High precision English base noun phrase identification based on "waterfall" model

Based on classical model that used by software exploitation in the subject of Software Engineering - waterfall model, a high precision model for English noun phrase identification is presented. In this model, three important features (interior structure, context information and boundary character) in base noun phrase identification are orderly used, and the relative method (the rule method, the transfer-based and error-driven method, and the boundary statistic method) are used from top to bottom. Thus the precision of base noun phrase identification improved steadily. The "waterfall" model combined the rule method and the statistical method so as to make them complement each other. The precision of base noun phrase identification achieves 98.10% and the F score is 95.25%. Compared to other method, our method achieves the highest precision and F score.