In this paper, we propose a new sequential labeling scheme, double sequential labeling, that we apply it on Chinese parsing. The parser is built with conditional random field (CRF) sequential labeling models. One focuses on the beginning of a phrase and the phrase type, while the other focuses on the end of a phrase. Our system, CYUT, attended 2012 the second CIPS-SGHAN conference Bake-off Task4, traditional Chinese parsing task, and got promising result on the sentence parsing task.
展开▼