Design And Implementation Of Intelligent Q?A System For Zanthoxylum Bungeanum Planting Based On Knowledge Graph

With the proposal of the National Rural Revitalization Strategy,agriculture will gradually become information-based in the future.In recent years,the scale of network data has increased exponentially,but the Zanthoxylum bungeanum(ZB)planting knowledge still exists mainly in the form of books,which makes it difficult for ZB planting practitioners to use the Internet to obtain timely and professional planting knowledge.In order to solve the above practical problems and help the personnel engaged in ZB planting to quickly obtain professional planting technology,this study establishes the ZB planting knowledge graph,designs and implements the intelligent question and answer system based on the knowledge graph.At present,the intelligent Q?A service in ZB planting in China is not mature,so this research is of great significance to speed up the informatization of ZB planting industry.The main work of this paper is as follows:(1)Study on the construction method of ZB planting knowledge graph.In view of the lack of professional knowledge graph in the field of ZB planting,the knowledge graph of ZB planting is constructed according to the top-down idea.In view of the time-consuming and laborious construction of traditional knowledge graph,Bi-LSTM-CRF model is used to complete the recognition of ZB planting named entities,and a ZB planting entity relationship classification model based on R-GCN is designed to complete the automatic extraction of ZB planting triple knowledge.The results show that the F1 index of entity recognition is improved by 11.43% compared with LSTM-CRF model;Compared with GCN,F1 index increased by 7.39%.(2)Study on the method of completing the knowledge graph of ZB planting.Aiming at the inherent incompleteness of knowledge graph,a complement model of ZB planting knowledge graph based on embedded representation and CNN is proposed.By fusing the information of different nodes of the triple,this method improves the feature expression ability of the triple,and forms a vector matrix with the original triple as the input of the convolution neural network.The experimental results show that on the data set FB15K-237 compared with Conv E model,the MRR index is improved by 6%,the Hit@1 index is increased by 12.7%.(3)Research on intelligent question answering method of ZB planting.In view of the short length of ZB planting questions and the difficulty of obtaining sufficient features by traditional models,the ZB planting question entity recognition model based on Bert-Bi-LSTM-CRF and the ZB planting question intention understanding model based on Bert-CNN-Softmax are designed.Build different types of cypher query statements according to different situations to complete the answer query.The experimental results show that the F1 value of question entity recognition is improved by 3.4% compared with Bi-LSTM-CRF model;Compared with Text RNN model,the F1 value of question classification is increased by 8.08%.(4)Design and implementation of ZB planting intelligent question and answer system based on knowledge graph.Aiming at the problem that it is difficult for ZB planting practitioners to use the Internet to obtain professional planting knowledge,an intelligent question and answer system for ZB planting based on knowledge graph is designed and implemented.Based on the B/S structure,the system takes the ZB planting knowledge graph as the answer data source,and uses the ZB planting question entity recognition model,question intention understanding model and ZB planting knowledge graph completion model to complete the functions of ZB planting intelligent question answering and ZB planting knowledge graph completion.It is convenient for users to obtain professional planting knowledge and greatly promote the information development of ZB planting industry.