Automatic Semantic Role Labeling for Chinese
Ke Wang · 2010
In this paper I propose a new method for labeling Chinese with semantic roles neither using syntactic parsing nor Part Of Speech tagging technologies. The whole task was divided into two subtasks, clustering and labeling. Clustering is aimed at partially replacing syntactic parsing, during which similar sentences are clustered together. In the labeling step, artificial neural networks is planted as many as the number of clusters, each of which takes charge of summing up features of chunks of a sentence and then labeling them with semantic roles. The experiment result shows this method is useful; and 83.8% correctness on average is achieved.