使用层叠隐马模型解决命名实体识别问题，含有训练语料及测试预料。-Implicit use of cascading Ma Named Entity Recognition Model to solve the problem, containing training materials and tests are expected words.

这是HMM算法里的训练功能的程序，是EM算法中的一种，即前向后向算法。-This is where the training function HMM algorithm procedure is an EM algorithm, the algorithm back before that.

这个是我自己整理的关于语音识别HMM模型训练部分的决策树相关学习总结-Decision tree learning summary of HMM model training for speech recognition