earticle

영어

Part of speech tagging is a basic step which is used for the next steps of natural language
understanding: syntactic analysis and semantic analysis. This is also can be applied for
Information processing such as information retrieval, question and answering. Etc. Both a
rule based approach and a statistical approach have been adopted for POS tagging. Since the
both approach has drawbacks, a hybrid approach were employed to overcome the shortcoming of the both approaches. In this paper, we propose a hybrid approach with which the rule and statistics are naturally combined into the viterbi algorithm, the probabity of rule including context information is combined into HMM thus the performance is expected better than the previous hybrid approaches.