In order to improve the accuracy of structured classification of medical records,the conditional random field(CRF)semi supervised dictionary segmentation algorithm and the implicit Dirichlet distribution(LDA)medical records text classifica-tion algorithm ard used to build a post structured system of medical records based on CRF mechanism and LDA.The results show that:when the number of topics is 40,the minimum perplexity of LDA topic modeling is-6.97,and compared with the initial perplexity,the perplexity of LDA topic modeling decreases by 9.76％.When the number of topics is 3,the minimum consistency value obtained is 0.361;when the number of topics is 40,the maximum consistency value obtained is 0.442,and compared with the minimum value,the increase in LDA topic modeling consistency value is 22.44％.In summary,it can be seen that the CRF mechanism combined with LDA's structured medical record system has good application effects.
关键词
条件随机场/半监督词典/隐式狄利克雷分布/病历文书/文本分类
Key words
conditional random field/semi supervised dictionary/implicit Dirichlet distribution/medical record document/text classification
维普
万方数据