SlideShare a Scribd company logo
Discrimina)ve	
  Approach	
  to	
  Fill-­‐in-­‐the-­‐Blank	
  Quiz	
  Genera)on	
  for	
  Language	
  Learners	
  
	
  	
  	
  	
  Keisuke	
  Sakaguchi1,	
  Yuki	
  Arase2,	
  Mamoru	
  Komachi1	
  
	
  	
  	
  	
  	
  	
  	
  	
  1	
  Nara	
  Ins9tute	
  of	
  Science	
  and	
  Technology	
  (NAIST),	
  Japan	
  	
  	
  2	
  MicrosoD	
  Research	
  Asia,	
  China	
  
keisuke-sa@is.naist.jp, yukiar@microsoft.com, komachi@tmu.ac.jp	
Prior work!
v Thesaurus	
  	
  (Sumita	
  et	
  al.)	
  
v Roundtrip	
  transla9on	
  	
  (Dahlmeier	
  and	
  Ng.)	
  
à re-­‐ranking	
  by	
  genera9ve	
  LMs	
  
References!
Summary!
v  Charles	
  Alderson,	
  Caroline	
  Clapham,	
  and	
  Dianne	
  Wall.	
  1995.	
  Language	
  Test	
  Construc/on	
  and	
  Evalua/on.	
  Cambridge	
  University	
  Press.	
  
v  Daniel	
  Dahlmeier	
  and	
  Hwee	
  Tou	
  Ng.	
  2011.	
  Correc9ng	
  seman9c	
  colloca9on	
  errors	
  with	
  L1-­‐induced	
  paraphrases.	
  In	
  Proceedings	
  of	
  the	
  2011	
  Conference	
  on	
  Empirical	
  Methods	
  in	
  Natural	
  Language	
  Processing,	
  pages	
  107–117,	
  Edinburgh,	
  Scotland,	
  UK.,	
  July.	
  	
  
v  Eiichiro	
  Sumita,	
  Fumiaki	
  Sugaya,	
  and	
  Seiichi	
  Yamamoto.	
  2005.	
  Measuring	
  Non-­‐na9ve	
  Speakers’	
  Proficiency	
  of	
  English	
  by	
  Using	
  a	
  Test	
  with	
  Automa9cally-­‐Generated	
  Fill-­‐in-­‐the-­‐Blank	
  Ques9ons.	
  In	
  Proceedings	
  of	
  the	
  2nd	
  Workshop	
  on	
  Building	
  Educa/onal	
  Applica/ons	
  Using	
  NLP,	
  pages	
  61–	
  68,	
  Ann	
  Arbor,	
  June.	
  	
  
Generate	
  more	
  reliable	
  and	
  valid	
  distractors	
  using	
  
1.	
  Large-­‐scale	
  ESL	
  corpus	
  	
  
2.	
  Discrimina9ve	
  models	
  
v Fill-­‐in-­‐the-­‐blank	
  quiz	
  for	
  ESL	
  learners.	
  
v Good	
  (seman9c)	
  distractors	
  (Alderson	
  et	
  al.)	
  
	
  -­‐	
  reliable:	
  exclusive	
  against	
  the	
  correct	
  answer	
  
	
  -­‐	
  valid:	
  discriminate	
  learners’	
  proficiency	
  
v Reliability	
  
-­‐	
  3	
  na9ve	
  speakers	
  
1.	
  Ra9o	
  of	
  Appropriate	
  	
  
Distractors	
  =	
  	
  
※	
  NAD:	
  #	
  of	
  quizzes	
  that	
  2+	
  
	
  par9cipants	
  agree	
  on	
  	
  
2.	
  Inter-­‐rater	
  agreement	
  κ	
  	
  
v Validity	
  
-­‐	
  23	
  Japanese	
  ESL	
  learners	
  
1.	
  Correla9on	
  Coefficient	
  (r)	
  	
  
Proposed Method!
v Confusion	
  Matrix	
  Method	
  
v Discrimina9ve	
  ESL	
  Method	
  
Confusion	
  matrix	
  from	
  ESL	
  corpus	
  (Lang-­‐8)	
  
Classifier	
  for	
  each	
  target	
  (trained	
  on	
  ESL	
  corpus)	
  
v Discrimina9ve	
  Simulated-­‐ESL	
  Method	
  
Classifier	
  for	
  each	
  target	
  	
  
(trained	
  on	
  Pseudo-­‐ESL	
  corpus)	
  
Features:	
  ±1	
  lemma,	
  ±2	
  lemma,	
  dependency	
  
Label:	
  generated	
  from	
  confusion	
  matrix	
  
Features:	
  ±1	
  lemma,	
  ±2	
  lemma,	
  dependency	
  
Label:	
  original	
  incorrect	
  verb	
  in	
  Lang-­‐8	
  	
  
Method	
 Corpus	
 Model	
 RAD	
 κ	
 r	
Confu9on	
  Mat.	
ESL	
 Genera9ve	
 94.5	
 0.55	
 0.71	
Disc.	
  ESL	
 ESL	
 Discrimina9ve	
 95.0	
 0.73	
 0.48	
Disc.	
  Sim-­‐ESL	
 Pseudo-­‐ESL	
 Discrimina9ve	
 98.3	
 0.69	
 0.76	
Thesaurus	
 Na9ve	
 Genera9ve	
   89.3	
 0.57	
 0.68	
Roundtrip	
 Na9ve	
 Genera9ve	
 93.6	
 0.53	
 0.67	
Experiment and Result!

More Related Content

Similar to ACL13_sakaguchi

Types of Adjectives 5th grade
Types of Adjectives 5th gradeTypes of Adjectives 5th grade
Types of Adjectives 5th grade
SelamAmanuel1
 
Inductive learning of long-distance dissimilation as a problem for phonology
Inductive learning of long-distance dissimilation as a problem for phonologyInductive learning of long-distance dissimilation as a problem for phonology
Inductive learning of long-distance dissimilation as a problem for phonology
Kevin McMullin
 
Tooltip-type, Frame-type, and Concordance Glossing in L2 Reading
Tooltip-type, Frame-type, and Concordance Glossing in L2 ReadingTooltip-type, Frame-type, and Concordance Glossing in L2 Reading
Tooltip-type, Frame-type, and Concordance Glossing in L2 Reading
engedukamall
 
Fast evaluation of Connectionist Language Models
Fast evaluation of Connectionist Language ModelsFast evaluation of Connectionist Language Models
Fast evaluation of Connectionist Language Models
Francisco Zamora-Martinez
 
MATH 485_Poster_Team V
MATH 485_Poster_Team VMATH 485_Poster_Team V
MATH 485_Poster_Team Vsana Habib
 
Essential Biology 10.2 Dihybrid Crosses & Gene Linkage (AHL)
Essential Biology 10.2 Dihybrid Crosses & Gene Linkage (AHL)Essential Biology 10.2 Dihybrid Crosses & Gene Linkage (AHL)
Essential Biology 10.2 Dihybrid Crosses & Gene Linkage (AHL)
Stephen Taylor
 
Nature of written langauge problems in children
Nature of written langauge problems in childrenNature of written langauge problems in children
Nature of written langauge problems in children
HillarySang4
 
NEST 2014 Question Paper
NEST 2014 Question PaperNEST 2014 Question Paper
NEST 2014 Question Paper
Eneutron
 
Colloquium talk on modal sense classification using a convolutional neural ne...
Colloquium talk on modal sense classification using a convolutional neural ne...Colloquium talk on modal sense classification using a convolutional neural ne...
Colloquium talk on modal sense classification using a convolutional neural ne...
Ana Marasović
 
Ekaterina vylomova-what-do-neural models-know-about-language-p1
Ekaterina vylomova-what-do-neural models-know-about-language-p1Ekaterina vylomova-what-do-neural models-know-about-language-p1
Ekaterina vylomova-what-do-neural models-know-about-language-p1
Katerina Vylomova
 
Master defence 2020 - Anastasiia Khaburska - Statistical and Neural Language ...
Master defence 2020 - Anastasiia Khaburska - Statistical and Neural Language ...Master defence 2020 - Anastasiia Khaburska - Statistical and Neural Language ...
Master defence 2020 - Anastasiia Khaburska - Statistical and Neural Language ...
Lviv Data Science Summer School
 
Mixed Effects Models - Crossed Random Effects
Mixed Effects Models - Crossed Random EffectsMixed Effects Models - Crossed Random Effects
Mixed Effects Models - Crossed Random Effects
Scott Fraundorf
 
Estimation of Severity of Speech Disability Through Speech Envelope
Estimation of Severity of Speech Disability Through Speech EnvelopeEstimation of Severity of Speech Disability Through Speech Envelope
Estimation of Severity of Speech Disability Through Speech Envelope
sipij
 
Painless grammar/Adviser Institute 2015
Painless grammar/Adviser Institute 2015Painless grammar/Adviser Institute 2015
Painless grammar/Adviser Institute 2015Candace Bowen
 
Thamme Gowda's PhD dissertation defense slides
Thamme Gowda's PhD dissertation defense slidesThamme Gowda's PhD dissertation defense slides
Thamme Gowda's PhD dissertation defense slides
Thamme Gowda
 
Using the Common Core
Using the Common CoreUsing the Common Core

Similar to ACL13_sakaguchi (20)

Language models
Language modelsLanguage models
Language models
 
Types of Adjectives 5th grade
Types of Adjectives 5th gradeTypes of Adjectives 5th grade
Types of Adjectives 5th grade
 
Inductive learning of long-distance dissimilation as a problem for phonology
Inductive learning of long-distance dissimilation as a problem for phonologyInductive learning of long-distance dissimilation as a problem for phonology
Inductive learning of long-distance dissimilation as a problem for phonology
 
Tooltip-type, Frame-type, and Concordance Glossing in L2 Reading
Tooltip-type, Frame-type, and Concordance Glossing in L2 ReadingTooltip-type, Frame-type, and Concordance Glossing in L2 Reading
Tooltip-type, Frame-type, and Concordance Glossing in L2 Reading
 
Anandkumar novel approach
Anandkumar novel approachAnandkumar novel approach
Anandkumar novel approach
 
Fast evaluation of Connectionist Language Models
Fast evaluation of Connectionist Language ModelsFast evaluation of Connectionist Language Models
Fast evaluation of Connectionist Language Models
 
MATH 485_Poster_Team V
MATH 485_Poster_Team VMATH 485_Poster_Team V
MATH 485_Poster_Team V
 
Essential Biology 10.2 Dihybrid Crosses & Gene Linkage (AHL)
Essential Biology 10.2 Dihybrid Crosses & Gene Linkage (AHL)Essential Biology 10.2 Dihybrid Crosses & Gene Linkage (AHL)
Essential Biology 10.2 Dihybrid Crosses & Gene Linkage (AHL)
 
Nature of written langauge problems in children
Nature of written langauge problems in childrenNature of written langauge problems in children
Nature of written langauge problems in children
 
NEST 2014 Question Paper
NEST 2014 Question PaperNEST 2014 Question Paper
NEST 2014 Question Paper
 
Syntactic parsing for arabic
Syntactic parsing for arabicSyntactic parsing for arabic
Syntactic parsing for arabic
 
Colloquium talk on modal sense classification using a convolutional neural ne...
Colloquium talk on modal sense classification using a convolutional neural ne...Colloquium talk on modal sense classification using a convolutional neural ne...
Colloquium talk on modal sense classification using a convolutional neural ne...
 
C14-1028
C14-1028C14-1028
C14-1028
 
Ekaterina vylomova-what-do-neural models-know-about-language-p1
Ekaterina vylomova-what-do-neural models-know-about-language-p1Ekaterina vylomova-what-do-neural models-know-about-language-p1
Ekaterina vylomova-what-do-neural models-know-about-language-p1
 
Master defence 2020 - Anastasiia Khaburska - Statistical and Neural Language ...
Master defence 2020 - Anastasiia Khaburska - Statistical and Neural Language ...Master defence 2020 - Anastasiia Khaburska - Statistical and Neural Language ...
Master defence 2020 - Anastasiia Khaburska - Statistical and Neural Language ...
 
Mixed Effects Models - Crossed Random Effects
Mixed Effects Models - Crossed Random EffectsMixed Effects Models - Crossed Random Effects
Mixed Effects Models - Crossed Random Effects
 
Estimation of Severity of Speech Disability Through Speech Envelope
Estimation of Severity of Speech Disability Through Speech EnvelopeEstimation of Severity of Speech Disability Through Speech Envelope
Estimation of Severity of Speech Disability Through Speech Envelope
 
Painless grammar/Adviser Institute 2015
Painless grammar/Adviser Institute 2015Painless grammar/Adviser Institute 2015
Painless grammar/Adviser Institute 2015
 
Thamme Gowda's PhD dissertation defense slides
Thamme Gowda's PhD dissertation defense slidesThamme Gowda's PhD dissertation defense slides
Thamme Gowda's PhD dissertation defense slides
 
Using the Common Core
Using the Common CoreUsing the Common Core
Using the Common Core
 

More from Keisuke Sakaguchi

EMNLP 2021 proScript (summary slides)
EMNLP 2021 proScript (summary slides)EMNLP 2021 proScript (summary slides)
EMNLP 2021 proScript (summary slides)
Keisuke Sakaguchi
 
EMNLP 2021 proScript
EMNLP 2021 proScriptEMNLP 2021 proScript
EMNLP 2021 proScript
Keisuke Sakaguchi
 
Acl18 sakaguchi
Acl18 sakaguchiAcl18 sakaguchi
Acl18 sakaguchi
Keisuke Sakaguchi
 
Ijcnlp17 sakaguchi
Ijcnlp17 sakaguchiIjcnlp17 sakaguchi
Ijcnlp17 sakaguchi
Keisuke Sakaguchi
 
ACL17_Sakaguchi
ACL17_SakaguchiACL17_Sakaguchi
ACL17_Sakaguchi
Keisuke Sakaguchi
 
TACL16_Sakaguchi
TACL16_SakaguchiTACL16_Sakaguchi
TACL16_Sakaguchi
Keisuke Sakaguchi
 
WMT14_sakaguchi
WMT14_sakaguchiWMT14_sakaguchi
WMT14_sakaguchi
Keisuke Sakaguchi
 
COLING12_sakaguchi
COLING12_sakaguchiCOLING12_sakaguchi
COLING12_sakaguchi
Keisuke Sakaguchi
 

More from Keisuke Sakaguchi (8)

EMNLP 2021 proScript (summary slides)
EMNLP 2021 proScript (summary slides)EMNLP 2021 proScript (summary slides)
EMNLP 2021 proScript (summary slides)
 
EMNLP 2021 proScript
EMNLP 2021 proScriptEMNLP 2021 proScript
EMNLP 2021 proScript
 
Acl18 sakaguchi
Acl18 sakaguchiAcl18 sakaguchi
Acl18 sakaguchi
 
Ijcnlp17 sakaguchi
Ijcnlp17 sakaguchiIjcnlp17 sakaguchi
Ijcnlp17 sakaguchi
 
ACL17_Sakaguchi
ACL17_SakaguchiACL17_Sakaguchi
ACL17_Sakaguchi
 
TACL16_Sakaguchi
TACL16_SakaguchiTACL16_Sakaguchi
TACL16_Sakaguchi
 
WMT14_sakaguchi
WMT14_sakaguchiWMT14_sakaguchi
WMT14_sakaguchi
 
COLING12_sakaguchi
COLING12_sakaguchiCOLING12_sakaguchi
COLING12_sakaguchi
 

Recently uploaded

María Carolina Martínez - eCommerce Day Colombia 2024
María Carolina Martínez - eCommerce Day Colombia 2024María Carolina Martínez - eCommerce Day Colombia 2024
María Carolina Martínez - eCommerce Day Colombia 2024
eCommerce Institute
 
0x01 - Newton's Third Law: Static vs. Dynamic Abusers
0x01 - Newton's Third Law:  Static vs. Dynamic Abusers0x01 - Newton's Third Law:  Static vs. Dynamic Abusers
0x01 - Newton's Third Law: Static vs. Dynamic Abusers
OWASP Beja
 
Burning Issue Presentation By Kenmaryon.pdf
Burning Issue Presentation By Kenmaryon.pdfBurning Issue Presentation By Kenmaryon.pdf
Burning Issue Presentation By Kenmaryon.pdf
kkirkland2
 
Bonzo subscription_hjjjjjjjj5hhhhhhh_2024.pdf
Bonzo subscription_hjjjjjjjj5hhhhhhh_2024.pdfBonzo subscription_hjjjjjjjj5hhhhhhh_2024.pdf
Bonzo subscription_hjjjjjjjj5hhhhhhh_2024.pdf
khadija278284
 
Obesity causes and management and associated medical conditions
Obesity causes and management and associated medical conditionsObesity causes and management and associated medical conditions
Obesity causes and management and associated medical conditions
Faculty of Medicine And Health Sciences
 
Gregory Harris' Civics Presentation.pptx
Gregory Harris' Civics Presentation.pptxGregory Harris' Civics Presentation.pptx
Gregory Harris' Civics Presentation.pptx
gharris9
 
Announcement of 18th IEEE International Conference on Software Testing, Verif...
Announcement of 18th IEEE International Conference on Software Testing, Verif...Announcement of 18th IEEE International Conference on Software Testing, Verif...
Announcement of 18th IEEE International Conference on Software Testing, Verif...
Sebastiano Panichella
 
Tom tresser burning issue.pptx My Burning issue
Tom tresser burning issue.pptx My Burning issueTom tresser burning issue.pptx My Burning issue
Tom tresser burning issue.pptx My Burning issue
amekonnen
 
Getting started with Amazon Bedrock Studio and Control Tower
Getting started with Amazon Bedrock Studio and Control TowerGetting started with Amazon Bedrock Studio and Control Tower
Getting started with Amazon Bedrock Studio and Control Tower
Vladimir Samoylov
 
Doctoral Symposium at the 17th IEEE International Conference on Software Test...
Doctoral Symposium at the 17th IEEE International Conference on Software Test...Doctoral Symposium at the 17th IEEE International Conference on Software Test...
Doctoral Symposium at the 17th IEEE International Conference on Software Test...
Sebastiano Panichella
 
Presentatie 4. Jochen Cremer - TU Delft 28 mei 2024
Presentatie 4. Jochen Cremer - TU Delft 28 mei 2024Presentatie 4. Jochen Cremer - TU Delft 28 mei 2024
Presentatie 4. Jochen Cremer - TU Delft 28 mei 2024
Dutch Power
 
Presentatie 8. Joost van der Linde & Daniel Anderton - Eliq 28 mei 2024
Presentatie 8. Joost van der Linde & Daniel Anderton - Eliq 28 mei 2024Presentatie 8. Joost van der Linde & Daniel Anderton - Eliq 28 mei 2024
Presentatie 8. Joost van der Linde & Daniel Anderton - Eliq 28 mei 2024
Dutch Power
 
International Workshop on Artificial Intelligence in Software Testing
International Workshop on Artificial Intelligence in Software TestingInternational Workshop on Artificial Intelligence in Software Testing
International Workshop on Artificial Intelligence in Software Testing
Sebastiano Panichella
 
AWANG ANIQKMALBIN AWANG TAJUDIN B22080004 ASSIGNMENT 2 MPU3193 PHILOSOPHY AND...
AWANG ANIQKMALBIN AWANG TAJUDIN B22080004 ASSIGNMENT 2 MPU3193 PHILOSOPHY AND...AWANG ANIQKMALBIN AWANG TAJUDIN B22080004 ASSIGNMENT 2 MPU3193 PHILOSOPHY AND...
AWANG ANIQKMALBIN AWANG TAJUDIN B22080004 ASSIGNMENT 2 MPU3193 PHILOSOPHY AND...
AwangAniqkmals
 
Acorn Recovery: Restore IT infra within minutes
Acorn Recovery: Restore IT infra within minutesAcorn Recovery: Restore IT infra within minutes
Acorn Recovery: Restore IT infra within minutes
IP ServerOne
 
somanykidsbutsofewfathers-140705000023-phpapp02.pptx
somanykidsbutsofewfathers-140705000023-phpapp02.pptxsomanykidsbutsofewfathers-140705000023-phpapp02.pptx
somanykidsbutsofewfathers-140705000023-phpapp02.pptx
Howard Spence
 
Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...
Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...
Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...
OECD Directorate for Financial and Enterprise Affairs
 
Media as a Mind Controlling Strategy In Old and Modern Era
Media as a Mind Controlling Strategy In Old and Modern EraMedia as a Mind Controlling Strategy In Old and Modern Era
Media as a Mind Controlling Strategy In Old and Modern Era
faizulhassanfaiz1670
 
Bitcoin Lightning wallet and tic-tac-toe game XOXO
Bitcoin Lightning wallet and tic-tac-toe game XOXOBitcoin Lightning wallet and tic-tac-toe game XOXO
Bitcoin Lightning wallet and tic-tac-toe game XOXO
Matjaž Lipuš
 
Supercharge your AI - SSP Industry Breakout Session 2024-v2_1.pdf
Supercharge your AI - SSP Industry Breakout Session 2024-v2_1.pdfSupercharge your AI - SSP Industry Breakout Session 2024-v2_1.pdf
Supercharge your AI - SSP Industry Breakout Session 2024-v2_1.pdf
Access Innovations, Inc.
 

Recently uploaded (20)

María Carolina Martínez - eCommerce Day Colombia 2024
María Carolina Martínez - eCommerce Day Colombia 2024María Carolina Martínez - eCommerce Day Colombia 2024
María Carolina Martínez - eCommerce Day Colombia 2024
 
0x01 - Newton's Third Law: Static vs. Dynamic Abusers
0x01 - Newton's Third Law:  Static vs. Dynamic Abusers0x01 - Newton's Third Law:  Static vs. Dynamic Abusers
0x01 - Newton's Third Law: Static vs. Dynamic Abusers
 
Burning Issue Presentation By Kenmaryon.pdf
Burning Issue Presentation By Kenmaryon.pdfBurning Issue Presentation By Kenmaryon.pdf
Burning Issue Presentation By Kenmaryon.pdf
 
Bonzo subscription_hjjjjjjjj5hhhhhhh_2024.pdf
Bonzo subscription_hjjjjjjjj5hhhhhhh_2024.pdfBonzo subscription_hjjjjjjjj5hhhhhhh_2024.pdf
Bonzo subscription_hjjjjjjjj5hhhhhhh_2024.pdf
 
Obesity causes and management and associated medical conditions
Obesity causes and management and associated medical conditionsObesity causes and management and associated medical conditions
Obesity causes and management and associated medical conditions
 
Gregory Harris' Civics Presentation.pptx
Gregory Harris' Civics Presentation.pptxGregory Harris' Civics Presentation.pptx
Gregory Harris' Civics Presentation.pptx
 
Announcement of 18th IEEE International Conference on Software Testing, Verif...
Announcement of 18th IEEE International Conference on Software Testing, Verif...Announcement of 18th IEEE International Conference on Software Testing, Verif...
Announcement of 18th IEEE International Conference on Software Testing, Verif...
 
Tom tresser burning issue.pptx My Burning issue
Tom tresser burning issue.pptx My Burning issueTom tresser burning issue.pptx My Burning issue
Tom tresser burning issue.pptx My Burning issue
 
Getting started with Amazon Bedrock Studio and Control Tower
Getting started with Amazon Bedrock Studio and Control TowerGetting started with Amazon Bedrock Studio and Control Tower
Getting started with Amazon Bedrock Studio and Control Tower
 
Doctoral Symposium at the 17th IEEE International Conference on Software Test...
Doctoral Symposium at the 17th IEEE International Conference on Software Test...Doctoral Symposium at the 17th IEEE International Conference on Software Test...
Doctoral Symposium at the 17th IEEE International Conference on Software Test...
 
Presentatie 4. Jochen Cremer - TU Delft 28 mei 2024
Presentatie 4. Jochen Cremer - TU Delft 28 mei 2024Presentatie 4. Jochen Cremer - TU Delft 28 mei 2024
Presentatie 4. Jochen Cremer - TU Delft 28 mei 2024
 
Presentatie 8. Joost van der Linde & Daniel Anderton - Eliq 28 mei 2024
Presentatie 8. Joost van der Linde & Daniel Anderton - Eliq 28 mei 2024Presentatie 8. Joost van der Linde & Daniel Anderton - Eliq 28 mei 2024
Presentatie 8. Joost van der Linde & Daniel Anderton - Eliq 28 mei 2024
 
International Workshop on Artificial Intelligence in Software Testing
International Workshop on Artificial Intelligence in Software TestingInternational Workshop on Artificial Intelligence in Software Testing
International Workshop on Artificial Intelligence in Software Testing
 
AWANG ANIQKMALBIN AWANG TAJUDIN B22080004 ASSIGNMENT 2 MPU3193 PHILOSOPHY AND...
AWANG ANIQKMALBIN AWANG TAJUDIN B22080004 ASSIGNMENT 2 MPU3193 PHILOSOPHY AND...AWANG ANIQKMALBIN AWANG TAJUDIN B22080004 ASSIGNMENT 2 MPU3193 PHILOSOPHY AND...
AWANG ANIQKMALBIN AWANG TAJUDIN B22080004 ASSIGNMENT 2 MPU3193 PHILOSOPHY AND...
 
Acorn Recovery: Restore IT infra within minutes
Acorn Recovery: Restore IT infra within minutesAcorn Recovery: Restore IT infra within minutes
Acorn Recovery: Restore IT infra within minutes
 
somanykidsbutsofewfathers-140705000023-phpapp02.pptx
somanykidsbutsofewfathers-140705000023-phpapp02.pptxsomanykidsbutsofewfathers-140705000023-phpapp02.pptx
somanykidsbutsofewfathers-140705000023-phpapp02.pptx
 
Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...
Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...
Competition and Regulation in Professional Services – KLEINER – June 2024 OEC...
 
Media as a Mind Controlling Strategy In Old and Modern Era
Media as a Mind Controlling Strategy In Old and Modern EraMedia as a Mind Controlling Strategy In Old and Modern Era
Media as a Mind Controlling Strategy In Old and Modern Era
 
Bitcoin Lightning wallet and tic-tac-toe game XOXO
Bitcoin Lightning wallet and tic-tac-toe game XOXOBitcoin Lightning wallet and tic-tac-toe game XOXO
Bitcoin Lightning wallet and tic-tac-toe game XOXO
 
Supercharge your AI - SSP Industry Breakout Session 2024-v2_1.pdf
Supercharge your AI - SSP Industry Breakout Session 2024-v2_1.pdfSupercharge your AI - SSP Industry Breakout Session 2024-v2_1.pdf
Supercharge your AI - SSP Industry Breakout Session 2024-v2_1.pdf
 

ACL13_sakaguchi

  • 1. Discrimina)ve  Approach  to  Fill-­‐in-­‐the-­‐Blank  Quiz  Genera)on  for  Language  Learners          Keisuke  Sakaguchi1,  Yuki  Arase2,  Mamoru  Komachi1                  1  Nara  Ins9tute  of  Science  and  Technology  (NAIST),  Japan      2  MicrosoD  Research  Asia,  China   keisuke-sa@is.naist.jp, yukiar@microsoft.com, komachi@tmu.ac.jp Prior work! v Thesaurus    (Sumita  et  al.)   v Roundtrip  transla9on    (Dahlmeier  and  Ng.)   à re-­‐ranking  by  genera9ve  LMs   References! Summary! v  Charles  Alderson,  Caroline  Clapham,  and  Dianne  Wall.  1995.  Language  Test  Construc/on  and  Evalua/on.  Cambridge  University  Press.   v  Daniel  Dahlmeier  and  Hwee  Tou  Ng.  2011.  Correc9ng  seman9c  colloca9on  errors  with  L1-­‐induced  paraphrases.  In  Proceedings  of  the  2011  Conference  on  Empirical  Methods  in  Natural  Language  Processing,  pages  107–117,  Edinburgh,  Scotland,  UK.,  July.     v  Eiichiro  Sumita,  Fumiaki  Sugaya,  and  Seiichi  Yamamoto.  2005.  Measuring  Non-­‐na9ve  Speakers’  Proficiency  of  English  by  Using  a  Test  with  Automa9cally-­‐Generated  Fill-­‐in-­‐the-­‐Blank  Ques9ons.  In  Proceedings  of  the  2nd  Workshop  on  Building  Educa/onal  Applica/ons  Using  NLP,  pages  61–  68,  Ann  Arbor,  June.     Generate  more  reliable  and  valid  distractors  using   1.  Large-­‐scale  ESL  corpus     2.  Discrimina9ve  models   v Fill-­‐in-­‐the-­‐blank  quiz  for  ESL  learners.   v Good  (seman9c)  distractors  (Alderson  et  al.)    -­‐  reliable:  exclusive  against  the  correct  answer    -­‐  valid:  discriminate  learners’  proficiency   v Reliability   -­‐  3  na9ve  speakers   1.  Ra9o  of  Appropriate     Distractors  =     ※  NAD:  #  of  quizzes  that  2+    par9cipants  agree  on     2.  Inter-­‐rater  agreement  κ     v Validity   -­‐  23  Japanese  ESL  learners   1.  Correla9on  Coefficient  (r)     Proposed Method! v Confusion  Matrix  Method   v Discrimina9ve  ESL  Method   Confusion  matrix  from  ESL  corpus  (Lang-­‐8)   Classifier  for  each  target  (trained  on  ESL  corpus)   v Discrimina9ve  Simulated-­‐ESL  Method   Classifier  for  each  target     (trained  on  Pseudo-­‐ESL  corpus)   Features:  ±1  lemma,  ±2  lemma,  dependency   Label:  generated  from  confusion  matrix   Features:  ±1  lemma,  ±2  lemma,  dependency   Label:  original  incorrect  verb  in  Lang-­‐8     Method Corpus Model RAD κ r Confu9on  Mat. ESL Genera9ve 94.5 0.55 0.71 Disc.  ESL ESL Discrimina9ve 95.0 0.73 0.48 Disc.  Sim-­‐ESL Pseudo-­‐ESL Discrimina9ve 98.3 0.69 0.76 Thesaurus Na9ve Genera9ve   89.3 0.57 0.68 Roundtrip Na9ve Genera9ve 93.6 0.53 0.67 Experiment and Result!