Research On The Construction Of Knowledge Graph Of Crop Diseases And Pests

There are many kinds of crop diseases and pests in China.According to statistics,there are more than 1400 kinds of common crop diseases and pests only.The impact of diseases and pests on crop output and farmers’ income is very serious every year.At the same time,because of crop diseases and pests data sources,data representation,storage,organization,management way is different,the resource is in a state of disorder and relatively chaos,make scientific research personnel,the people difficult to effectively query and use various crop diseases and pests information resources,which leads to knowledge is hard to be used efficiently.With the fourth data intensive scientific research paradigm and knowledge graph is put forward,using the knowledge graph technology was carried out on the crop diseases and pests of multi-source heterogeneous data integration,fully excavate implicit knowledge in the text,to help users to clarify the knowledge structure,improve the efficiency of knowledge acquisition and dissemination,promote agricultural intelligent knowledge service ability is of great significance.In this regard,this paper draws on the integration of professional and high-quality knowledge resources from multiple sources and carriers and introduces deep learning algorithm to carry out knowledge extraction,so as to explore the construction methods and technical routes of semi-automated knowledge graph in the agricultural field.At the same time,this paper tries to integrate the knowledge graph of crop diseases and pests with the general knowledge graph and the traditional knowledge organization system.Based on the analysis of its application scenarios,a prototype knowledge question answering system is designed and implemented for application verification,thus providing relevant ideas for the practice of new intelligent knowledge service.This paper first introduced the origin and status quo of knowledge graph,and then analyzed the research status and existing problems of crop diseases and pests related ontologies and knowledge graph.Secondly,this paper introduces the concept of knowledge graph and knowledge graph construction need to apply to the relevant technical methods and methods involved include: ontology construction of related knowledge,entity recognition,knowledge storage,knowledge fusion method,related concepts and methods for the development of this research provides the theoretical support.Third,in the construction of knowledge graph model layer,this paper analyzed and studied the objectives and processes of building the ontology of crop diseases and insect pests,mainly based on the THESAURUS of Agricultural Science,the authoritative monograph of Crop Diseases and Pests in China and the literature resources of the National Agricultural Library.The domain ontology of crop diseases and pests was constructed,which included 13 first-level classes,25 second-level classes,20 third-level classes,15 first-level object attributes and 36 first-level data attributes.The ontology was formalized based on OWL+SKOS language.Fourthly,in terms of the construction of the data layer of the knowledge graph of crop diseases and pests,this paper first introduced the construction process of the data layer.Then,based on the BERT-Bi LSTMCRF model,entity link and other technical methods,it completed the semi-automatic knowledge extraction and knowledge fusion experiment of structured and unstructured data,and applied Neo4 j to carry out knowledge storage.A knowledge graph of crop diseases and pests containing 9 types of entities,9 relationships,a total of 16842 nodes and 24303 triples was formed.Finally,based on the constructed knowledge graph of crop diseases and pests,a prototype knowledge question and answer system was designed and implemented for application verification.The results show that the knowledge graph constructed in this paper realizes the formal and standardized description organization and deep integration of knowledge of multi-source diseases and pests,which lays a foundation for carrying out indepth knowledge mining.