Chinese Medicine Named Entity Recognition

The Chinese medicine concerns named entity recognition.The medicine named entity of the inquiry information between doctor and patient in the internet is identified by using the linguistic rules template and the hidden Markov model of statistical learning.The research in the absence of relevant knowledge database support and experimental data from the "xun yi wen yao website" in the different inquiry information.We improve the probability smoothing methods to deal with data sparsity.This study is expected to reduce the dependence on the dictionary and difficulty of handling unknown words and data sparsity.Promoting the application of named entity recognition has practical significance to the development of natural language processing technology.In the process of experiment,we preprocess the text data and complete the annotation.The hidden Markov model is trained by using the tagged corpus.The observation set and the state set are determined,and the observation probability matrix,the state transition probability matrix and the initial state probability matrix are calculated through supervised training algorithm respectively.Output of the optimal state sequence using viterbi algorithm and the sequence by pattern matching.Split the special state into prefix and suffix word,then extract the corresponding Chinese medicine name.The precision rate,recall rate and F1 value of the experimental results were calculated.Using different probabilistic smoothing methods to optimize the model parameters.Correct the word boundary error through using sentential form and trigger word rules,which can further improve the recognition effect.The experimental results on real data sets show that the precision rate is about 80%,which indicates that the whole process is effective.