Research And Design Of Sheep Breeding Knowledge Question Answering System Based On Knowledge Grap

At present,most of the knowledge of sheep breeding exists in unstructured form,which is difficult to obtain quickly and accurately.In view of this problem,this study uses knowledge graph to organize knowledge in the field of sheep breeding,uses deep learning model to understand the semantic information of natural language questions,and designs a sheep breeding question and answer system based on knowledge graph,so as to realize the deep integration of science and technology and industry and promote the intelligent development process of breeding industry.The main research contents and achievements are as follows:(1)Build a knowledge graph of sheep farming based on natural language processing.A model based on deep learning knowledge graph data layer is proposed.The model uses the BIO annotation strategy to perform text vectorization processing through BERT preprocessing,and the processing results generate global and local features of text in bidirectional recurrent units and convolutional neural networks,and then connects the two-way long short-term memory network to automatically extract text information,uses conditional random fields to constrain the rationality of labels at the sentence level,completes the recognition of sheep breeding entities,extracts sheep breeding triplet data according to label categories,and forms a knowledge graph.The experimental results show that the accuracy of tridata extraction in sheep breeding reaches95.98%.(2)Design a knowledge question and answer system for sheep breeding based on knowledge graph.The BERT model is used to represent the character-level feature vector of sheep breeding text to solve the problem of question diversification.The convolutional neural network is used to extract the features of the question and complete the understanding of the question intention.For the user’s question entity recognition problem,the BERT-Bi LSTM-CRF model is used to perform entity recognition for the question and the cosine similarity calculation is used to complete the question entity recognition.The effect of question intention comprehension was verified by experiment,and the value exceeded 99.80%.(3)Develop a knowledge question and answer system for sheep farming based on knowledge graph.A knowledge question and answer system for sheep breeding based on knowledge graph is designed.The system takes the sheep breeding knowledge graph as the data base and FLASK as the development framework,and uses the deep learning model to realize the sheep breeding entity recognition and knowledge question and answer functions,which can save the time of users to obtain information and contribute to the intelligent development of the sheep breeding industry.