基于主动学习隐马尔可夫模型的文本信息抽取

Text Information Extraction Based on Active Hidden Markov Model

对只能获得部分标记的训练文本,将主动学习方法应用到文本信息抽取中,提出了一种基于主动学习隐马尔可夫模型的文本信息抽取方法.在该方法中,通过主动学习,仅将对隐马尔可夫模型的训练最有价值的训练文本挑选出来进行标记.实验表明,通过选择模型信任值的最佳门槛值,该方法在保证文本信息抽取性能的前提下,大大减少了用户标记训练文本的工作量.

An active learning was used in text information extraction for training text, which was partly labeled. And an approach of text information extraction based on active hidden Markov model was proposed. In this approach, only the most valuable training text for training the model was selected out to label through active learning. Experimental results show that, by selecting the optimal threshold for active model parameters, the approach can reduce the user workload for labeling without affecting the performance of text information extraction.