interlingua-based mt interlingua-based machine translation syntactic transfer-based mt – couples...
TRANSCRIPT
Interlingua-based MT
Interlingua-based Machine Translation
• Syntactic transfer-based MT – Couples the syntax of the two
languages
• What if we abstract away the syntax
– All that remains is meaning – Meaning is the same across
languages – Simplicity: Only N components
needed to translate among N languages
• Two “small” problems:– What is meaning?– How do we represent meaning?
Direct MT
Interlingua
Transfer-basedMT
Source Target
Parsing
Semantic Interpretation
Semantic Generation
Syntactic Generation
Syntactic Structure
Syntactic Structure
English analyzer
Spanish analyzer
Japanese analyzer
Spanish Generator
Japanese Generator
English generator
Interlingual representation
Example of Interlingua Machine Translation
)2(_);2,(1);1,( ecallcollecteIMakeeeINeed
need
I make
to call
a collect
indefssDefinitene
collectattributes
call
Theme
IAgent
InfinitiveTense
MakeEvent
Theme
IAgent
presentTense
NeedEvent
:
::
:
:
:
:
:
:
:
必要があります (need)
私は (I)
かける (make)
コールを (call)
コレクト (collect)
Interlingua representation
Ingredients of a semantic representation• language neutral
– Syntactic variations should result is the same semantics
• sense of a word• deep semantic role labels• scope of quantifiers, adverbials, adjectives• polarity information
Distinguish between
surface structure (syntactic structure) and
deep structure (semantic structure) of sentences.
Different forms of semantic representation:
logic formalisms
ontology / semantic representation languages • Case Frame Structures (Filmore)• Conceptual Dependy Theory (Schank)• Description Logic (DL) and similar KR languages • Ontologies
Text Meaning Representation
• Lexicon has two components– Syntactic part– Semantic constraints part
• Given a sentence, the syntactic part analyzes the input syntactically and the semantic constraints create semantic expressions that can be evaluated.
• Ontology specifies the type hierarchy– Used for checking selectional restrictions – Selectional restrictions used for word-sense disambiguation
• e.g. accident is an event; organization has humans
Constructing a Semantic Representation
General approach:
Start with surface structure derived from parser.
Map surface structure to semantic structure: Use phrases as sub-structures. Find concepts and representations for central phrases (e.g. VP,
NP, then PP) Assign phrases to appropriate roles around central concepts
(e.g. bind PP into VP representation).
Semantic Representation
Semantic Representations are based on some form of (formal) Representation Language.
• Semantics Networks
• Conceptual Dependency Graphs
• Case Frames
• Ontologies
• DL and similar KR languages
Important note: Difference between relations between text strings and referents in the world.
Ontology (Interlingua) approach
Ontology: a language-independent classification of objects, events, relations
A Semantic Lexicon, which connects lexical items to nodes (concepts) in the ontology
An analyzer that constructs Interlingua representations and selects an appropriate one
Semantic Lexicon
Provides a syntactic context for the appearance of the lexical item
Provides a mapping for the lexical item to a node in the ontology (or more complex associations)
Provides connections from the syntactic context to semantic roles and constraints on these roles
Constructing an InterLingua Representation
For each syntactic analysis:
Access all semantic mappings and contexts for each lexical item.
Create all possible semantic representations.
Test them for coherency of structure and content.
Input: John makes toolsSyntactic Analysis:
Basic Semantic Dependency - Example
cat verbroot maketense present
subject root johncat noun-proper
object root toolcat nounnumber plural
John-n1syn-struc
root johncat noun-proper
sem-struchuman
name johngender male
tool-n1syn-struc
root toolcat n
sem-structool
Lexicon Entries for John and tool
Relevant extract from the specification of the ontological concept used to describe the appropriate meaning of make:
manufacturing-activity...
agent human theme artifact
…
Ontological Representation - Example
who
what
The basic semantic dependency component of the “Text Meaning Representation” (TMR) for:
John makes tools
manufacturing-activity-7
agent human-3theme set-1
element toolcardinality > 1
…
Semantic Dependency Component
try-v3syn-struc
root trycat vsubj root $var1
cat nxcomp root $var2
cat vform OR infinitive gerund
sem-strucset-1 element-type refsem-1
cardinality >=1refsem-1 sem event
agent ^$var1effect refsem-2
modalitymodality-type epiteucticmodality-scope refsem-2modality-value < 1
refsem-2 value ^$var2sem event
Means “non finished action; outcome unclear”
semantic representation of “try-v3”
REQUEST-INFO-130 THEME DEVELOP-2601.PURPOSE DEVELOP-2601.REASON TEXT-POINTER why INSTANCE-OF REQUEST-INFO
DEVELOP-2601THEME SET-2555AGENT NATION-97PHASE CONTINUOUS
TIME FIND-ANCHOR-TIME INSTANCE-OF DEVELOP
TEXT-POINTER developing
NATION-97HAS-NAME Iraq
INSTANCE-OF NATIONTEXT-POINTER Iraq
SET-2555 ELEMENT-TYPE WEAPONCARDINALITY > 1
INSTRUMENT-OF KILL-1864 THEME-OF DEVELOP-2601 INSTANCE-OF WEAPON
TEXT-POINTER weapons
KILL-1864 THEME SET-2556 INSTRUMENT SET-2555 INSTANCE-OF KILL
TEXT-POINTER destruction
SET-2556 THEME-OF KILL-1225 ELEMENT-TYPE HUMAN
CARDINALITY > 100 INSTANCE-OF HUMAN
TEXT-POINTER mass
“Why is Iraq developing weapons of mass destruction?”
Word sense Disambiguation
Methods
Constraint checking
make sure the constraints imposed on context are met
Graph traversal
is-a links are inexpensive
other links are more expensive
the “cheapest” structure is the most coherent one
Hunter-gatherer processing
find (hunt) and eliminate (kill) unlikely interpretations
collect (gather) remaining interpretations
Ontological Semantics: An example semantic representation language
slides from S. Nirenberg
Ontological semantics is a computationally tractabletheory of meaning in natural language as well as asuite (OntoSem) of implemented NLP programs and a set ofstatic knowledge resources that support these programs.
Ontological semantics deals directly with extraction,representation and manipulation of text meaning.
Ontosem text analyzers produce interpreted knowledge ready to be used in reasoning-heavy applications that include question answering, cross-document and cross-lingual text summarization, question answering, machine translation and others.
Support of intelligent human-computer interaction in domain- and task-oriented environments is squarelywithin the purview of ontological semantics.
Ontological semantics concentrates on contentof representations and is adaptable to a number of different representation formats.
Ontological semantics is both a producer and aconsumer of knowledge: deriving text meaning isitself a knowledge-intensive task
OntoSem
• is devoted to processing naturally occurring texts• strives for high-quality results first followed by concern for broad coverage• expects “unexpected” inputs• seeks quality heuristics of any provenance (knowledge- based or probabilistic, cooccurrence-based)• does not grant syntax a privileged position among the providers of heuristics for semantic processing• does not make a strong distinction between semantics and pragmatics• is applicable to any natural language
Ontological-semantic analyzers take natural language texts as inputs and generate machine-tractable text meaning representations (TMRs) that form the basis of various reasoning processes.
Sample Input Sentence:
Iran, Iraq and North Korea on Wednesday rejected an accusation by President Bush that they are developing weapons of mass destruction.
The TMR (presented graphically) for the above isas follows:
Output: A Text Meaning Representation (TMR)
This presentation is simplified; the system, in fact, derives much more from text;event instances are shown in ellipses; object instances, in rectangles; only caserole and set membership relations are shown (as labels on links); numerical constraints can be fuzzy, as in the cardinality of SET-1226.
DENY-1224 ;; speech act AGENT SET-1224 THEME DEVELOP-1224 TIME < FIND-ANCHOR-TIME WEDNESDAY-1224 INSTANCE-OF DENY
TEXT-POINTER reject
ACCUSE-1224 ;; President BushÕs accusation AGENT HUMAN-15691
BENEFICIARY SET-1224 THEME DEVELOP-1224 INSTANCE-OF ACCUSE
TEXT-POINTER accusation
DEVELOP-1224 ;; developing weaponsTHEME SET-1225THEME-OF DENY-1224 ACCUSE-1224
AGENT SET-1224PHASE CONTINUOUS
PURPOSE WARN-1224TIME FIND-ANCHOR-TIME
INSTANCE-OF DEVELOPTEXT-POINTER developing
Word Sense Disambiguation
Instances of OntologicalConcepts
Semantic Dependencies(fillers of ontological properties mentioned intext; not simply relationsamong textual strings)
Triggers for further context-dependent processing
Many additional properties stored with concepts underlying instances
A pretty-printed fragment ofthe actual TMR representationfor sample input
Ontological-semantic systems centrally rely on the followingstatic knowledge resources:
a language-independent ontology that includes knowledge about types of entities in the world, e.g., ATHLETE, WELD or SPEED; ontology-oriented lexicons (and onomasticons, or lexicons of proper names) for each natural language in the system; and a fact repository containing instances of ontological concepts, e.g., Andre Agassi (ATHLETE-3176) or the Apollo 13 mission (SPACEFLIGHT-142)
A Sample Screen of the Ontology/Lexicon/Fact Repository Browsing and Editing Environment
(diagnosis (diagnosis-n1 (cat n) (anno (def "") (ex "The diagnosis (of cancer) (by the specialist) was made
quickly") (comments ""))
(syn-struc ((root $var0) (cat n) ; diagnosis (pp-adjunct ((root of) (root $var1) (cat prep) (opt +) ; of (obj ((root $var2) (cat n))))) ; disease (pp-adjunct ((root by) (root $var3) (cat prep) (opt +) ; by (obj ((root $var4) (cat n))))))) ; someone (sem-struc (DIAGNOSE ; the ontological mapping (agent (value ^$var4)) ; the case roles (theme (value ^$var2))) (^$var1 (null-sem +)) ; blocks compositional analysis of preps (^$var3 (null-sem +)))) )
(cancer (cancer-n1 (cat n) (anno (def "a disease") (ex "") (comments "") ) (syn-struc ((n ((root $var1) (cat n) (opt +))) ; animal part as modifier (root $var0) (cat n) ; cancer )) (sem-struc (CANCER (location (value ^$var1) (sem animal-part))) ) )
(cancer-n2 (cat n) (anno (def "a sign of the zodiac") (ex "") (comments "") ) (syn-struc ((root $var0) (cat n) )) (sem-struc (CANCER-ZODIAC) ) ) )
Currently Available Static Knowledge Sources for English:
• Ontology of about 6,500 concepts (about 95,000 property-value pairs)• English lexicon of about 40,000 entries• Fact repository of about 20,000 facts (outside medical domain)• English Onomasticon of about 350,000 entries• Tokenization knowledge, morphological and syntactic grammars for a number of languages
Preprocessor
InputText
SyntacticAnalyzer
Grammar:Ecology
MorphologySyntax
Lexicon andOnomasticon
Static Knowledge Resources
SemanticAnalyzer
Ontology andFact Repository
TMR
Processing Modules
The analyzer’s conceptual architecture
(in reality, not strictly pipelined)
The basic (“who did what to whom”) semantic dependency is derived, in the general case, on the basis of
a) lexical-semantic expectations (selectional restrictions) recorded in the ontology and the lexicon and
b) syntactic dependency derived from the results of syntactic analysis.
The beginnings of system evaluation
Run I: “raw”Run II: preprocessor output correct; Run III: preprocessor and syntactic analysis output correct
Sentences 1 2 3 4 5 6 AverageWords 28 33 8 24 33 26 25.33Senses 79 86 29 150 96 76 86Words in / not in lexicon 28/0 32/1 5/3 24/0 31/2 24/2 24/1.33Syntactic ambiguity count 192 32 16 19 63 47 61.5Overall ambiguity count >1.7M >149M 64 >199M >418M >268K >120MWS disambiguation I 52% 48% 50% 46% 30% 50% 48.0%Semantic dependencies I 67% 33% 17% 40% 33% 29% 36.5%WS disambiguation II 96% 68% 67% 83% 88% 54% 76.0%Semantic dependencies II 69% 50% 63% 33% 69% 29% 52.2%WS disambiguation III 96% 100% 67% 88% 90% 100% 90.2%Semantic dependencies III 85% 100% 63% 90% 100% 86% 87.3%
In addition to the basic semantic dependency, TMRs also include parameterized information provided by the microtheoriesof aspect, modality (including speaker attitudes), time, styleand others.
Most of these microtheories have been implemented. All would benefit from further work. We are also actively looking intopossibilities of borrowing some microtheories -- either in toto or partially.
FrameNet: Another example of semantic representation
Frame Semantics (Fillmore 1976, 1977, ..)
• Frame: a conceptual structure or prototypical situation
• Frame elements (roles) – Identify participants of the situation– Are local to their frame
• Frame evoking elements (verbs, nouns, adjectives) introduce frames
• E.g. VERDICT:
[The jury]Judge convicted [him]Defentant [on the counts of theft]Charges.
On Thursday [a jury]Judge found [the youth]Defendant [guilty of wounding Mr Lay] Finding
Berkeley FrameNet Project
• Database of frames for core lexicon of English
• Current release: 610 frames, about 9000 lexical units
Types of Relations
FrameNet Relations
• Frame hierarchy: inherits
• Subframes
Contextual Relations between instantiated frames and roles
• Syntactic and/or semantic embedding
• Discourse relations
• Anaphoric relations
Inferences
• On the basis of both
A Case Study
In the first trial in the world in connection with the terrorist attacks of 11 September 2001, the Higher Regional Court of Hamburg has passed down the maximum sentence. Mounir al Motassadeq will spend 15 years in prison. The 28-year-old Moroccan was found guilty as an accessory to murder in more than 3000 cases.
FrameNet „as a Net“– Frame-to-Frame Relations –
Subframe relation
• Super frame represents complex event
• Subframes represent sub-events
• Subframes usually inherit some roles of the super frame
Criminalprocess
Arraignment Arrest Sentencing Trial
Charge
JudgeDefendant
Defense
Court
Jury
Offense
Prosecution
Charge
Defendant
... ... ... ...
Local Roles
In the first trial in the world in connection with [the [terrorist]Assailant attacks of [11 September 2001]Time]Case, [the Higher Regional Court of Hamburg]Court has passed down the [maximum]Type sentence.
Local Roles
[Mounir al Motassadeq]Inmates will spend [15 years]Duration in prison.
Local Roles
[The 28-year-old Moroccan]Defendant was found [guilty]Finding as [an accessory to [murder]FocalEntity [in more than 3000 cases]Victim ]Charge.
Unfilled Roles
Target Frame Frame roles Filler (given vs. Induced)
trial TRIAL CASE terrorist attacks (1)
CHARGE accessory to murder (2)
COURT Higher Regional Court (3)
DEFENDANT ... 28-year-old Moroccan (4)
attacks ATTACK ASSAILANT terrorist (5)
VICTIM ... (6) TIME (exth.) 11 September 2001(7)
sentence SENTENCING CONVICT Mounir al Motassadeq (8) COURT Higher Regional Court (9) TYPE ... maximum sentence (10)
prison PRISON INMATES ... Mounir al Motassadeq (11) DURATION (exth.) 15 years (12)
found VERDICT CASE terrorist attacks (13)
CHARGE accessory to murder (14) DEFENDANT 28-year-old Moroccan (15) FINDING ... guilty (16)
accessory ASSISTANCE CO-AGENT (17)
FOCAL_ENTITY murder (18)
HELPER ... 28-year-old Moroccan (19)
murder KILLING KILLER (20)
VICTIM ... m.t. 3000 cases (21)
Target Frame Frame roles Filler (given vs. Induced)
trial TRIAL CASE terrorist attacks (1)
CHARGE accessory to murder (2)
COURT Higher Regional Court (3)
DEFENDANT ... 28-year-old Moroccan (4)
attacks ATTACK ASSAILANT terrorist (5)
VICTIM ... (6)
TIME (exth.) 11 September 2001 (7)
sentence SENTENCING CONVICT Mounir al Motassadeq (8)
COURT Higher Regional Court (9)
TYPE ... maximum sentence (10)
prison PRISON INMATES ... Mounir al Motassadeq (11)DURATION (exth.) 15 years (12)
Found VERDICT CASE terrorist attacks (13)
CHARGE accessory to murder (14)
DEFENDANT 28-year-old Moroccan (15)
FINDING ... guilty (16)
accessory ASSISTANCE CO-AGENT (17)
FOCAL_ENTITY murder (18)
HELPER ... 28-year-old Moroccan (19)
murder KILLING KILLER (20)
VICTIM ... m.t. 3000 cases (21)
Target Frame Frame roles Filler (given vs. Induced)
trial TRIAL CASE terrorist attacks (1)
CHARGE accessory to murder (2)
COURT Higher Regional Court (3)
DEFENDANT ... 28-year-old Moroccan (4)
attacks ATTACK ASSAILANT terrorist (5)
VICTIM ... (6)
TIME (exth.) 11 September 2001 (7)
sentence SENTENCING CONVICT Mounir al Motassadeq (8)
COURT Higher Regional Court (9)
TYPE ... maximum sentence (10)
prison PRISON INMATES ... Mounir al Motassadeq (11)
DURATION (exth.) 15 years (12)
found VERDICT CASE terrorist attacks (13)
CHARGE accessory to murder (14)
DEFENDANT 28-year-old Moroccan (15)
FINDING ... guilty (16)
accessory ASSISTANCE CO-AGENT (17)
FOCAL_ENTITY murder (18)
HELPER ... 28-year-old Moroccan (19)
murder KILLING KILLER (20)
VICTIM ... m.t. 3000 cases (21)
Linking Frames and Roles in Context
At the instance level
• given frame instances f1:F1 and f2:F2, where
– f1 and f2 stand in a contextual relation (syn, sem, discourse)
– frame types F1 and F2 stand in some frame relation
=> identify role instances (referents) of f1 and f2 (r1 (= r0) = r2)
frame relation context-related instances inferred relation
Linking Frames and Roles in Context
In the first trial in the world in connection with the terrorist attacks of 11 September 2001, the Higher Regional Court of Hamburg has passed down the maximum sentence.
Criminal Process
Trial
SentencingCourt
frame relation
Linking Frames and Roles in Context
In the first trial (f1) in the world in connection with the terrorist attacks of 11 September 2001, [the Higher Regional Court of Hamburg] (r2) has passed down the maximum sentence (f2).
The Higher Regional Court of Hamburg
Functional Embedding
Criminal Process
Trial
SentencingCourt
frame relation context-related instances
Linking Frames and Roles in Context
The Higher Regional Court of Hamburg
Functional Embedding
Criminal Process
Trial
SentencingCourt
frame relation context-related instances inferred relation
In the first trial (f1) in the world in connection with the terrorist attacks of 11 September 2001, [the Higher Regional Court of Hamburg] (r2=r0= r1) has passed down the maximum sentence (f2).
Linking Frames and Roles in Context
At the type level (more involved)
• If instances of frame roles f1:F1 and f2:F2 are often found co-referent within particular contextual relations
=> Hypothesize a frame relation between F1 and F2
(no) frame relation context-related instances inferred relation
Linking Frames and Roles in Context
(no) frame relation context-related instances inferred relations
… the Higher Regional Court of Hamburg has passed down the Maximum sentence. [Mounir al Motassadeq] will spend 15 years in prison.
Sentencing
Prison
Convict
Inmates Discourse Relation
• New Frame Relation
• (Role Binding: Convict=Inmates)
(Co-reference)
CRIMINAL PROCESS
SENTENCING (1) TRIAL (1)
VERDICT (3)Defendant
Defendant
KILLING (3)
Inferred RelationContextual Relation Killer
Subframe/FE
PRISON (2)
Inmates Duration
ASSISTANCE (3)
Helper Co_agentFocal_entityVictim
Convict Type Court CaseCharge
CaseCharge
Court
Finding
(1) sentence number
Frame, Contextual, and Inferred Relations
CRIMINAL PROCESS
SENTENCING TRIAL
VERDICT
Defendant
Defendant(the Moroccan)
KILLING
InferenceContextual Relations
Killer
Hierarchy/Subframe/FE
PRISON
Inmates(Motus.)
Duration(15Y)
ASSISTANCE
Helper Co_agentGoal(murder)
Victim(3000)
Convict Duration(maximum)
Court(Hmbg.)
Case(9/11)
Charge
CaseCharge(accessory)
In the first trial .. the higher Regional Court .. has passed down the maximum sentence.Mounir al Motussadeq will spend 15 years in prison.The 28-year-old Moroccan was found guilty as an accessory to murder in .. 3000 cases.
Statistical Semantic Role Labeling
References
Jurafsky, D. & J. H. Martin, Speech and Language Processing, Prentice-Hall, 2000. (Chapters 9 and 10)
Helmreich, S., From Syntax to Semantics, Presentation in the 74.419 Course, November 2003.
Nirenburg, S. & V. Raskin, Ontological Semantics, MIT Press, 2004.
Wordnet, http://wordnet.princeton.edu/
Suggested Upper Merged Ontology (SUMO), http://www.ontologyportal.org/