CHINESE ADDRESS PARSING AND ANNOTATION METHOD

摘要

The present invention relates to a Chinese address parsing and annotation method. The present method comprises: step 11, manually parsing and annotating selected address data and using same as training data; step 12, using specified single Arabic numerical characters or English letter characters to replace single Arabic numerical characters or English letter characters and multiple consecutive Arabic numerical characters or English letter characters that appear; step 13, converting training data to a format required by a CRF++ tool; step 14, defining a characteristic template; step 15, establishing a parsing and annotation model; step 16, using the specified single Arabic numerical characters or English letter characters to replace single Arabic numerical characters or English letter characters and multiple consecutive Arabic numerical characters or English letter characters that appear in the address; step 17, using the CRF++ tool for parsing and annotation; step 18, restoring the replaced Arabic numerical characters or English letter characters. The Chinese address parsing and annotation method of the present invention is highly accurate.