Research On Intelligent Question Answering System Of Apple Diseases And Insect Pests Based On Knowledge Graph

Nowadays,as the largest apple producing and consuming country in the world,China has a wide planting area and many kinds of coverage,while the occurrence of apple diseases and pests is the key factor that directly affects apple yield and quality.The species of apple diseases and insect pests are gradually increasing and the threat of diseases and insect pests is increasingly rising due to the transformation of apple planting mode and the continuous expansion of planting area.When practitioners encounter problems of apple diseases and insect pests,how to quickly and accurately obtain relevant knowledge of apple diseases and insect pests and answer questions has become the main task.These natural language questions are recognized and analyzed and the straightest answers are obtained by question answering system.The emergence and development of knowledge atlas has laid a solid foundation for the high-quality knowledge base of intelligent question-answering system and promoted its wide application in all walks of life.In view of the above considerations,in this study,based on the knowledge graph of apple diseases and insect pests,an intelligent question and answer system for apple diseases and insect pests was designed and developed by integrating the application of deep learning technology in the natural language of apple diseases and insect pests.The main work and achievements are as follows:(1)Research on the construction of knowledge graph of apple diseases and insect pests.The construction of knowledge graph in this paper is based on the top-down design concept,which is composed of four stages: mode layer construction,data acquisition,data layer construction and knowledge storage.The ontology layer is constructed according to the characteristics of apple diseases and insect pests.Unstructured data is obtained from books and semi-structured data is obtained by Scrapy technology.The data layer extracts entities by using the BIOES annotation method and applying Bi LSTM-CRF model,establishing the triplet of apple diseases and pests.Finally,the load CSV method is used to store the triple form into the Neo4 j database to complete the knowledge storage.By adjusting the training parameters of the model,the F1 value of Bi LSTM-CRF model is 85.97%,which is increased by 10.64%and 8.85% respectively compared with the traditional HMM model and CRF model.(2)Research on algorithm design of question answering system.Based on the knowledge graph of apple diseases and pests,the text vector set is constructed based on Word2 Vec word vector model,the question intention is classified through Text CNN model,and then the question entity is extracted in the form of Aho-Corasick algorithm and Bi LSTM+CRF.The intention and question entity are combined into question triples.Finally,cypher is used to construct query sentences to retrieve answers from question triples.The questions of entity type,entity relationship type and entity attribute type are tested and evaluated respectively,and the average F1 value is 85.08%.(3)Research on the construction of Q&A: a system based on apple pest knowledge graph.Adopted MVC model and combined Flask D3 visualization technology,The system completes the construction and interaction of the front and back end of Knowledge graph Q&A module,Relational retrieval module and Visual module.which realized the solution and visualization of hazard symptoms,hazard parts,control methods and therapeutic agents in the field of apple diseases and pests.