Exploiting clause boundary information as features for Chinese functional chunk parsing

This paper aims at using clause boundary information to improve the performance of the Chinese functional chunk parsing. Firstly, we propose a rule-based method using both punctuation and predicate information to automatically exploit the boundary of the clause, then, we apply these information to the Chinese functional chunk parsing which is based on CRF model as features. The experiment results show that the use of the clause boundary information can contribute to significant performance improvements which outperform the other methods. Specifically, these features can better deal with comparably long and complex chunk.