Construction Of Knowledge Map Of Citrus Diseases And Insect Pests

With the deepening and expansion of the application of information technology in all fields of society,traditional agriculture has entered the era of information technology transformation and upgrading.Hunan is a large province of citrus planting in China.Citrus pests and diseases will seriously affect the yield and quality of citrus,bring great economic losses and reduce farmers’ income.At present,the relevant information of citrus diseases and insect pests are often recorded in professional books or field articles,so it is difficult for citrus farmers to find relevant information in a short time.In order to solve the above problems,this paper constructed the knowledge base of citrus diseases and insect pests by using knowledge map related technology,and improved the knowledge map of citrus diseases and insect pests under the guidance of relevant experts and literatures by studying the deep learning model and ontology editor Protege,and studied and built the query system of citrus diseases and insect pests based on knowledge map.At present,knowledge service covering the whole citrus industry chain in China is still in its infancy,and there is no mature and practical citrus pest service system.The study of this paper is of great significance to accelerate the construction of intelligent citrus industry service system.The main research contents of this paper include:(1)The initial pattern layer was created according to the entity structure levels of citrus diseases and insect pests recorded in agricultural Science Thesaurus,and part of citrus diseases and insect pests data were imported into Neo4 j graph database to realize an initial agricultural knowledge map of citrus diseases and insect pests.Based on the pest and disease entities in the Thesaurus of Agricultural Sciences,a data collection tool was designed to obtain entity-related data sets from multiple sources such as Baidu Encyclopedia and agricultural books.The obtained structured and unstructured data were stored in CSV files,and then data cleaning and noise reduction were carried out.(2)Bi LSTM-CRF deep learning algorithm is used to extend the data layer of knowledge graph.Deep learning algorithm was trained by using BIO annotation strategy to automatically recognize text entities and extract triad data of citrus pests and diseases based on label information.During the experiment,batch＿size and num＿epochs will be set to 64 and 100 respectively.Through the training and comparison of multiple entity recognition models Bi LSTM,Bi LSTM-CRF,LSTM and LSTM-CRF,Bi LSTM-CRF has the best effect,and its accuracy rate,recall rate and F1 value reach 91.07%,80.95% and 85.71%,respectively.(3)Based on the extracted triplet data,the ontology information database of citrus diseases and insect pests was constructed by Protege ontology editor,and then the ontology information of citrus diseases and insect pests was extended by knowledge reasoning technology.Meanwhile,SWRL rule library and Pellet reasoner are written to complete the reasoning and completion of knowledge atlas under the guidance of experts in related fields.Through the command operation of the database,the knowledge in the ontology information base is stored in the Neo4 j graph database to realize the transformation from the ontology database to the knowledge map.Finally,the new knowledge of citrus diseases and insect pests was edited into several natural sentences,and then annotated for the expansion of the data set.Bi LSTM-CRF was used for training.Experimental results show that compared with Bi LSTM-CRF model before data expansion,the accuracy,recall rate and F1 value of Bi LSTM-CRF model after data expansion are improved by 0.31%,0.59% and 0.47%,respectively.(4)A citrus disease and insect pest query system based on knowledge map was designed and implemented with representation layer,business processing layer and data access layer as data source.The system could guide citrus planting and production effectively and provide scientific basis for the control of citrus disease and insect pest.