Recognizing Chinese Proper Nouns with Transformation-based Learning and Ontology
Peifeng Li, Qiaoming Zhu and Lei Wang
The 10th Congress of the Italian Association for Artificial Intelligence (AIIA 2007)
Roma, Italy, September 10-13, 2007
Abstract
An approach based on transformation-based error-driven learning is proposed to recognize Chinese proper nouns. Firstly, that approach redefines the tag set of Chinese words according to the usage of proper nouns and its context, and then it extracts characteristic information (CI) of proper nouns and merges them based on Ontology. Secondly, it tags the training corpus following new definitions of Multi-dimension Attribute Point (MAP), and then extracts rules to recognize proper nouns by using the transformation-based learning approach. Finally, proper nouns are recognized by utilizing the rule set and Ontology. The experimental result in the open test shows that the precision is 92.5% and the recall is 86.3%.