Accurate Learning for Chinese Function Tags from Lexical Evidence

Data-driven function label assignment hasbeen studied for English using Penn Treebank data. In thispaper, we address the question of whether such methodscan be applied to other languages and Treebank resources.In addition to simple extension of previous method, wealso proposed an effective way for assigning function labels to languages that lack of suffcient parse resources.We investigated three machine-learning algorithms to testour method which achieved the best F-score of 93.76 ongold-standard PoS (Part-of-speech) tagged Chinese text a statistically significant improvement over existing Chinese function labeling systems.