Research On Dairy Cow Disease Diagnosis Methods Based On Knowledge Graph And Deep Learning

Efficient and accurate dairy cow disease diagnosis has important practical significance for improving the health level of dairy cows and the management level of pasture.Dairy cow disease has its own characteristics.The individual difference of dairy cow disease is strong,and there are fewer epidemic diseases in clusters.For the sick cow,the clinical diagnosis is generally required.For clinical diagnosis of sick dairy cows,both large-scale farms and retail farmers rely heavily on professional veterinarians.The diagnosis of dairy cow disease differs from that of human diseases,which requires veterinarians to possess general medical knowledge.However,small and medium-sized farms in China’s vast rural areas are extremely short of high-level employees.The diagnosis accuracy and timeliness can not be guaranteed,which results in a great financial loss.At present,ancillary diagnosis of animal diseases is still at the stage of expert systems.However,these expert systems rely heavily on manual rules that are difficult to obtain and expand,and it is hard for nonprofessional users to operate them.Furthermore,the reasoning ability of these systems is so closely related to the number and quality of rules that it is almost impossible to find the complex relationships implied between symptoms and diseases,and between symptoms and symptoms,diseases and pathogens,etc.Therefore,it is urgent to explore an efficient method to assist the intelligent diagnosis of dairy cow diseases.This method will help veterinarians solve the problem of missing knowledge,lower the threshold for the application of diagnosis of dairy cow diseases,and improve the intelligent level of dairy farming.Therefore,this paper studies the efficient diagnosis method of dairy cow disease based on the knowledge graph of dairy cow disease and deep learning algorithm.The method takes the description texts of sick dairy cows as the research object,and extracts the disease feature keywords from the texts based on knowledge graph of dairy disease.Then,the disease feature entities and their context entities are found in the dairy cow disease domain knowledge graph through entity links.After the entity is vectorized,the entity vector is input into the deep learning model,and the dairy cow disease diagnosis result is obtained.The research work carried out in this paper is as follows:(1)Construction of dairy cow disease domain knowledge graph.The professional knowledge of dairy cow disease is obtained from professional books,CNKI literature and encyclopedia website,and it is preprocessed as the data source of knowledge graph construction.According to data characteristics and application requirements,the knowledge graph model layer of dairy cow disease field is designed.Aiming at the situation that knowledge extraction tasks are prone to errors and redundant information,a joint extraction method of entity relationship is adopted,and a named entity recognition model combining MHA and Attention mechanism is studied to improve the accuracy of entity recogni tion.The experimental results show that the F1 value of the named entity recognition model in the dairy cow disease field constructed in this paper reaches 91.81%.(2)Dairy cow disease diagnosis model based on knowledge graph and deep learning.For the feature words in the dairy cow disease status text report,link the dairy cow disease knowledge graph to mine hidden structural knowledge as an additional supplement.The feature words and tacit knowledge are transformed into word vectors by word embeddin g,and the Bi LSTM-CNN hybrid network framework is used to capture global long-term disease features and local features,so as to learn more discriminative feature representations and improve the accuracy of dairy cow disease diagnosis.The experimental results show that the accuracy of the dairy cow disease diagnosis model is 94.57%,and the F1 value reaches 94.89%.(3)Implementation and verification of dairy cow disease auxiliary diagnosis system.Based on the knowledge graph of dairy cow disease domain,an auxiliary diagnosis system of dairy cow disease is developed with the dairy cow disease diagnosis model integrating knowledge graph and deep learning as the logical core.It provides dairy cow disease diagnosis services and knowledge assistance service s for dairy farmers and veterinarians.The system has been applied in many pastures and farmers,which further verifies the effectiveness of the model.This paper proposes an effective dairy cow disease diagnosis method based on knowledge graph and deep learning techniques to achieve both data and knowledge driven dairy cow disease diagnosis.This method provides technical support for efficient,timely and accurate diagnosis of cow diseases,and also provides new ideas and methods for the research of intelligent auxiliary diagnosis of animal diseases.