animacy) etc.
How is this different from classical NLP
The burden is on the data as opposed to the human.
corpus
Text data
Linguist
Computer
rules
rules/probabilities
Classical NLP
Statistical NLP
Classification appears as sequence labeling
A set of Sequence Labeling Tasks: smaller to larger units
Words: Part of Speech tagging
Named Entity tagging
Sense marking
Phrases: Chunking
Sentences: Parsing
Paragraphs: Co-reference annotating
Example of word labeling: POS Tagging
<s>
Come September, and the IIT campus is abuzz with new and returning students.
</s>
<s>
Come_VB September_NNP ,_, and_CC the_DT IIT_NNP campus_NN is_VBZ abuzz_JJ with_IN new_JJ and_CC returning_VBG students_NNS ._.
</s>
Example of word labeling: Named Entity Tagging
<month_name>
September
</month_name>
<org_name>
IIT
</org_name>
Example of word labeling: Sense Marking
Word Synset WN-synset-no
come {arrive, get, come} 01947900
.
.
.
abuzz {abuzz, buzzing, droning} 01859419
Example of phrase labeling: Chunking
Come July, and is
abuzz with .
the IIT campus
new and returning students
Example of Sentence labeling: Parsing
[S1[S[S[VP[VBCome][NP[NNPJuly]]]]
[,,]
[CC and]
[S [NP [DT the] [JJ UJF] [NN campus]]
[VP [AUX is]
[ADJP [JJ abuzz]
[PP[IN with]
[NP[ADJP [JJ new] [CC and] [ VBG returning]]
[NNS students]]]]]]
[..]]]
Parsing of Sentences
Are sentences flat linear structures? Why tree?
Is there a principle in branching
When should the constituent give rise to children?
What is the hierarchy building principle?
Structure Dependency: A Case Study
Interrogative Inversion
(1) John will solve the problem.
Will John solve the problem?
Declarative Interrogative
(2) a. Susan must leave. Must Susan leave?
b. Harry can swim. Can Harry swim?
c. Mary has read the book. Has Mary read the book?
d. Bill is sleeping. Is Bill sleeping?
……………………………………………………….
The section, “Structure dependency a case study” here is adopted from atalk given by Howard Lasnik (2003) in Delhi university.
Interrogative inversionStructure Independent (1st attempt)
(3)Interrogative inversion process
Beginning with a declarative, invert the first and secondwords to construct an interrogative.
Declarative Interrogative
(4) a. The woman must leave. *Woman the must leave?
b. A sailor can swim. *Sailor a can swim?
c. No boy has read the book. *Boy no has read the book?
d. My friend is sleeping. *Friend my is sleeping?
Interrogative inversioncorrect pairings
Compare the incorrect pairings in (4) with thecorrect pairings in (5):
Declarative Interrogative
(5) a. The woman must leave. Must the woman leave?
b. A sailor can swim. Can a sailor swim?
c. No boy has read the book. Has no boy read the book?
d. My friend is sleeping. Is my friend sleeping?
Interrogative inversionStructure Independent (2nd attempt)
(6) Interrogative inversion process:
Beginning with a declarative, move the auxiliaryverb to the front to construct an interrogative.
Declarative Interrogative
(7) a. Bill could be sleeping. *Be Bill could sleeping?
Could Bill be sleeping?
b. Mary has been reading. *Been Mary has reading?
Has Mary been reading?
c. Susan should have left. *Have Susan should left?
Should Susan have left?
Structure independent (3rd attempt):
(8) Interrogative inversion process
Beginning with a declarative, move the firstauxiliary verb to the front to construct aninterrogative.
Declarative Interrogative
(9) a. The man who is here can swim. *Is the man who here can swim?
b. The boy who will play has left. *Will the boy who play has left?
Structure Dependent Correct Pairings For the above examples, fronting the second
auxiliary verb gives the correct form:
Declarative Interrogative
(10) a.The man who is here can swim. Can the man who is here swim?
b.The boy who will play has left. Has the boy who will play left?
Natural transformationsarestructure dependent
(11) Does the child acquiring English learn these properties?
(12) We are not dealing with a peculiarity of English. Noknown human language has a transformational processthat would produce pairings like those in (4), (7) and(9), repeated below:
(4) a. The woman must leave. *Woman the must leave?
(7) a. Bill could be sleeping. *Be Bill could sleeping?
(9) a. The man who is here can swim. *Is the man who here can swim?
Deeper trees needed for capturing sentence structure
NP
PPAP
big
The
of poems
with the blue cover
[The big book of poems with theBlue cover] is on the table.
book
This wont do! Flat structure!
PP
Other languages
NP
PPAP
big
The
of poems
with the blue cover
[niil jilda vaalii kavita kii kitaab]
book
English
NP
PPAP
niil jilda vaalii kavita kii
kitaab
PP
badii
Hindi
PP
Other languages: contd
NP
PPAP
big
The
of poems
with the blue cover
[niil malaat deovaa kavitar bai ti]
book
English
NP
PPAP
niil malaat deovaa kavitar
bai
PP
motaa
Bengali
PPti
PPs are at the same level: flat with respect to the head word “book”
NP
PPAP
big
The
of poems
with the blue cover
[The big book of poems with theBlue cover] is on the table.
book
No distinction in terms of dominance or c-command
PP
“Constituency test of Replacement” runs into problems
One-replacement:
I bought the big [book of poems with the blue cover] not the small [one]
One-replacement targets book of poems with the blue cover
Another one-replacement:
I bought the big [book of poems] with the blue cover not the small [one] with the red cover
One-replacement targets book of poems
More deeply embedded structureNP
PP
AP
big
The
of poems
with the blue cover
N’1
Nbook
PP
N’2
N’3
To target N1’
I want [NPthis [N’big book of poems with the red cover] and not [Nthat [None]]
Bar-level projections Add intermediate structures
NP (D) N’
N’ (AP) N’ | N’ (PP) | N (PP)
() indicates optionality
New rules produce this treeNP
PP
AP
big
The
of poems
with the blue cover
N’1
Nbook
PP
N’2
N’3
N-bar
As opposed to this tree
NP
PPAP
big
The
of poems
with the blue coverbook
PP
V-bar
What is the element in verbs corresponding to one-replacement for nouns
do-so or did-so
As opposed to this tree
NP
PPAP
big
The
of poems
with the blue coverbook
PP
I [eat beans with a fork]
VP
NP
beans
eat
with a fork
PP
No constituent that groups together V and NP and excludesPP
Need for intermediate constituents
I [eat beans] with a fork but Ram [does so] with a spoon
V2’
NP
beans
eat
with a fork
PP
VP
V1’
V
VPV’V’ V’ (PP)V’ V (NP)
How to target V1’
I [eat beans with a fork], and Ram [does so] too.
V2’
NP
beans
eat
with a fork
PP
VP
V1’
V
VPV’V’ V’ (PP)V’ V (NP)
Parsing Algorithms
A simplified grammar
S NP VP
NP DT N | N
VP V ADV | V
A segment of English Grammar
S’(C) S
S{NP/S’} VP
VP(AP+) (VAUX) V (AP+)
({NP/S’}) (AP+) (PP+) (AP+)
NP(D) (AP+) N (PP+)
PPP NP
AP(AP) A
Example Sentence
People laugh
1 2 3
Lexicon:
People - N, V
Laugh - N, V
These are positions
This indicate that both
Noun and Verb is
possible for the word
“People”
Top-Down Parsing
State Backup State Action
-----------------------------------------------------------------------------------------------------
1. ((S) 1) - -
2. ((NP VP)1) - -
3a. ((DT N VP)1) ((N VP) 1) -
3b. ((N VP)1) - -
4. ((VP)2) - Consume “People”
5a. ((V ADV)2) ((V)2) -
6. ((ADV)3) ((V)2) Consume “laugh”
5b. ((V)2) - -
6. ((.)3) - Consume “laugh”
Termination Condition : All inputs over. No symbols remaining.
Note: Input symbols can be pushed back.
Position of input pointer
Discussion for Top-Down Parsing This kind of searching is goal driven. Gives importance to textual precedence (rule
precedence). No regard for data, a priori (useless expansions
made).
Bottom-Up Parsing
Some conventions:
N12
S1? -> NP12 ° VP2?
Represents positions
End position unknownWork on the LHS done, while
the work on RHS remaining
Bottom-Up Parsing (pictorial representation)
S -> NP12 VP23 °
People Laugh
1 2 3
N12 N23
V12 V23
NP12 -> N12 ° NP23 -> N23 °
VP12 -> V12 ° VP23 -> V23 °
S1? -> NP12 ° VP2?
Problem with Top-Down Parsing
• Left Recursion
• Suppose you have A-> AB rule.
Then we will have the expansion as follows:
• ((A)K) -> ((AB)K) -> ((ABB)K) ……..
Combining top-down and bottom-up strategies
Top-Down Bottom-Up Chart Parsing
Combines advantages of top-down & bottom-
up parsing.
Does not work in case of left recursion.
e.g. – “People laugh”
People – noun, verb
Laugh – noun, verb
Grammar – S NP VP
NP DT N | N
VP V ADV | V
Transitive Closure
People laugh
1 2 3
S NP VP NP N VP V
NP DT N S NPVP S NP VP
NP N VP V ADV success
VP V
Arcs in Parsing
Each arc represents a chart which records
Completed work (left of )
Expected work (right of )
Example
People laugh loudly
1 2 3 4
S NP VP NP N VP V VP V ADV
NP DT N S NPVP VP VADV S NP VP
NP N VP V ADV S NP VP
VP V
Advantage of Combination of Bottom Up & Top Down parsing over either of top down / bottom down
In top down bottom up parsing
1. Like top down parsing productions are brought, but inline top down parsing rules are not necessarily expanded
2. Unlike bottom up parsing uncontrolled lexical options (parts of speech) are not even considered.
Dealing With Structural Ambiguity
Multiple parses for a sentence
The man saw the boy with a telescope.
The man saw the mountain with a telescope.
The man saw the boy with the ponytail.
At the level of syntax, all these sentences are ambiguous. But semantics can disambiguate 2nd & 3rd sentence.
Prepositional Phrase (PP) Attachment Problem
V – NP1 – P – NP2
(Here P means preposition)
NP2 attaches to NP1 ?
or NP2 attaches to V ?
Parse Trees for a Structurally Ambiguous Sentence
Let the grammar be –
S NP VP
NP DT N | DT N PP
PP P NP
VP V NP PP | V NP
For the sentence,
“I saw a boy with a telescope”
Parse Tree - 1S
NP VP
N V NP
Det N PP
P NP
Det N
I saw
a boy
with
a telescope
Parse Tree -2S
NP VP
N V NP
Det N
PP
P NP
Det N
I saw
a boy with
a telescope
