The Semantic Role Labeling problem can be formulated as a sentence tagging problem. A sentence can be represented as a sequence of words, as phrases (chunks), or as a parsing tree. The basic units of a sentence are words, phrases, and constituents in these representations, respectively. Pradhan et al. (2004) established that Constituent-by-Constituent (C-by-C) is better than Phrase-by-Phrase (P-by-P), which is better than Word-by-Word (W-by-W). This is probably because the boundaries of the constituents coincide with the arguments; therefore, C-by-C has the highest argument identification F-score among the three approaches.
We exploits full parsing information by representing it as features of argument classification models and as constraints in integer linear learning programs. In addition, to take advantage of SVM-based and Maxi-mum Entropy-based argument classification models, we incorporate their scoring matrices, and use the combined matrix in the above-mentioned integer linear pro-grams. The experimental results show that full parsing information not only in-creases the F-score of argument classification models by 0.7%, but also effectively removes all labeling inconsistencies, which increases the F-score by 0.64%. The ensemble of SVM and ME also boosts the F-score by 0.77%. Our system achieves an F-score of 76.53% in the development set and 76.38% in Test WSJ.
　

Demo site URL: