Research On Crop Diseases And Pests Question And Answer System Based On Knowledge Graph

With the continuous development of digital agriculture,agricultural big data is exploding and the demand of farmers for agricultural information services is increasing,but there is often a large amount of redundant information in search engines that people need to distinguish by themselves,and crop pest and disease data also shows a highly scattered and multi-source heterogeneous situation for farmers,agricultural technicians and related practitioners.To address the above problems,this study classifies various knowledge of crop pests and diseases,and uses a deep learning model to extract triadic data,constructs a knowledge graph based on crop pests and diseases,and designs and implements a crop pest and disease Q&A system based on this knowledge graph.The main research work is as follows:(1)Construction of crop pest and disease knowledge graph.Based on relevant information in the field of crop pests and diseases,this study used various strategies to obtain unstructured data and semi-structured data,and applied Scrapy crawler technology to collect a large amount of information about semi-structured data from popular science websites.The entity extraction of semi-structured text data was achieved and the data was stored into a graph database.In the comparison experiments of crop pest entity extraction,the F1 values of CRF,BiLSTM,BiLSTM-CRF and HMM were 82.74%,75.74%,87.30% and 71.14%,respectively,and the BiLSTM-CRF model got the best results,and it was proved experimentally that in the absence of a large number of artificial features,the BiLSTM-CRF model can automatically and efficiently extract text features in the absence of a large number of manual features.After that,the obtained triads were formed into a knowledge graph and stored in the Neo4 j graph database to finally form a crop pest knowledge graph.(2)Knowledge graph-based question and answer method for crop pests and diseases.The input questions are classified into interrogative sentences and keywords are extracted.In this study,the method of extracting interrogative entities takes the form of Aho-Corasick automaton matching,and query templates are written using the Cypher language in the graph database.Finally,the search results were combined into a smooth and coherent natural language output using the preconstructed answer templates.(3)Development of a knowledge graph-based crop pest and disease Q&A system.The data layer of the system architecture is designed,the model layer and the display layer are three-layer structure,and the separation of the display module and the functional module is realized to realize the user’s question and answer and visualization of the crop pests and diseases related to the symptoms of hazards,the parts of hazards,the prevention and control methods,the treatment methods,etc.to meet the user’s fast and accurate access to information..