Research And Implementation Of Question Answering System Based On Diseases And Pests Knowledge Graph Of Lycium Barbarum

"Internet+agriculture" has made important achievements in promoting the development of agricultural production,disease control and sales through cutting-edge technologies such as big data and artificial intelligence."Intelligent agriculture" has become an important measure of science and technology to assist agriculture,which of agricultural production and agricultural disease control.Furthermore "Intelligent agriculture" improves agricultural production efficiency and level by using intelligent information.The industry chain of Lycium barbarum,as a characteristic industry in Ningxia region,which will surely promote the development of intelligent platform of Lycium barbarum cultivation,disease control,expert consultation and sales with the influence of factors such as the increase of planting scale,variety innovation and the change of sales mode.Knowledge graph has gradually developed to a new knowledge management and knowledge creation technology from the information representation form used by Google to enhance the search engine effect in the early stage.And knowledge graph is an important basis for improving the semantic retrieval,intelligent questions and answers and other intelligent applications.Through investigation and analysis,firstly this paper preprocessed the unstructured text data of Lycium barbarum pests and diseases,extracted the pests and diseases knowledge and put it into the Neo4j database as the data layer,which provides the data basis for the visual retrieval of lycium barbarum knowledge and intelligent QA system.Secondly,knowledge extraction is the most basic process and key steps in the construction of knowledge graph,so this paper determined the entity class based on the characteristics of the diseases and insect pests of Lycium barbarum in Ningxia.Then this paper proposed a tagging rules considering physical boundary in the field of plant diseases and insect pests,and used the rules to conduct the entities annotations of disease name,symptoms,and agriculture drug.In addition,considering the fuzzy boundary of pest and disease entities,this paper studied and designed entity recognition model of the pest and disease based on ALBERT model.Because of using the ALBERT pre-training language model to replace Word2vec tool as the input of the model word vector,this model improved the recognition accuracy and time efficiency compared with the traditional Bilstm-CRF model and the Bert-Bilstm-CRF model.The question-and-answer system is a common and important application form in the industrial chain processes such as auxiliary Lycium barbarum cultivation,pest and disease control,sales and traceability.Compared with the traditional retrieval system,which has information overload and is difficult to meet the individual and specific needs of users,the QA system based on knowledge graph is more in line with the actual needs of users in the way of visualization and knowledge reasoning.Based on the Lycium barbarum knowledge graph,this paper designed and implemented the QA system of Lycium barbarum disease,which includes the display knowledge graph of Lycium barbarum disease and insect pest,disease relationship retrieval and knowledge QA function.In the knowledge QA part,the answer of the question is obtained by transforming the questions input by the user into the query statement of knowledge graph.In the part of knowledge graph display and disease relation retrieval,the system visualized the retrieval results and related knowledge structure according to the knowledge graph.In this paper,a question and answer system based on knowledge graph is designed to provide an application platform for the knowledge service of Lycium barbarum's industry chain and promote the development of agricultural intelligent informatization.