HMM-based Korean named entity recognition for information extraction

This paper presents the HMM (Hidden Markov Model) based named entity recognition method for information extraction. In Korean language, named entities have the distinct characteristics unlike other languages. Many named entities can be decomposed into more than one word. Moreover, there are contextual relationship between named entities and their surrounding words. There are many internal and external evidences in named entities. To overcome data sparseness problem, we used multi-level back-off methods. The experimental result shows the F-measure of 87.6% in the economic article domain.