adapting text instead of the model : an open domain approach
DESCRIPTION
Adapting Text instead of the Model : An Open Domain Approach. Gourab Kundu, Dan Roth University of Illinois at Urbana-Champaign. Motivating Example # 1. predicate. Semantic role. Original Sentence. Wrong. Scotty gazed at ugly gray slums . AM-LOC. Transformed Sentence. - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/1.jpg)
1
Adapting Text instead of the Model : An Open Domain Approach
Gourab Kundu, Dan RothUniversity of Illinois at Urbana-Champaign
![Page 2: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/2.jpg)
2
Motivating Example #1
Scotty gazed at ugly gray slums .Original Sentence
Scotty looked at ugly gray slums .
Transformed Sentence
AM-LOC
A1
predicate
Semantic role
Wrong
Correct!
![Page 3: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/3.jpg)
3
Motivating Example #2
Original SentenceHe was discharged from the hospital after a two-day checkup and he and his parents had what Mr. Mckinley described as a “celebration lunch” in the campus.
AM-TMPPredicate Wron
g
![Page 4: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/4.jpg)
4
Transformed SentenceHe was discharged from the hospital after a two-day examination and he and his parents had what Mr. Mckinley described as a “celebration lunch” in the campus.
Motivating Example #2
Predicate AM-TMP
Correct!
![Page 5: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/5.jpg)
5
Research Question
Can text perturbation be done in an automatic way to yield better NLP analysis?
We study this question in the context of semantic role labeling.
We focus on improving the performance of SRL on a different domain
![Page 6: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/6.jpg)
6
Outline
Overview of Domain Adaptation Overview of Adaptation Using Transformations (ADUT) Transformation Functions Combination Strategy Experimental Results Conclusion
![Page 7: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/7.jpg)
7
Domain Adaptation
Models trained on one domain perform significantly worse on another domain Semantic Role Labeling: WSJ domain (76%), Fiction domain (65%)
Important Problem for wide scale NLP Adaptation is a problem for many tasks of NLP There are many different domains where natural language varies Labeling is expensive and time consuming
![Page 8: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/8.jpg)
8
Current Approaches to Domain Adaptation
Labeled Adaptation Uses labeled data from new domain
Unlabeled Adaptation Uses unlabeled data from new domain
Combined Adaptation Combines labeled and unlabeled data
• ChelbaAc04, Adaptation of a maximum entropy capitalizer: Little data can help a lot
• Daume07, Frustratingly Easy domain adaptation• FinkelMa09, Hierarchical Bayesian domain
adaptation
• BlitzerMcPe06, Domain Adaptation with Structural Correspondence Learning
• HuangYa09, Distributional Representations for Handling Sparsity in Supervised Sequence Labeling
• JiangZh07, Instance Weighting for Domain Adaptation in NLP
• ChangCoRo10, The necessity of combining adaptation methods
![Page 9: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/9.jpg)
Limitations: Need to retrain the model -- can take a long time
Limitations (Retraining takes time)
9
NLP Tool 1
NLP Tool N
NLP Tool 2
Model
Retrain
Retrain
Target Domain Unlabeled Data
Source DomainUnlabeled Data
SRL: 20 hours
![Page 10: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/10.jpg)
10
Limitations: Need to retrain other people’s tools -- may need implementation
Limitations (Some tools are hard to retrain)
NLP Tool 1
NLP Tool N
NLP Tool 2
Model
Target Domain Unlabeled Data
Source DomainUnlabeled Data
No option for retraining
![Page 11: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/11.jpg)
11
Limitations: Need significant unlabeled data -- may not be available (e.g. website)
Limitations (Insufficient Unlabeled Data)
NLP Tool 1
NLP Tool N
NLP Tool 2
Model
Target Domain Unlabeled Data
Source DomainUnlabeled Data
May not be sufficient
![Page 12: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/12.jpg)
12
Outline
Overview of Domain Adaptation Overview of Adaptation Using Transformations (ADUT) Transformation Functions Combination Strategy Experimental Results Conclusion
![Page 13: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/13.jpg)
13
ADaptation Using Transformations (ADUT)
t1
Combination Module
Transformation Module
Transformed Sentences
t2
tk
Model Outputs
o1
o2
ok
Output o… …
Tool Tool
Model
Sentence s
Old System
Traditional Approach: Adapt model for the new text Our Approach: Adapt text for the old model
![Page 14: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/14.jpg)
14
Transformation Functions
Definition: A Function that maps an instance to a set of instances
Example: Replacement of a word with synonyms that are common in training data
Properties: Label (Semantic role) Preserving Output examples are more likely to appear in Old Domain than input
example
![Page 15: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/15.jpg)
15
Categorization of Transformation Functions
Resource Based Transformation Uses resources and prior knowledge
Learned Transformations Learned from training data
![Page 16: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/16.jpg)
16
Resource Based Transformation
Replacement of Infrequent Predicate Replacement/Removal of Quoted String Replacement of Unknown Word (Word Cluster, WordNet) Sentence Simplification
![Page 17: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/17.jpg)
17
Replacement of Infrequent Predicate (VerbNet)
Intuition: Model makes better prediction over frequent predicates.
Scotty gazed at ugly gray slums .Input Sentence
Scotty looked at ugly gray slums .
Transformed Sentence
![Page 18: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/18.jpg)
18
Replacement/Removal of Quoted String
Intuition: Parser works better on simplified quoted sentences.
Input Sentence“We just sit quiet” , he said .
Transformed Sentences
We just sit quiet.
He said, “This is good”.
He said, “We just sit quiet”.
![Page 19: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/19.jpg)
19
Replacing Unknown Word(Word Cluster, WordNet)
Intuition: Parser & Model works better on known words.
Input Sentence
He was released after a two-day checkup.
Transformed SentenceHe was released after a two-day examination.
![Page 20: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/20.jpg)
20
Sentence Simplification (1)
Intuition: Parser & Model work better on simplified sentences.
Transformed Sentence
The science teacher and the students discussed the issue.
Input SentenceThe science teacher and the students discussed the issue at the classroom .
Delete PP
![Page 21: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/21.jpg)
21
Sentence Simplification (2)
Transformed Sentence
The teacher discussed the issue.
Input Sentence
The science teacher and the students discussed the issue. Simplify NP
![Page 22: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/22.jpg)
A2
Learned Transformation Rules
Motivation: Identify a specific context in the input sentence Transfer the candidate argument to a simpler context in which the SRL
is more robust
was entitled to a discount .
-2 -1 0 1 2.
Input SentenceMr. Mckinley
NP, Mckinley AUX, was PP, to
pattern p=[-2,NP,][-1,AUX,][1,,to]
![Page 23: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/23.jpg)
-2
Rule: predicate p=entitle pattern p=[-2,NP,][-1,AUX,][1,,to] Location of Source Phrase ns=-2 Replacement Sentence st=“But he did not sing.” Location of Replacement Phrase nt=-3 Label Correspondence function f={(A0,A2),(Ai,Ai, i0)}
Context Component of Rules
was entitled to a discount .
-1 0 1 2.
Input SentenceMr. Mckinley
![Page 24: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/24.jpg)
24
Replacement Component of Rules
Motivation:
Rule: predicate p=entitle pattern p=[-2,NP,][-1,AUX,][1,,to] Location of Source Phrase ns=-2
Replacement Sentence st=“But he did not sing.” Location of Replacement Phrase nt=-3 Label Correspondence function f={(A0,A2),(Ai,Ai, i0)}
did not sing .
-4 -3 -2 -1 0 1
Replacement SentenceBut he
![Page 25: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/25.jpg)
Rule: predicate p=entitle pattern p=[-2,NP,][-1,AUX,][1,,to] Location of Source Phrase ns=-2 Replacement Sentence st=“But he did not sing.” Location of Replacement Phrase nt=-3
Label Correspondence function f={(A0,A2),(Ai,Ai, i0)}
Semantic Role mapping component of Rule
25
was entitled to a discount .
-2 -1 0 1 2
Input Sentence Transformed Sentencedid not sing .
-4 -3 -2 -1 0 1
Replacement SentenceMr. Mckinley But he
Gold AnnotationA2 Apply SRL SystemA0
![Page 26: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/26.jpg)
26
for each phrase p in input sentence sfor each rule τ Є R
if τ applies to psentence t = transform(τ, p)r = semantic role of p in t using SRL modelsemantic role of p in s = map(τ, r)
Transforming a sentence by using rules
R is the set of rules, learned from training data
![Page 27: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/27.jpg)
27
Learning Transformation Rules
Input: Predicate p, Semantic role r R Get Initial Rules (p, r) repeat
S Expand Rules (R) Sort R S based on accuracy∪ R Top rules in R S∪
![Page 28: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/28.jpg)
28
Learning Transformation Rules
Input: Predicate p, Semantic role r R Get Initial Rules (p, r) repeat
S Expand Rules (R) Sort R S based on accuracy∪ R Top rules in R S∪
![Page 29: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/29.jpg)
29
Get Initial Rules (entitle, A2)
was entitled to a discount .did not sing .
Replacement SentenceMr. MckinleyBut he
A2
Replacement Sentence I asked the man .
![Page 30: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/30.jpg)
30
Learning Transformation Rules
Input: Predicate p, Semantic role r R Get Initial Rules (p, r) repeat
S Expand Rules (R) Sort R S based on accuracy∪ R Top rules in R S∪
![Page 31: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/31.jpg)
31
Expand Rule ( )𝜏Rule :
st = “But he did not sing .”nt = -3p = asksp = [-1,NP,I][0,VBD,asked][1,NP,man]ns = 1f = {(A0,A2), (Ai,Ai, i0)}
Neighbor Rule of :st = .st
nt = p = .psp=[-1,NP,][0,VBD,asked][1,NP,man]ns = .ns
f = .f
He asked the man
Does not apply
Applies
![Page 32: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/32.jpg)
32
Learning Transformation Rules
Input: Predicate p, Semantic role r R Get Initial Rules (p, r) repeat
S Expand Rules (R) Sort R S based on ∪ accuracy R Top rules in R S∪
![Page 33: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/33.jpg)
33
Calculate Accuracy ( )𝜏 Example: A rule is correct
33
was entitled to a discount .
-2
-1 0 1 2.
Input Sentence Transformed Sentencedid not sing .
-4 -3 -2 -1 0 1
Replacement SentenceMr. Mckinley But he
Gold Annotation
A2
Apply SRL System
A0
A2 = f (A0)Correct!
-2
![Page 34: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/34.jpg)
34
Calculate Accuracy ( )𝜏 Example: A rule makes a mistake
34
was entitled a big success .-2
-1 0 1 2.
Input Sentence Transformed Sentencedid not sing .
-4 -3 -2 -1 0 1
Replacement SentenceThe movie But he
Gold Annotation
A1
Apply SRL System
A0
A2 = f (A0)Wrong
-2
![Page 35: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/35.jpg)
35
Outline
Overview of Domain Adaptation Overview of Adaptation Using Transformation (ADUT) Transformation Functions Combination Strategy Experimental Results Conclusion
![Page 36: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/36.jpg)
Combination using Integer Linear Programming
Step1: Compute distribution of scores over labels for argument candidates Our SRL system classifies each phrase as a semantic role The system assigns a probability distribution over semantic roles for each argument For same argument in different sentences, compute the average
3636
Transformed Sentence 1
Scotty looked at ugly gray slums .
Transformed Sentence 2
Scotty gazed at ugly gray slums .
A1 0.3
AM-LOC 0.4
A1 0.6
AM-LOC 0.1
A1 0.45
AM-LOC 0.25Average
![Page 37: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/37.jpg)
37
Inference via Integer Linear Programming
37
Input Sentence
Scotty gazed at ugly gray slums.
Goal: Find maximum likely semantic role assignment to all arguments without violating the constraints
Solve an ILP
Example of a Constraint:Two arguments can not overlap
![Page 38: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/38.jpg)
38
Outline
Overview of Domain Adaptation Overview of Adaptation Using Transformations (ADUT) Transformation Functions Combination Strategy Experimental Results Conclusion
![Page 39: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/39.jpg)
39
Results for Single Parse System (F1)
Charniak Parse based SRL Stanford Parse based SRL
65.5
62.9
69.3(+3.8)
65.7(+2.8)
Baseline ADUT
![Page 40: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/40.jpg)
40
Results for Multi Parse System (1)
F1
67.8(-2.7)68.8(-1.7) 69.2(-1.3)
70.5
73.8(+3.3) (Retrain)
Punyakanok08 Toutanova08 Surdeanu07 (Cons)ADUT-Combined Huang10
![Page 41: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/41.jpg)
41
Effect of each Transformation
F1
65.566.1
66.8 6766.4 66.2
69.3
Baseline Replacement of Unknown wordsReplacement of Predicate Replacement of QuotesSentence Simplification Transformation By RulesTogether
![Page 42: Adapting Text instead of the Model : An Open Domain Approach](https://reader035.vdocuments.us/reader035/viewer/2022062814/56816731550346895ddbdcb2/html5/thumbnails/42.jpg)
42
Conclusion
Current Work We suggested a framework for adapting text to yield better SRL analysis We showed that adaptation is possible without retraining and unlabeled
data We showed that simple transformations yield 13% error reduction for SRL
Future Work: Applying framework to other domains and tasks Using unlabeled data to improve transformations
Thank You.