We describe a system for the CoNLL-2005 shared task of Semantic Role Labeling.The system implements a two-layerarchitecture to first identify the argumentsand then to label them for each predicate.The components are implemented asSVM classifiers using libSVM. Featureswere adapted and tuned for the system,including a reduced set for the identifierclassifier. Experiments were conducted tofind kernel parameters for the Radial BasisFunction (RBF) kernel. An algorithmwas defined to combine the results of theargument labeling classifier according tothe constraints of the argument labelingproblem.
展开▼