Paper: The Integration of Dependency Relation Classification and Semantic Role Labeling Using Bilayer Maximum Entropy Markov Models

ACL ID W08-2135
Title The Integration of Dependency Relation Classification and Semantic Role Labeling Using Bilayer Maximum Entropy Markov Models
Venue International Conference on Computational Natural Language Learning
Session Main Conference
Year 2008
Authors

This paper describesa system to solve thejointlearningof syntacticandseman- tic dependencies. An directedgraphical modelis put forward to integratedepen- dency relationclassificationandsemantic role labeling. We presenta bilayerdi- rected graph to expressprobabilisticre- lationshipsbetweensyntacticand seman- tic relations. MaximumEntropy Markov Modelsareimplementedto estimatecon- ditionalprobabilitydistributionandto do inference. The submittedmodel yields 76.28% macro-average F1 performance, forthejointtask,85.75%syntacticdepen- denciesLASand66.61%semanticdepen- denciesF1.