span based constituency parsing with provably optimal
play

Span-Based Constituency Parsing with Provably Optimal Dynamic - PowerPoint PPT Presentation

Nov 06, 2023 •102 likes •823 views

Span-Based Constituency Parsing with Provably Optimal Dynamic Oracles James Cross and Liang Huang Oregon State University EMNLP, Austin, TX November 2, 2016 Dependency vs. Constituency S ROOT NP VP PRP MD VBP S subj dobj dobj aux

  1. Span-Based Constituency Parsing with Provably Optimal Dynamic Oracles James Cross and Liang Huang Oregon State University EMNLP, Austin, TX November 2, 2016

  2. Dependency vs. Constituency S ROOT NP VP PRP MD VBP S subj dobj dobj aux I do like VP I do like eating fish VBG NP PRP MD VBP VBG NN eating NN fish search F 1 search UAS Carreras et al. (2008) cubic 91.1 Zhang & Nivre 2011 beam 92.9 Chen & Manning 2014 greedy 91.8 Shindo et al. (2012) cubic 91.1 Zhou et al. (2015) beam 93.3 Thang et al. (2015) (A*) ~cubic 91.1 Weiss et al. (2015) beam 94.0 Watanabe et al. (2015) beam 90.7 our work (ACL 2016) greedy 93.4 Vinyals et al. (2015) (WSJ) beam 90.5 Andor et al. (2016) beam 94.4 Red = Neural 2

  3. Dependency vs. Constituency S ROOT NP VP PRP MD VBP S subj dobj dobj aux I do like VP I do like eating fish VBG NP PRP MD VBP VBG NN eating NN fish search F 1 search UAS Carreras et al. (2008) cubic 91.1 Zhang & Nivre 2011 beam 92.9 Chen & Manning 2014 greedy 91.8 Shindo et al. (2012) cubic 91.1 Zhou et al. (2015) beam 93.3 Thang et al. (2015) (A*) ~cubic 91.1 Weiss et al. (2015) beam 94.0 Watanabe et al. (2015) beam 90.7 our work (ACL 2016) greedy 93.4 Vinyals et al. (2015) (WSJ) beam 90.5 Andor et al. (2016) beam 94.4 This Work greedy 91.3 Red = Neural 2

  4. Outline • Span-Based Constituency Parsing • Bi-Directional LSTM Span Features • Provably Optimal Dynamic Oracle • Experiments 3

  5. Span-Based Parsing • Previous work uses tree structures on stack • We simplify to operate directly on sentence spans • Simple-to-implement linear-time parsing Stack Queue NP VP’ previous work I/PRP do/MD like/VBP eating/VBG fish/NN Stack Queue our work I/PRP do/MD like/VBP eating/VBG fish/NN 0 1 3 4 5 4

  6. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish current I/PRP do/MD like/VBP eating/VBG fish/NN t = {} brackets 0 1 2 3 4 5 5

  7. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish current I/PRP do/MD like/VBP eating/VBG fish/NN t = {} brackets 0 1 2 3 4 5 Shift I/PRP do/MD like/VBP eating/VBG fish/NN 4 0 1 2 3 5 5

  8. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish current I/PRP do/MD like/VBP eating/VBG fish/NN t = {} brackets 0 1 2 3 4 5 Shift Label-NP I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 4 0 1 2 3 5 5

  9. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish current I/PRP do/MD like/VBP eating/VBG fish/NN t = {} brackets 0 1 2 3 4 5 Shift Label-NP I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 4 0 1 2 3 5 Shift I/PRP do/MD like/VBP eating/VBG fish/NN 0 1 2 3 4 5 5

  10. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish current I/PRP do/MD like/VBP eating/VBG fish/NN t = {} brackets 0 1 2 3 4 5 Shift Label-NP I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 4 0 1 2 3 5 Shift I/PRP do/MD like/VBP eating/VBG fish/NN No-Label t = { 0 NP 1 } 0 1 2 3 4 5 5

  11. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish current I/PRP do/MD like/VBP eating/VBG fish/NN t = {} brackets 0 1 2 3 4 5 Shift Label-NP I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 4 0 1 2 3 5 Shift I/PRP do/MD like/VBP eating/VBG fish/NN No-Label t = { 0 NP 1 } 0 1 2 3 4 5 Shift I/PRP do/MD like/VBP eating/VBG fish/NN 0 1 2 3 4 5 5

  12. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish current I/PRP do/MD like/VBP eating/VBG fish/NN t = {} brackets 0 1 2 3 4 5 Shift Label-NP I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 4 0 1 2 3 5 Shift I/PRP do/MD like/VBP eating/VBG fish/NN No-Label t = { 0 NP 1 } 0 1 2 3 4 5 Shift No-Label I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 0 1 2 3 4 5 5

  13. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 0 1 2 3 4 5 6

  14. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 0 1 2 3 4 5 Combine I/PRP do/MD like/VBP eating/VBG fish/NN 4 0 1 3 5 6

  15. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 0 1 2 3 4 5 Combine No-Label I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 4 0 1 3 5 6

  16. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 0 1 2 3 4 5 Combine No-Label I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 4 0 1 3 5 Shift I/PRP do/MD like/VBP eating/VBG fish/NN 0 1 3 4 5 6

  17. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 0 1 2 3 4 5 Combine No-Label I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 4 0 1 3 5 Shift I/PRP do/MD like/VBP eating/VBG fish/NN No-Label t = { 0 NP 1 } 0 1 3 4 5 6

  18. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 0 1 2 3 4 5 Combine No-Label I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 4 0 1 3 5 Shift I/PRP do/MD like/VBP eating/VBG fish/NN No-Label t = { 0 NP 1 } 0 1 3 4 5 Shift I/PRP do/MD like/VBP eating/VBG fish/NN 0 1 3 4 5 6

  19. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 0 1 2 3 4 5 Combine No-Label I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 } 4 0 1 3 5 Shift I/PRP do/MD like/VBP eating/VBG fish/NN No-Label t = { 0 NP 1 } 0 1 3 4 5 Shift Label-NP I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 , 4 NP 5 } 0 1 3 4 5 6

  20. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 , 4 NP 5 } 0 1 3 4 5 7

  21. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 , 4 NP 5 } 0 1 3 4 5 Combine I/PRP do/MD like/VBP eating/VBG fish/NN 0 1 3 5 7

  22. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 , 4 NP 5 } 0 1 3 4 5 Combine Label-S-VP t = { 0 NP 1 , 4 NP 5 , I/PRP do/MD like/VBP eating/VBG fish/NN 3 S 5 , 3 VP 5 } 0 1 3 5 7

  23. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP Label- X Label VBG NP ` (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 , 4 NP 5 } 0 1 3 4 5 Combine Label-S-VP t = { 0 NP 1 , 4 NP 5 , I/PRP do/MD like/VBP eating/VBG fish/NN 3 S 5 , 3 VP 5 } 0 1 3 5 Combine I/PRP do/MD like/VBP eating/VBG fish/NN 0 1 5 7

  24. S Shift Structural NP VP (even step) PRP MD VBP S Combine I do like VP ` Label- X Label VBG NP (odd step) No-Label eating NN fish I/PRP do/MD like/VBP eating/VBG fish/NN t = { 0 NP 1 , 4 NP 5 } 0 1 3 4 5 Combine Label-S-VP t = { 0 NP 1 , 4 NP 5 , I/PRP do/MD like/VBP eating/VBG fish/NN 3 S 5 , 3 VP 5 } 0 1 3 5 Combine t = { 0 NP 1 , 4 NP 5 , I/PRP do/MD like/VBP eating/VBG fish/NN Label-VP 3 S 5 , 3 VP 5 , 0 1 5 1 VP 5 } 7

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Linear Time Constituency Parsing with RNNs and Dynamic Programming Juneki Hong 1 Liang Huang 1,2 1

Linear Time Constituency Parsing with RNNs and Dynamic Programming Juneki Hong 1 Liang Huang 1,2 1

Linear Time Constituency Parsing with RNNs and Dynamic Programming Juneki Hong 1 Liang Huang 1,2 1 Oregon State University 2 Baidu Research Silicon Valley AI Lab Span Parsing is SOTA in Constituency Parsing Cross+Huang 2016 introduced Span

691 views • 53 slides
A Minimal Span-Based Neural Constituency Parser Mitchell Stern, Jacob Andreas, Dan Klein CS 546

A Minimal Span-Based Neural Constituency Parser Mitchell Stern, Jacob Andreas, Dan Klein CS 546

A Minimal Span-Based Neural Constituency Parser Mitchell Stern, Jacob Andreas, Dan Klein CS 546 Paper Presentation Boyin Zhang Outline 1. Introduction 2. Background 3. Model 4. Algorithms 5. Training Details 6. Experiments 7.

660 views • 22 slides
Constituency Parsing CMSC 723 / LING 723 / INST 725 M ARINE C ARPUAT marine@cs.umd.edu T

Constituency Parsing CMSC 723 / LING 723 / INST 725 M ARINE C ARPUAT marine@cs.umd.edu T

Constituency Parsing CMSC 723 / LING 723 / INST 725 M ARINE C ARPUAT marine@cs.umd.edu T odays Agenda Grammar-based parsing with CFGs CKY algorithm Dealing with ambiguity Probabilistic CFGs Strategies for improvement

1.05k views • 68 slides
Domain Adaptation for Constituency Parsing Using Partial Annotations Vidur Joshi Matthew Peters

Domain Adaptation for Constituency Parsing Using Partial Annotations Vidur Joshi Matthew Peters

Domain Adaptation for Constituency Parsing Using Partial Annotations Vidur Joshi Matthew Peters Mark Hopkins Constituency Parsing is Useful Textual Entailment (Bowman et al., 2016) Semantic Parsing (Hopkins et al., 2017) Sentiment Analysis

1.01k views • 82 slides
Constituency Parsing Data structures and algorithms for Computational Linguistics III ar

Constituency Parsing Data structures and algorithms for Computational Linguistics III ar

Constituency Parsing Data structures and algorithms for Computational Linguistics III ar ltekin ccoltekin@sfs.uni-tuebingen.de University of Tbingen Seminar fr Sprachwissenschaft Winter Semester 20192020 Introduction CKY

661 views • 65 slides
An Empirical Comparison of Unsupervised Constituency Parsing Methods Jun Li, Yifan Cao, Jiong

An Empirical Comparison of Unsupervised Constituency Parsing Methods Jun Li, Yifan Cao, Jiong

An Empirical Comparison of Unsupervised Constituency Parsing Methods Jun Li, Yifan Cao, Jiong Cai, Yong Jiang, Kewei Tu {lijun2, caoyf, caijiong, tukw}@shanghaitech.edu.cn yongjiang.jy@alibaba-inc.com Background Goal : To learn a

339 views • 32 slides
Dependency Parsing Dr. Besnik Fetahu Parsing so far Use context free grammars to

Dependency Parsing Dr. Besnik Fetahu Parsing so far Use context free grammars to

Dependency Parsing Dr. Besnik Fetahu Parsing so far Use context free grammars to determine constituents in a clause or sentence Use CFGs to parse entire sentences into constituency based parse trees, e.g. syntactic parse trees In

461 views • 42 slides
Constituency Parsing Spring 2020 2020-03-24 Adapted from slides from Danqi Chen and Karthik

Constituency Parsing Spring 2020 2020-03-24 Adapted from slides from Danqi Chen and Karthik

SFU NatLangLab CMPT 825: Natural Language Processing Constituency Parsing Spring 2020 2020-03-24 Adapted from slides from Danqi Chen and Karthik Narasimhan (with some content from David Bamman, Chris Manning, Mike Collins, and Graham Neubig)

712 views • 52 slides
Policy Gradient as a Proxy for Dynamic Oracles in Constituency Parsing Daniel Fried and Dan Klein

Policy Gradient as a Proxy for Dynamic Oracles in Constituency Parsing Daniel Fried and Dan Klein

Policy Gradient as a Proxy for Dynamic Oracles in Constituency Parsing Daniel Fried and Dan Klein Parsing by Local Decisions S VP NP NP nap . a The cat took (S (NP The cat ) (VP = log ; ) =

590 views • 18 slides
Straight to the Tree: Constituency Parsing with Neural Syntactic Distance Yikang Shen*, Zhouhan

Straight to the Tree: Constituency Parsing with Neural Syntactic Distance Yikang Shen*, Zhouhan

Straight to the Tree: Constituency Parsing with Neural Syntactic Distance Yikang Shen*, Zhouhan Lin*, Athul Paul Jacob, Alessandro Sordoni, Aaron Courville, Yoshua Bengio University of Montreal, Microsoft Research, University of Waterloo

508 views • 35 slides
Dependency Parsing & Feature-based Parsing Ling571 Deep Processing Techniques for NLP

Dependency Parsing & Feature-based Parsing Ling571 Deep Processing Techniques for NLP

Dependency Parsing & Feature-based Parsing Ling571 Deep Processing Techniques for NLP February 2, 2015 Roadmap Dependency parsing Graph-based dependency parsing Maximum spanning tree CLE Algorithm Learning

899 views • 87 slides
Constituency Parsing Karl Stratos Rutgers University Karl Stratos CS 533: Natural Language

Constituency Parsing Karl Stratos Rutgers University Karl Stratos CS 533: Natural Language

CS 533: Natural Language Processing Constituency Parsing Karl Stratos Rutgers University Karl Stratos CS 533: Natural Language Processing 1/37 Project Logistics 1. Proposal (due 3/24) 2. Milestone (due 4/15) 3. Presentation (tentatively

869 views • 37 slides
Parse Trees Statistical NLP Spring 2011 Lecture 15: Parsing I The move followed a round of

Parse Trees Statistical NLP Spring 2011 Lecture 15: Parsing I The move followed a round of

Parse Trees Statistical NLP Spring 2011 Lecture 15: Parsing I The move followed a round of similar increases by other lenders, reflecting a continuing decline in that market Dan Klein UC Berkeley Phrase Structure Parsing Constituency

265 views • 5 slides
Optimal Parsing Strategies for Linear Context-Free Rewriting Systems Daniel Gildea Computer

Optimal Parsing Strategies for Linear Context-Free Rewriting Systems Daniel Gildea Computer

Optimal Parsing Strategies for Linear Context-Free Rewriting Systems Daniel Gildea Computer Science Department University of Rochester Overview Factorization lowers rank of LCFRS rules Binarization minimizes parsing complexity

353 views • 24 slides
Second Generation Model-based Testing Provably Strong Testing Methods for the Certification of

Second Generation Model-based Testing Provably Strong Testing Methods for the Certification of

CyPhyAssure Spring School Second Generation Model-based Testing Provably Strong Testing Methods for the Certification of Autonomous Systems Part II of III Provably Strong Testing Methods for Autonomous Systems Jan Peleska University of Bremen

634 views • 51 slides
Natural Language Processing with Deep Learning CS224N/Ling284 Christopher Manning Lecture 18:

Natural Language Processing with Deep Learning CS224N/Ling284 Christopher Manning Lecture 18:

Natural Language Processing with Deep Learning CS224N/Ling284 Christopher Manning Lecture 18: Tree Recursive Neural Networks, Constituency Parsing, and Sentiment Lecture Plan: Lecture 18: Tree Recursive Neural Networks, Constituency Parsing,

878 views • 73 slides
CS 225 Data Structures Se Septembe ber 23 23 St Stac acks, Que ueue ues and and

CS 225 Data Structures Se Septembe ber 23 23 St Stac acks, Que ueue ues and and

CS 225 Data Structures Se Septembe ber 23 23 St Stac acks, Que ueue ues and and Design gn G G Carl Evans CS 225 So Far CS 225 So F r List ADT Linked Memory Implementation (Linked List) O(1) insert/remove at

691 views • 13 slides
CS 225 Data Structures Sept. 27 Queues and It Iterators Queue.h #ifndef QUEUE_H 1 2

CS 225 Data Structures Sept. 27 Queues and It Iterators Queue.h #ifndef QUEUE_H 1 2

CS 225 Data Structures Sept. 27 Queues and It Iterators Queue.h #ifndef QUEUE_H 1 2 #define QUEUE_H 3 4 template <class QE> 5 class Queue { 6 public: 7 8 9 10 11 12 13 14 15 16 17 private: 18 19 20 }; 21 22

417 views • 9 slides
Topic 16 Queues Queues A sharp tool, like stacks "FISH queue : n. A line [acronym, by

Topic 16 Queues Queues A sharp tool, like stacks "FISH queue : n. A line [acronym, by

Topic 16 Queues Queues A sharp tool, like stacks "FISH queue : n. A line [acronym, by analogy with FIFO (First In, pointing out that processing of a particular sequence of events or requests has stopped dead. Also FISH mode and FISHnet;

426 views • 4 slides
Resilient Food Systems, Resilient Cities Presented by Kim Zeuli Resilience A

Resilient Food Systems, Resilient Cities Presented by Kim Zeuli Resilience A

Resilient Food Systems, Resilient Cities Presented by Kim Zeuli Resilience A resilient Growing food Processing system Distribution 60% of NYCs produce; 50% meat and fish 28% of site in

181 views • 15 slides
Object Oriented Programming and Design in Java Session 20 Instructor: Bert Huang Announcements

Object Oriented Programming and Design in Java Session 20 Instructor: Bert Huang Announcements

Object Oriented Programming and Design in Java Session 20 Instructor: Bert Huang Announcements Homework 4 due Monday, Apr. 19th ( next class) Review Homework tips Data Structures Lists, Stacks, Queues Sets, HashSet

599 views • 22 slides
Services and Scale Jeff Chase Duke University A simple, familiar example request GET

Services and Scale Jeff Chase Duke University A simple, familiar example request GET

D D u k e S y s t t e m s Services and Scale Jeff Chase Duke University A simple, familiar example request GET /images/fish.gif HTTP/1.1 reply client (initiator) server s = socket( ); sd = socket( ); bind(s, name);

904 views • 75 slides
B-Tree and Heaps Todays announcements HW3 out, due Nov 15, 23:59 MT2 Nov 7, 19:00-21:00

B-Tree and Heaps Todays announcements HW3 out, due Nov 15, 23:59 MT2 Nov 7, 19:00-21:00

B-Tree and Heaps Todays announcements HW3 out, due Nov 15, 23:59 MT2 Nov 7, 19:00-21:00 WOOD 2 Todays Plan B-Tree wrap Priority queues Warm up: If m = 128, then a B-Tree of height 4 must contain at least 1. 300 items 2.

538 views • 10 slides
SEA - RIOUS SCHOOL OF FISH! Proudly presents FISH! H! PHILOSO LOSOPHY PHY WORKSH

SEA - RIOUS SCHOOL OF FISH! Proudly presents FISH! H! PHILOSO LOSOPHY PHY WORKSH

PUNGGO GOL L GREEN EN PRIMARY Y SEA - RIOUS SCHOOL OF FISH! Proudly presents FISH! H! PHILOSO LOSOPHY PHY WORKSH SHOP FO FOR PARENTS nd January 2018 2 nd THE FISHy Hy! ! STO TORY Discovering the FISH!

618 views • 31 slides

More recommend