Construction And Application Of Knowledge Graph Of Crop Diseases And Insect Pests Based On ALBERT

Crop diseases and insect pests have the characteristics of wide spread and easy to cause disasters,which have a negative impact on the healthy development of our country’s agriculture.With the increasing application of equipment such as soil detectors,temperature and humidity sensors,and drones to the agricultural field,a large amount of agricultural information data is generated.However,the source and composition of these data are relatively chaotic,and it is difficult to use them effectively.Most of the knowledge of pest control is in professional literature and books,which requires high access and is time-consuming and labor-intensive.There are a lot of redundant data in the results obtained by traditional retrieval methods,and it is difficult for farmers and plant protection experts to obtain the required information accurately and efficiently.In order to integrate disease and insect pest information resources and improve the accuracy of crop disease and insect pest control,this paper uses knowledge graph to manage crop disease and insect pest knowledge,and designs and implements an automatic question answering system based on crop disease and insect pest knowledge graph.The main work of this paper is as follows:(1)Construction of knowledge graph of corn,rice and wheat pests and diseases.In this paper,we first use technologies such as text scanning and crawling to collect data on diseases and insect pests of corn,rice and wheat,and combine Numpy and Pandas tool libraries to preprocess the data;then use Protégé tools to design the boundaries of ontology model management knowledge extraction,and use deep learning and Natural language processing and other technologies realize semi-automatic knowledge extraction of crop pests and diseases field data;finally,the extracted triple knowledge is stored through Neo4 j graph database.(2)In view of the difficulty and low efficiency of manual construction of the knowledge graph data layer of corn,rice and wheat pests and diseases,this paper studies the semi-automatic construction method of the data layer.In this paper,annotation tools are used to assist in labeling to improve labeling efficiency and accuracy.At the same time,combined with the ALBERT-Bi LSTM-CRF model,the semi-automatic extraction of crop pest and disease knowledge is realized.Experiments show that the joint model used in this paper has an accuracy rate of 89.32%.(3)Research on question answering algorithm based on knowledge graph of corn,rice and wheat pests and diseases.In this paper,the method based on deep learning is used to realize the question and answer function.First,the trained ALBERT-Bi LSTM-CRF model is used to extract the key entities of the question sentence,and then combined with the question sentence classification results of the ALBERT-Text CNN model,the corresponding query sentence is generated,and finally for the user Returns an answer that follows natural language logic.Experiments have confirmed that the accuracy rate of the question classification model used in this paper reaches 90.98%.(4)Develop an intelligent question answering service system based on the knowledge graph of corn,rice and wheat pests and diseases.Combined with the knowledge graph of crop diseases and insect pests constructed in this paper,a disease and insect pest knowledge service platform is built to realize the visual display and knowledge question and answer functions of crop disease and insect pest knowledge.