| China is rich in flower resources,the flower industry chain is gradually maturing,and the flower market is growing rapidly.However,in terms of flower production process and benefits,there are still problems such as high professional knowledge requirements,high production management costs,and low availability of production knowledge.At the same time,with the development of information technology,a large number of flower production knowledge is scattered in the network,and it is difficult for flower producers to use search engines to quickly acquire and extract the required knowledge.Therefore,it is necessary to effectively manage and use flower production knowledge to improve the level of flower production,reduce the loss of flower in production,and achieve efficient and high-quality flower production.However,the traditional relational database knowledge management method cannot effectively represent and store these knowledge,and there are problems such as the inability to integrate heterogeneous data,the inability to efficiently express the relationship between data,and the inability to refine knowledge.According to the above analysis,the knowledge of flower production is organized and managed by means of knowledge graph and knowledge representation through semantic network.Based on the knowledge graph of flower production,a knowledge reasoning method is proposed to form an auxiliary decision-making system to make auxiliary decisions for all businesses in the flower production process.（1）The knowledge management method of flower production based on knowledge graph is proposed.Collect,sort and analyze the knowledge data of flower production,including the knowledge data of flower planting and flower diseases and pests,extract the key elements of flower production,build the ontology model of flower production on the basis of reusing the existing knowledge system.According to different data types of knowledge,the corresponding knowledge extraction method is adopted.Among them,the structured flower planting data is automatically extracted in triples by constructing the mapping relationship with the ontology model.For unstructured flower diseases and pests data,the extraction framework of integrating head-end entity separation"01"tagging method,A Lite Bidirectional Encoder Representations from Transformers（ALBERT）and cascade tagging model with part of speech features(CasPOSRel)is proposed to extract triples.The experiments show that the F1 value of the tagging methods,pretrained model and extraction model in proposed extraction framework is increased by 0.88,4.90 and 8.57 percentage points compared with baseline methods,and the F1 value of the extraction result is 95.07%.Then use the custom RDF2PG mapping algorithm to store the extracted triples into the Neo4j database according to the ontology model in the RDF diagram to complete the storage and management of flower production knowledge.（2）The design and implementation methods of flower production decision rules based on Drools and knowledge graph are proposed.Taking cut flowers of Anthurium andraeanum as the research object,through the analysis of the flower production process,the paper summarizes the content of the decision business in each flower production process,and designs the rules of the decision content.Based on the characteristics of knowledge storage and management of flower production knowledge graph and ensuring the efficiency of decision making,this paper proposes a knowledge reasoning method that does not directly infer on the storage structure,but uses the Drools rule engine to infer after acquiring the required knowledge.The reasoning of knowledge graph and rule engine is tested by an example,which verifies the reliability and flexibility of the rules written and the reasoning method used in this paper.（3）A flower production auxiliary decision-making system based on knowledge graph is built.Taking the cut flower production of Anthurium andraeanum as an example,through the overall design of the auxiliary decision-making system for flower production,the above research content is integrated through the Spring Boot framework,and the main functions of the auxiliary decision-mkaing system are realized.It can provide decision ideas for the business content of flower production environment,work content,disease and pest control,and provide intelligent scientific decision tools for flower production.Through effective management of flower knowledge and efficient decision making with knowledge,a flower production assistant decision system is formed,which provides practical and feasible implementation plan in the field of intelligent flower production,and has certain reference value for promoting modern flower production. |
