Research On Knowledge Map Construction And Knowledge Extraction Method Of Agricultural Diseases And Pests

China is the world’s largest producer of agricultural products and the world’s largest food consumer.However,due to the wide variety of diseases and pests and the difficulty of prevention and control,more than 10% of our agricultural food is lost due to diseases and pests every year,and there are more than 1,400 common diseases and pests.Although the innovation and popularization of science and technology promote the steady progress of the construction of agricultural diseases and insect pests,it is still a difficult problem to solve how to conduct unified management and accurate inquiry of the widely distributed and complex data information in the field of agricultural diseases and insect pests.In this paper,the knowledge map was constructed based on the knowledge of agricultural diseases and pests,and the knowledge was stored using the Neo4 j technology of the map database.Meanwhile,the accurate and fast semi-automatic entity relationship joint extraction model was proposed,which made a beneficial exploration for the construction and intelligent application of the knowledge map in the field of agricultural diseases and pests.Specific research contents are as follows:(1)Establishment of domain knowledge map of agricultural diseases and pests.This paper analyzes the important problems existing in agricultural development at present,and introduces the theoretical basis of knowledge graph and its key technologies.Aiming at the existing problems,the framework and research ideas of intelligent system of agricultural diseases and pests based on knowledge map were proposed,and the specific implementation scheme was given from the perspectives of knowledge extraction,semantic expression and reasoning mechanism,combining with various kinds of diseases and pests involved in the agricultural production process.At the same time,the corresponding functional modules were established according to the application requirements of different levels,and the construction process of agricultural pest map network model was elaborated from the perspectives of database,knowledge base and reasoning engine,so as to further improve the system functions and provide a reference scheme for visual interaction interface.Finally,based on the data of agricultural pests and diseases and combined with the existing research results,the specific implementation plan is planned to realize the construction of its knowledge map,and then the effectiveness and feasibility of the model is verified by designing experiments.(2)A semi-automatic knowledge extraction model based on XLnet-Bi LSTM-CNN-CRF is proposed.In view of the low efficiency of the existing text knowledge extraction models and the shortcomings of the extracted data,a joint entity relation extraction method based on knowledge representation is proposed in this paper,which can realize effective processing of multi-class label data sets.According to the feature of correlation between word vectors in different types of text,XLnet is used for data pre-training,Bi LSTM is used to process the training data,and then CRF is combined to extract the features most relevant to the target category.Convolutional neural network CNN is combined to extract the local features of the current word and text classification.Finally,CRF module is used to decode the output results of Bi LSTM module,train and learn the probability and constraint conditions of label transfer,prevent illegal labels,and obtain the final prediction annotation sequence.The experimental results show that LSTM improves the recognition accuracy to a certain extent,while CRF can effectively avoid overfitting phenomenon.In addition,the algorithm also has good generalization ability and strong learning speed,which can well meet the requirements of practical application.(3)Design and implement agricultural disease and pest knowledge map system based on natural language query.In this study,knowledge graph technology,GRPCframe,React and other frameworks are integrated to realize the development and interaction of the front and back end of the system.Using Spark and Naive Bayes to realize knowledge query based on natural language questions,the common problem expression statements and feature words in the field of agricultural diseases and pests are predefined,and this is taken as a training sample.The problem classification model is trained by Spark,and Cypher query language is generated according to feature words of query statements and classification labels.Search answers on Neo4 j to realize the intelligent query and answer of natural language questions.To sum up,this paper has carried out a series of related work on the subject of agricultural diseases and insect pests.Firstly,the research status and application of knowledge graph at home and abroad are described,and the deficiencies in this field are pointed out.Then the purpose and significance of the research are explained.Secondly,the construction technology and characteristics of knowledge map in the field of agricultural diseases and pests are elaborated,the current mainstream text knowledge extraction framework is discussed,and an entity relationship joint extraction model based on XLnet+Bi LSTM+CNN+CRF is proposed,and the feasibility and efficiency of the proposed method are verified by experiments.To help people quickly and effectively from a large number of jumbled text content to obtain valuable information resources.Finally,the knowledge map system in the field of agricultural diseases and insect pests is designed and developed.Through demand analysis and architecture design,the knowledge map system can accurately query knowledge and display it visually through examples,so as to provide scientific guidance for the prevention and control of agricultural diseases and insect pests.