SlideShare a Scribd company logo
Translation Memory Matching
and Retrieval using Deep
Semantics
Rohit Gupta (ESR4)
University of Wolverhampton
Long Short Term Memory
Networks
An	
  Unfolded	
  LSTM	
  Network	
  
Input	
  
Gate	
  
Forget	
  
Gate	
  
Output	
  
Gate	
  
Motivation
S1:	
  without	
  removing	
  punctuations,	
  S2:	
  removed	
  punctuations	
  
Motivation
S1:	
  without	
  removing	
  punctuations,	
  S2:	
  removed	
  punctuations	
  
Motivation
METEOR:	
  higher	
  is	
  better,	
  BLEU:	
  higher	
  is	
  better,	
  TER:	
  lower	
  is	
  better	
  
*	
  results	
  are	
  significant	
  (p<0.01)	
  
#Segments:	
  the	
  number	
  of	
  segments	
  for	
  which	
  edit-­‐distance	
  and	
  LSTM	
  match	
  differ	
  
	
  
Motivation
METEOR:	
  higher	
  is	
  better,	
  BLEU:	
  higher	
  is	
  better,	
  TER:	
  lower	
  is	
  better	
  
*	
  results	
  are	
  significant	
  (p<0.01)	
  
#Segments:	
  the	
  number	
  of	
  segments	
  for	
  which	
  edit-­‐distance	
  and	
  LSTM	
  match	
  differ	
  
	
  
METEOR	
  
edit-­‐
distance	
  
Motivation
METEOR:	
  higher	
  is	
  better,	
  BLEU:	
  higher	
  is	
  better,	
  TER:	
  lower	
  is	
  better	
  
*	
  results	
  are	
  significant	
  (p<0.01)	
  
#Segments:	
  the	
  number	
  of	
  segments	
  for	
  which	
  edit-­‐distance	
  and	
  LSTM	
  match	
  differ	
  
	
  
METEOR	
  
edit-­‐
distance	
  
METEOR	
  
our	
  LSTM	
  
model	
  
Thank you!

More Related Content

DOCX
Mca1010 fundamentals of computer and it
PDF
Chatbots and Deep Learning
PPTX
NLP_KASHK:Minimum Edit Distance
PDF
BERT - Part 2 Learning Notes
PDF
Santanu Pal - ESR 2 USAAR
PDF
ESR10 Joachim Daiber - EXPERT Summer School - Malaga 2015
PDF
5. Communication - Eunice Lawton (USFD)
PDF
ESR4 Rohit Gupta - EXPERT Summer School - Malaga 2015
Mca1010 fundamentals of computer and it
Chatbots and Deep Learning
NLP_KASHK:Minimum Edit Distance
BERT - Part 2 Learning Notes
Santanu Pal - ESR 2 USAAR
ESR10 Joachim Daiber - EXPERT Summer School - Malaga 2015
5. Communication - Eunice Lawton (USFD)
ESR4 Rohit Gupta - EXPERT Summer School - Malaga 2015

Viewers also liked (9)

PDF
Gianluca Giulinin - FAO
PDF
ESR8 Liangyou Li - EXPERT Summer School - Malaga 2015
PPT
Problems
PDF
Juanjo Arevelillo - Hermes Traducciones
PDF
Anna Zaretskaya - ESR 1 UMA
PDF
Tony O'Dowd - KantanMT
PDF
16. Anne Schumann (USAAR) Terminology and Ontologies 1
PDF
3. Natalia Konstantinova (UoW) EXPERT Introduction
PDF
18. Alessandro Cattelan (Translated) Terminology
Gianluca Giulinin - FAO
ESR8 Liangyou Li - EXPERT Summer School - Malaga 2015
Problems
Juanjo Arevelillo - Hermes Traducciones
Anna Zaretskaya - ESR 1 UMA
Tony O'Dowd - KantanMT
16. Anne Schumann (USAAR) Terminology and Ontologies 1
3. Natalia Konstantinova (UoW) EXPERT Introduction
18. Alessandro Cattelan (Translated) Terminology
Ad

Similar to Rohit Gupta - ESR 4 - UoW (7)

PDF
Text Tokenization
PPTX
Advance sql session - strings
PDF
Text summarization
PDF
IMPROVE THE QUALITY OF IMPORTANT SENTENCES FOR AUTOMATIC TEXT SUMMARIZATION
PPTX
Natural Language Query to SQL conversion using Machine Learning Approach
PDF
Conversational transfer learning for emotion recognition
PPTX
Aspect Extraction Performance With Common Pattern of Dependency Relation in ...
Text Tokenization
Advance sql session - strings
Text summarization
IMPROVE THE QUALITY OF IMPORTANT SENTENCES FOR AUTOMATIC TEXT SUMMARIZATION
Natural Language Query to SQL conversion using Machine Learning Approach
Conversational transfer learning for emotion recognition
Aspect Extraction Performance With Common Pattern of Dependency Relation in ...
Ad

More from RIILP (20)

PDF
Gabriella Gonzalez - eTRAD
PDF
Manuel Herranz - Pangeanic
PDF
Carla Parra Escartin - ER2 Hermes Traducciones
PDF
Lianet Sepulveda & Alexander Raginsky - ER 3a & ER 3b Pangeanic
PDF
Chris Hokamp - ESR 9 DCU
PDF
Carolina Scarton - ESR 7 - USFD
PDF
Hernani Costa - ESR 3 - UMA
PDF
Liangyou Li - ESR 8 - DCU
PDF
Liling Tan - ESR 5 USAAR
PDF
Sandra de luca - Acclaro
PDF
ER1 Eduard Barbu - EXPERT Summer School - Malaga 2015
PDF
ESR1 Anna Zaretskaya - EXPERT Summer School - Malaga 2015
PDF
ESR2 Santanu Pal - EXPERT Summer School - Malaga 2015
PDF
ESR3 Hernani Costa - EXPERT Summer School - Malaga 2015
PDF
ESR5 Liling Tan - EXPERT Summer School - Malaga 2015
PDF
ESR6 Varvara Logacheva - EXPERT Summer School - Malaga 2015
PDF
ESR7 Carolina Scarton - EXPERT Summer School - Malaga 2015
PDF
ESR11 Hoang Cuong - EXPERT Summer School - Malaga 2015
PDF
ESR12 Hanna Bechara - EXPERT Summer School - Malaga 2015
PDF
9. Ethics - Juan Jose Arevalillo Doval (Hermes)
Gabriella Gonzalez - eTRAD
Manuel Herranz - Pangeanic
Carla Parra Escartin - ER2 Hermes Traducciones
Lianet Sepulveda & Alexander Raginsky - ER 3a & ER 3b Pangeanic
Chris Hokamp - ESR 9 DCU
Carolina Scarton - ESR 7 - USFD
Hernani Costa - ESR 3 - UMA
Liangyou Li - ESR 8 - DCU
Liling Tan - ESR 5 USAAR
Sandra de luca - Acclaro
ER1 Eduard Barbu - EXPERT Summer School - Malaga 2015
ESR1 Anna Zaretskaya - EXPERT Summer School - Malaga 2015
ESR2 Santanu Pal - EXPERT Summer School - Malaga 2015
ESR3 Hernani Costa - EXPERT Summer School - Malaga 2015
ESR5 Liling Tan - EXPERT Summer School - Malaga 2015
ESR6 Varvara Logacheva - EXPERT Summer School - Malaga 2015
ESR7 Carolina Scarton - EXPERT Summer School - Malaga 2015
ESR11 Hoang Cuong - EXPERT Summer School - Malaga 2015
ESR12 Hanna Bechara - EXPERT Summer School - Malaga 2015
9. Ethics - Juan Jose Arevalillo Doval (Hermes)

Recently uploaded (20)

PDF
Global Data and Analytics Market Outlook Report
PPTX
Qualitative Qantitative and Mixed Methods.pptx
PPTX
Pilar Kemerdekaan dan Identi Bangsa.pptx
PDF
REAL ILLUMINATI AGENT IN KAMPALA UGANDA CALL ON+256765750853/0705037305
PPTX
retention in jsjsksksksnbsndjddjdnFPD.pptx
PPTX
CYBER SECURITY the Next Warefare Tactics
PDF
Data Engineering Interview Questions & Answers Cloud Data Stacks (AWS, Azure,...
PPTX
FMIS 108 and AISlaudon_mis17_ppt_ch11.pptx
PPTX
SAP 2 completion done . PRESENTATION.pptx
PDF
Data Engineering Interview Questions & Answers Batch Processing (Spark, Hadoo...
PDF
Navigating the Thai Supplements Landscape.pdf
PPTX
STERILIZATION AND DISINFECTION-1.ppthhhbx
PPTX
Introduction to Inferential Statistics.pptx
PPTX
Microsoft-Fabric-Unifying-Analytics-for-the-Modern-Enterprise Solution.pptx
PPTX
Copy of 16 Timeline & Flowchart Templates – HubSpot.pptx
PDF
Transcultural that can help you someday.
PPTX
Managing Community Partner Relationships
PDF
Jean-Georges Perrin - Spark in Action, Second Edition (2020, Manning Publicat...
PDF
Microsoft 365 products and services descrption
PDF
Introduction to Data Science and Data Analysis
Global Data and Analytics Market Outlook Report
Qualitative Qantitative and Mixed Methods.pptx
Pilar Kemerdekaan dan Identi Bangsa.pptx
REAL ILLUMINATI AGENT IN KAMPALA UGANDA CALL ON+256765750853/0705037305
retention in jsjsksksksnbsndjddjdnFPD.pptx
CYBER SECURITY the Next Warefare Tactics
Data Engineering Interview Questions & Answers Cloud Data Stacks (AWS, Azure,...
FMIS 108 and AISlaudon_mis17_ppt_ch11.pptx
SAP 2 completion done . PRESENTATION.pptx
Data Engineering Interview Questions & Answers Batch Processing (Spark, Hadoo...
Navigating the Thai Supplements Landscape.pdf
STERILIZATION AND DISINFECTION-1.ppthhhbx
Introduction to Inferential Statistics.pptx
Microsoft-Fabric-Unifying-Analytics-for-the-Modern-Enterprise Solution.pptx
Copy of 16 Timeline & Flowchart Templates – HubSpot.pptx
Transcultural that can help you someday.
Managing Community Partner Relationships
Jean-Georges Perrin - Spark in Action, Second Edition (2020, Manning Publicat...
Microsoft 365 products and services descrption
Introduction to Data Science and Data Analysis

Rohit Gupta - ESR 4 - UoW

  • 1. Translation Memory Matching and Retrieval using Deep Semantics Rohit Gupta (ESR4) University of Wolverhampton
  • 2. Long Short Term Memory Networks An  Unfolded  LSTM  Network   Input   Gate   Forget   Gate   Output   Gate  
  • 3. Motivation S1:  without  removing  punctuations,  S2:  removed  punctuations  
  • 4. Motivation S1:  without  removing  punctuations,  S2:  removed  punctuations  
  • 5. Motivation METEOR:  higher  is  better,  BLEU:  higher  is  better,  TER:  lower  is  better   *  results  are  significant  (p<0.01)   #Segments:  the  number  of  segments  for  which  edit-­‐distance  and  LSTM  match  differ    
  • 6. Motivation METEOR:  higher  is  better,  BLEU:  higher  is  better,  TER:  lower  is  better   *  results  are  significant  (p<0.01)   #Segments:  the  number  of  segments  for  which  edit-­‐distance  and  LSTM  match  differ     METEOR   edit-­‐ distance  
  • 7. Motivation METEOR:  higher  is  better,  BLEU:  higher  is  better,  TER:  lower  is  better   *  results  are  significant  (p<0.01)   #Segments:  the  number  of  segments  for  which  edit-­‐distance  and  LSTM  match  differ     METEOR   edit-­‐ distance   METEOR   our  LSTM   model