Research on Information Extraction Based on Second-Order HMM
Xiong Ling · Journal of Intelligence · 2011
Hidden Markov model is an effective approach for information extraction.The second-order hidden Markov model could get more contextual information than the first-order hidden Markov model does.Since the transition and emission probabilities depend on the historical states in the second-order hidden Markov model which has better performance of recognition for states,this paper proposes an improved text information extraction model based on the second-order hidden Markov model.The traditional viterbi algorithm is improved and a smooth algorithm is introduced to solve the zero probability problem.The improved model is proved to be more effective and precise than the second-order HMM.