论文部分内容阅读
This paper proposed a new method of semi-automatic extraction for semantic structures from unlabelled corpora in specific domains. The approach is statistical in nature. The extracted structures can be used for shallow parsing and semantic labeling. By it