基于模拟退火算法和隐马尔可夫模型的文本信息抽取

Text Information Extraction Based on Simulated Annealing Algorithm and Hidden Markov Model

本文提出了一种基于改进的模拟退火算法和二阶隐马尔可夫模型相结合的信息抽取方法.其中,改进的模拟退火算法在适当的时机增加了重升温过程,提高了局部寻优的搜索效率;二阶隐马尔可夫模型充分考虑了概率与历史状态的关联性,增加了信息抽取的可靠性.实验结果表明,新算法在精确度、召回率和时间性能指标上比基于模拟退火算法和一阶隐马尔可夫模型的信息抽取方法有所提高.

An improved combined method embedded with modified simulated annealing algorithm and second-order hidden Markov model was presented.The modified simulated annealing algorithm improves the temperature at the proper time,which increased the efficiency of local search optimization.The second-order hidden Markov model considers the relevance of probability and historical status which improved the reliability of information extraction.Experimental results show that the new method has a better improvement in precision,recall and time performance than the method combined with simulated annealing algorithm and first-order hidden Markov model.