Research On Data Analysis Technology Of Chinese Paper Based On Neural Network

In recent years,the Internet has been fully developed,network communication and transmission technology has also developed rapidly,people's ability to create and transmit information has been greatly enhanced,and the network is full of more and more massive virtual information and resources.The traditional text analysis technology is more and more weak in the face of complex and massive text.In recent years,the development of deep learning technology has greatly improved people's ability of data processing and utilization.This paper studies and explores the extraction,classification and similarity calculation of key sentences in the text,and proposes an effective scheme to analyze and process the Chinese text by improving and combining the text Abstract extraction technology and deep learning technology.In the aspect of text key sentence extraction,aiming at the needs of the follow-up work and the defects of TF-IDF algorithm,we improved it appropriately,added DAC coefficient to measure the distribution of words among classes,so that the final weight can better measure the importance of words,and then get a higher quality text summary.With the help of word2 vec technology,a semantic based similarity calculation scheme is proposed.Training skip gram network to get the vector mapping table of words,quantifying the words,then combining the weight of words to build the text vector,then using cosine distance formula to calculate the similarity between texts.The input data form of the network is constructed by using the key sentences and word vector mapping table obtained by the text Abstract extraction technology in the previous paper.The task of growing text classification is completed by combining the convolution neural network,and the convolution neural network used in this paper is analyzed and explained in detail.In order to verify the feasibility and effectiveness of the method,experiments areset up to test the performance of the proposed algorithm.The results show that the text analysis technology with the help of deep learning technology has certain advantages compared with the traditional technology,and the improvement made in the extraction of text summary also has certain gains for the final results.At last,this paper constructs a text analysis system which can deal with Chinese text.In this paper,TF-IDF algorithm is improved to obtain higher quality text summarization.Combining with word2 vec technology,we propose a text similarity calculation scheme based on semantics,and a high-efficiency long text classification scheme based on convolution neural network technology.