Introduction and Background Error Detection Re-generation Surface Realisation by Network Exploration Conclusion and Perspectives Pairing Model-Theoretic Syntax and Semantic Network for Writing Assistance Jean-Philippe Prost and Mathieu Lafourcade LIRMM, Universit´ e Montpellier 2 CSLP@Context, 2011 J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection Re-generation Surface Realisation by Network Exploration Conclusion and Perspectives The Problem Syntax/Semantics interface for Property Grammar (PG) through writing assistance Example *L’avocat le dossier de son client ( The lawyer his client’s file ) Case of (likely) missing word: L’avocat X le dossier de son client where X is of category V Expected surface realisation: plaide ( pleads ) J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection Re-generation Surface Realisation by Network Exploration Conclusion and Perspectives Outline 1 Introduction and Background 2 Error Detection 3 Re-generation 4 Surface Realisation by Network Exploration The Lexical Network Completion Message Propagation Algorithm 5 Conclusion and Perspectives J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection Re-generation Surface Realisation by Network Exploration Conclusion and Perspectives Sketch of the Process 1 Error detection with approximated parse(s) 2 (unrealised) Re-generation by tree transduction 3 Surface realisation with lexical network choice of functional and semantic roles completion message propagation in the network J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection Re-generation Surface Realisation by Network Exploration Conclusion and Perspectives Property Grammar (Blache, 2001) Model-Theoretic Semantics for PG (Duchier et al., 2009) S NP1 VP Models are labelled trees L’avocat V NP2 plaide NP3 PP le dossier de son client The grammar is a constraint system over tree nodes NP : D ≺ N VP : △ V NP : N ⇒ D , . . . J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection Re-generation Surface Realisation by Network Exploration Conclusion and Perspectives Model-Theoretic Semantics for PG Strong semantics (i.e. well-formedness) τ : σ | = G a syntax tree τ is a strong model of property grammar G , with realization σ , iff it is admissible and R τ ( ε ) = σ and I − G ,τ = ∅ J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection Re-generation Surface Realisation by Network Exploration Conclusion and Perspectives Model-Theoretic Semantics for PG S NP1 VP D N V Loose semantics *NP2 Les employ´ es rendent The employees N AP deliver rapport Adv A report tr` es complet very complete fitness F G ,τ = I + G ,τ / I 0 G ,τ loose models τ : σ | ≈ G τ ∈ argmax iff ( F G ,τ ′ ) τ ′ ∈A G ,σ J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection Re-generation Surface Realisation by Network Exploration Conclusion and Perspectives Getting Approximated Parses S NP1 VP D N V *NP2 Les employ´ es rendent The employees N AP deliver rapport Adv A report tr` es complet very complete Use of robust parsers’s combined output as set of models Most robust parsers are not capable of deciding about the well-formedness of the input sentence J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection Re-generation Surface Realisation by Network Exploration Conclusion and Perspectives Characterising Approximated Parses Characterisation of a model: set of pairs (pertinent instance of property, truth value) I 0 G ,τ = { r ∈ I τ [ [ G ] ] | P τ ( r ) } I + G ,τ = { r ∈ I 0 G ,τ | S τ ( r ) } I − G ,τ = { r ∈ I 0 G ,τ | ¬ S τ ( r ) } J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection Re-generation Surface Realisation by Network Exploration Conclusion and Perspectives Re-generation by Tree Transduction Basic tree operations, where τ is a tree, and c , c 1 , c 2 are node labels (i.e. categories): Node insertion , denoted by τ ↓ c Node deletion , denoted by τ ∤ c τ Node permutation , denoted by c 1 ↔ c 2 Transduction: Property Violated instances Tree operation Requirement I τ [ [ c 0 : c 1 ⇒ s 2 ] ] τ ↓ s 2 Obligation I τ [ [ c 0 : △ c 1 ] ] τ ↓ c 1 τ Linearity I τ [ [ c 0 : c 1 ≺ c 2 ] ] c 1 ↔ c 2 I τ [ Uniqueness [ c 0 : c 1 !] ] τ ∤ c 1 Exclusion I τ [ [ c 0 : c 1 �⇔ c 2 ] ] τ ∤ c 1 ∪ τ ∤ c 2 J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection Re-generation Surface Realisation by Network Exploration Conclusion and Perspectives Re-generated Model S NP1 VP D N V NP2 Les employ´ es rendent The employees deliver D N AP X rapport Adv A report tr` es complet very complete J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection The Lexical Network Re-generation Completion Message Surface Realisation by Network Exploration Propagation Algorithm Conclusion and Perspectives The Lexical Network rezoJDMFR Nodes and directed relations Weights and types Example cat isa cat loc − → animal − → sofa cat part cat can − → pur − → claw Many relation types including semantic roles agent, patient, instrument Other relations Typical location, manner, entailment isa, partof, substance, synonym, antonym, . . . J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection The Lexical Network Re-generation Completion Message Surface Realisation by Network Exploration Propagation Algorithm Conclusion and Perspectives A game for building the network JeuxDeMots http://jeuxdemots.org/ Users associate terms given a relation A popular consensus filtered by pairs of players In 4 years time over 1 , 200 , 000 relations among 230 , 000 terms evaluation through a guessing game (AKI, tip-of-the-tongue ) term found in more than 75% of cases while the typical human score is around 46% J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection The Lexical Network Re-generation Completion Message Surface Realisation by Network Exploration Propagation Algorithm Conclusion and Perspectives J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection The Lexical Network Re-generation Completion Message Surface Realisation by Network Exploration Propagation Algorithm Conclusion and Perspectives J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection The Lexical Network Re-generation Completion Message Surface Realisation by Network Exploration Propagation Algorithm Conclusion and Perspectives Completion Message � a , :R , b � :R denotes an oriented semantic relation, and a and b its oriented elements. J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection The Lexical Network Re-generation Completion Message Surface Realisation by Network Exploration Propagation Algorithm Conclusion and Perspectives Gathering Functional and Semantic Roles VP (Verb Phrase) � � V S head Obligation : △ V � V .subcat � subcat Uniqueness : V [main past part] ! : NP ! NP1 VP : PP ! Linearity : V ≺ NP : V ≺ Adv L’avocat : V ≺ PP V NP2 Requirement : V [past part] ⇒ V [aux] Exclusion : Pro [acc] �⇔ NP : Pro [dat] �⇔ Pro [acc] X NP3 PP PRED role � � role OBJ | P-OBJ | A-OBJ � � Dependency : V [ role PAT ] � NP arg subcat le dossier de son client cat NP PRED role � � role OBJ | P-OBJ | A-OBJ � � : V � PP [ role PAT ] subcat arg PP cat NP2 is in a Patient relationship with V NP2 ’s head is inherited from NP3 ’s: dossier ( file ) Message ♯ 1 = � X , :PAT , dossier � J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection The Lexical Network Re-generation Completion Message Surface Realisation by Network Exploration Propagation Algorithm Conclusion and Perspectives Gathering Functional and Semantic Roles S (Utterance) Obligation : △ VP Uniqueness : NP ! : VP ! Linearity : NP ≺ VP role PRED Dependency : VP 0 � � � pos � � NP role AGT subcat role SUBJ arg NP cat NP1 is in an Agent relationship with VP by inheritance, NP1 ’s head is avocat ( lawyer ) Message ♯ 2= � avocat , :AGT , X � J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Introduction and Background Error Detection The Lexical Network Re-generation Completion Message Surface Realisation by Network Exploration Propagation Algorithm Conclusion and Perspectives Completion Messages {� X , :PAT , dossier � , � avocat , :AGT , X �} J-P. Prost and M. Lafourcade Pairing MTS and Semantic Network
Recommend
More recommend
