Traditional algorithms of named entity recognition rely on a largenumber of artificial features and external knowledge To solve this problem

wepropose a novel neural network structure which combines the bidirectionalLSTM(Long ShortTerm Memory)

CNN(Convolution Neural Network) andCRF(Conditional Random Field) This algorithm can also automatically get bothcharacterlevel and wordlevel representations It is a truly endtoend system

requiring no feature engineering and data processing Thus

it can be easilyapplied to various fields of named entity recognition task Finally

theexperimental results show that the F1 measure of the algorithm in the medicaland news field is 9097% and 9219%

respectively