45 Research on Internet Chinese News Geography Name Text Mining

It is a complicated text mining task for classify Internet news geography name. Proper noun recognition is one of difficult problem in NLP, in order to achieve proper result, we present a comprehend solution based on geography name knowledge base matching, heuristics decision making and geography name hierarchy induction. To overcome geography knowledge base adoption to Internet news geography name feature's appearance, we add geography knowledge base's self-learning function. According to this solution, we developed a prototype system, and integrated it to SmartICIN, realized trade and geography name feature automatic classification. We tested it using ChinaInfoBank news corpus, experiment result shows it can be accepted more than 85%.