a latent variable model of synchronous parsing for
play

A Latent Variable Model of Synchronous Parsing for Syntactic and - PowerPoint PPT Presentation

A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation A Latent Variable Model of Synchronous Parsing for Syntactic and Semantic Dependencies James Henderson 1 Paola Merlo 2 Gabriele Musillo 1 2


  1. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation A Latent Variable Model of Synchronous Parsing for Syntactic and Semantic Dependencies James Henderson 1 Paola Merlo 2 Gabriele Musillo 1 2 Ivan Titov 3 1 Dept Computer Science, Univ Geneva 2 Dept Linguistics, Univ Geneva 3 Dept Computer Science, Univ Illinois at U-C CoNLL 2008 university-logo

  2. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Outline A Latent Variable Model of Synchronous Parsing 1 Probability Model 2 Machine Learning Method 3 Evaluation 4 university-logo

  3. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Motivation for synchronous parsing Syntax and semantics are separate structures , with different generalisations Obj Sub John broke the vase. A0 A1 Sub The vase broke. A1 Syntax and semantics are highly correlated , and therefore should be learned jointly Synchronous parsing provides a single joint model of two separate structures university-logo

  4. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Motivation for latent variables The correlations between syntax and semantics are partly lexical , and independence assumptions are hard to specify a priori The dataset is new, and there was little time for feature engineering Latent variables provide a powerful mechanism for discovering correlations both within and between the structures university-logo

  5. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Outline A Latent Variable Model of Synchronous Parsing 1 Probability Model 2 Machine Learning Method 3 Evaluation 4 university-logo

  6. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Outline A Latent Variable Model of Synchronous Parsing 1 Probability Model 2 Machine Learning Method 3 Evaluation 4 university-logo

  7. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation The Probability Model A generative, history-based model of the joint probability of syntactic and semantic synchronous derivations synchronised at each word . university-logo

  8. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Syntactic and semantic dependencies example ROOT Hope seems doomed to failure P ( T d , T s ) university-logo

  9. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Syntactic and semantic derivations Define two separate derivations , one for the syntactic structure and one for the semantic structure. P ( T d , T s ) = P ( D 1 d , ..., D m d d , D 1 s , ..., D m s s ) Actions of an incremental shift-reduce style parser similar to MALT [Nivre et al., 2006] Semantic derivations are less constrained, because their structures are less constrained Assumes each dependency structure is individually planar (“projective”) university-logo

  10. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Synchronisation granularity Use an intermediate synchronisation granularity, between full predications and individual actions. b t e t d , shift t , D b t s , ..., D e t C t = D d , ..., D d d s s , shift t s d , ..., D m d P ( D 1 d , D 1 s , ..., D m s s ) = P ( C 1 , . . . , C n ) Synchronisation at each word prediction Results in one shared input queue Allows two separate stacks university-logo

  11. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Synchronous parsing example ROOT Hope P ( C 1 ) university-logo

  12. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Synchronous parsing example ROOT Hope seems P ( C 1 ) P ( C 2 | C 1 ) university-logo

  13. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Synchronous parsing example ROOT Hope seems doomed P ( C 1 ) P ( C 2 | C 1 ) P ( C 3 | C 1 , C 2 ) university-logo

  14. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Synchronous parsing example ROOT Hope seems doomed to P ( C 1 ) P ( C 2 | C 1 ) P ( C 3 | C 1 , C 2 ) P ( C 4 | C 1 , C 2 , C 3 ) university-logo

  15. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Synchronous parsing example ROOT Hope seems doomed to failure P ( C 1 ) P ( C 2 | C 1 ) P ( C 3 | C 1 , C 2 ) P ( C 4 | C 1 , C 2 , C 3 ) P ( C 5 | C 1 , C 2 , C 3 , C 4 ) university-logo

  16. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope university-logo

  17. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems university-logo

  18. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems university-logo

  19. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems university-logo

  20. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems university-logo

  21. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems doomed university-logo

  22. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems doomed university-logo

  23. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems doomed university-logo

  24. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems doomed university-logo

  25. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems doomed to university-logo

  26. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems doomed to university-logo

  27. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems doomed to university-logo

  28. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems doomed to failure university-logo

  29. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems doomed to failure university-logo

  30. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Derivation example ROOT Hope seems doomed to failure university-logo

  31. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Projectivisation Allows crossing links between syntax and semantics Use the HEAD method [Nivre et al., 2006] to projectivise syntax Use syntactic dependencies to projectivise semantic dependencies university-logo

  32. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Projectivising semantic dependencies C w1 w2 w3 w4 w5 A B C w1 w2 w3 w4 w5 B A/C university-logo

  33. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Outline A Latent Variable Model of Synchronous Parsing 1 Probability Model 2 Machine Learning Method 3 Evaluation 4 university-logo

  34. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation The Machine Learning Method Synchronous derivations are modeled with an Incremental Sigmoid Belief Network ( ISBN ). ISBNs are Dynamic Bayesian Networks for modeling structures , with vectors of latent variables annotating derivation states that represent features of the derivation history . Use the neural network approximation of ISBNs [Titov and Henderson, ACL 2007] (“Simple Synchrony Netowrks”) university-logo

  35. A Latent Variable Model of Synchronous Parsing Probability Model Machine Learning Method Evaluation Statistical dependencies in the ISBN Connections between latent states reflect locality in the syntactic or semantic structure , thereby specifying the domain of locality for conditioning decisions Explicit conditioning features of the history are also specified t−c t t−1 S S S t−c t−1 t D D D university-logo

Recommend


More recommend