Application Of Literature Classification Based On Convolution Neural Network In Rice Resistance Gene Database

Rice plays an important role in China's food crops.Genetic studies of rice under abiotic stress have important implications for China's food security.With the continuous development of science and technology,the number of technical means for acquiring rice information has increased.The data on the biological information of rice has grown exponentially.How to quickly and accurately classify literature containing stress resistant genes from a large amount of rice biological information and collect it? Outbreak resistance genetic data of rice has gradually become a difficult problem to be solved.The research of this dissertation is based on the topic of "Resistance Gene Bank of the Main Crops Entrusted by Crop Resilience Breeding and Disaster Reduction Engineering Laboratory",using the convolutional neural network model to classify the abstracts of rice literature and screen out the rice literature containing rice stress resistance data.,And put forward the data of stress-resistant genes in the literature to build a rice stress-resistant gene database so as to better serve scientific researchers.The content of this dissertation is as follows:(1)This dissertation compares and analyzes the classification model to determine the classification model suitable for literature classification.The experiment selected 2000 papers of abiotic literature under non-biological stress conditions were divided into test data sets and training data sets.The models were constructed using K-nearest neighbors,Na?ve Bayes method and convolutional neural network methods,and the test data sets were classified.The experimental results were compared based on three evaluation criteria: accuracy rate,recall rate,and F-measurement value.It was found that the convolutional neural network model was the best among the three text classification models.(2)The literature was classified using a convolutional neural network model.In this disssertation,the reptile technique is used to collect 2830 rice abstracts under non-biological stress conditions as text data sets.Firstly,the text data sets are coded and preprocessed.The word data is obtained by word2 vec training text data sets,and gene types are used as text classification.Input features;then construct a vector matrix and enter the CNN neural network model,and output the classification results through the convolution layer,pooled layer,and full-connection layer.(3)According to the convolutional neural network literature classification,related literatures containing stress resistance genes were constructed and a database of rice stress resistance genes was constructed.First,the classified literature is studied and analyzed to extract the rice stress-resistance gene information data contained in the literature;then,JSON data exchange format is used for data transmission,MySQL database is used to store information,and AJAX technology is used for front-end and back-end data.Interaction,construction and display of rice stress gene database.In this dssertation,we use the convolutional neural network model to classify the rice literature,improve the retrieval efficiency and accuracy of the literature,and facilitate the collection and study of rice resistance gene literature;we construct the rice stress resistance gene database,which is beneficial to rice resistance gene data.The management,renewal,and maintenance will also better serve scientific researchers in information acquisition.