Named Entity Recognition In The Food Of Online Healthy Texts Based On Deep Learning

Deep learning is now a hot topic on machine learning in recent years.It utilizes the hierarchical thinking of the human brain,abstracts and learns training data using multi-layer network.As time goes on,a series of improvements and studies have been made to enable training it.Nowadays,neural networks like convolutional neural network,recurrent neural network have been proved to be effective in many areas such as text generation,speech synthesis,face recognition,behavior recognition and other tasks,at the same time,often achieve state of art in these tasks.With the huge improvement of living standards,Chinese people has begun to raise their concerns about their health conditions and related diseases.At the same time,the Internet of our country grows rapidly,the boom of online health website provides more health information and literature for average people.However,these online health documents exist in free text format,and entity recognition is an effective method to help with further data mining.This article aims at food material named entity recognition which is the most closely related to people's normal life.The main contribution of this paper includes several following points:1.The relevant difficulties of Chinese entity recognition are pointed out by analyzing relevant works.2.Analyzing and comparing the similarities and differences between the traditional method and the deep learning way,and points out the shortcomings of the traditional method and the improvement that the deep learning method can bring.3.With lack of related corpus,constructing entity corpus with relevant html page by means of web crawler4.Testing the performance of different recurrent neural network.5.Based on the previous work of convolutional neural network and recurrent neural network,achieved better results with new relate deep neural network.