Intro Source Target Mapping Evaluation Conclusion Mapping between Dependency Structures and Compositional Semantic Representations LREC 2010 Max Jakob, Mark´ eta Lopatkov´ a, Valia Kordoni ´ UFAL at Charles University in Prague, Czech Republic German Research Centre for Artificial Intelligence (DFKI GmbH) Dept. of Computational Linguistics, Saarland University, Germany 1/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion Motivation Research presented in the paper map PDT annotation �→ RMRS structures manually annotated corpora are valuable barrier: difference in formal descriptions → usage of a resource remains limited ⇒ precisely relate formalisms to overcome these limitations Benefits: flexibility, availability, RMRSs for Czech 2/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion Prague Dependency Treebank 2.0 Czech newspaper and magazine articles theoretical backgr.: Functional Generative Description theory (Sgall et al. , 1986) 3 annotation layers “Byl by ˇ sel do lesa.” “He would have gone into the woods.” 3/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion Input of the mapping (1) Tectogrammatical trees: t-ln94208-126-p2s4 root (Hajiˇ c et al. , 2006) highest level of honit enunc . abstraction PRED v sub-layers: structure and pes asi kočka dependencies ACT MOD PAT n.denot atom n.denot morphological “Pes asi hon´ ı koˇ cku.” categories “The dog probably coreferences chases a cat.” topic-focus articulation 4/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion Input of the mapping (2) PDT Valency Dictionary: (Hajiˇ c et al. , 2003) separate data source from PDT comprises obligatory and optional valency modifications does not contain free modifications minout (to pass) ACT, PAT (the bullet passed/missed the victim) ACT (the holidays have passed) 5/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion Output of the mapping (Robust) Minimal Recursion Semantics: (Copestake et al. , 2005; Copestake, 2007) flat, underspecified representation no semantic theory < [ l 0 , e 2] , { l 1: every q( x 1 , h 1 , h 2), l 2: white adj( x 1), l 2: cat n( x 1), l 3: probably adv( e 1 , h 3), l 4: eat v( e 2 [ tense : past ] , x 1 , x 2), l 5: a q( x 2 , h 4 , h 5), l 6: mouse n 1( x 2) } , { h 1 = q l 2, h 3 = q l 4, h 4 = q l 6 } > “Every white cat probably ate a mouse.” 6/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion Mapping PDT �→ RMRS adapt theoretical background of PDT rule-based approach t-ln94208-126-p2s4 root < [ l 0 , a 1 , e 1], { l 1: a 1: honit v 1( e 1), l 2: a 2: pes n.denot( x 1), honit enunc . l 4: a 4: koˇ cka n.denot( x 2), PRED l 3: a 3: asi atom( e 2), v �→ l 1: a 5:MOD( e 1) } , { a 1:ACT( x 1), pes asi kočka a 1:PAT( x 2), ACT MOD PAT a 3:ARG1( h 1), n.denot atom n.denot a 5:ARG1( e 2) } , { h 1 = q l 1 } > 7/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion node-RMRS represents a subtree as RMRS honit PRED < [ l 3 , a 1 , e 1], { l 1: a 1: honit v 1( e 1), l 2: a 2: pes n.denot( x 1), l 3: a 3: koˇ cka n.denot( x 2), l 4: a 4: asi atom( e 2), l 1: a 5:MOD( e 1) } , { a 1:ACT( x 1), a 1:PAT( x 2), a 4:ARG1( h 1), a 5:ARG1( e 2) } , { h 1 = q l 1 } > ✟✟✟✟✟✟✟✟✟✟✟✟✟✟ ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍ pes asi koˇ cka ACT MOD PAT < [ l 2 , a 2 , x 1], < [ l 4 , a 4 , e 2], < [ l 3 , a 3 , x 2], { l 2: a 2: pes n.denot( x 1) } , { l 4: a 4: asi atom( e 2) } , { l 3: a 3: koˇ cka n.denot( x 2) } , { } , { a 4:ARG1( h 1) } , { } , { } > { } > { } > 8/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion node-RMRS Initialization < [ l 3 , a 3 , x 2], { l 3: a 3: koˇ cka n.denot( x 2 [ number : sg , gender : fem ] ), l 4: a 4:udef q( x 2) } , { a 4:RSTR( h 1), a 4:BODY( h 1) } , { h 1 = q l 3 } > relation name: lemma POS index variable features: morphological categor. set hook values introduce a quantifier for nominal objects 9/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion node-RMRS Composition valency modification add argument to governing lexical EP free modification add EP that relates lexical EPs coordination add a coordination EP add constraints, update hook build union of all involved sets 10/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion Algorithm Sketch Input: tectogrammatical tree (, valency dictionary) Output: RMRS structure get node-RMRS(node) 1: initialize node-RMRS 2: for all relevant dependent nodes dep node-RMRS ← get node-RMRS (dep.) 3: treat dep node-RMRS 4: as a member of a coordination or as a valency modification or as a free modification merge dep node-RMRS with node-RMRS 5: 6: return node-RMRS 11/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion Evaluation no manual checking ← corpus size & lack of sufficient Czech skills Structurally valid MRSs 1 must be a net (Flickinger et al. , 2005). 2 must have at least one configuration . Precision 40120/44725 89.70 % Recall 40120/49431 81.16 % (skipped 4706 trees: 9.52 %) 12/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion Conclusion mapping of PDT dependency trees onto flat RMRS structures is feasible mapped : structure and dependencies, morphological categories, some grammatical coreferences future work : word order, quantifier representation, rest of grammatical coreference, textual coreference, topic-focus articulation Benefits: treebank data available in (R)MRS towards formalism independence compositional semantics structures for Czech 13/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Intro Source Target Mapping Evaluation Conclusion Paper Reference Max Jakob, Mark´ eta Lopatkov´ a, Valia Kordoni. Mapping between Dependency Structures and Compositional Semantic Representations. In Proceedings of the Seventh International Conference on Language Resources and Evaluation (LREC 2010) , Valletta, Malta. 14/14 Jakob, Lopatkov´ a, Kordoni Mapping between Dependencies & Compositional Semantics
Recommend
More recommend
