Animal disease-related information serves as an important reference. This paper focuses on the study of Named Entity Recognition (NER) in animal disease texts using a dataset constructed from animal disease books. However, the dataset exhibits imbalanced distribution of different types of entities, which affects the learning effectiveness of the models. In this method, firstly, a small amount of data was enhanced based on the template data augmentation method, and then the overall data was enhanced based on the improved EDA method. On the enhanced dataset, the accuracy of NER based on the BiLSTM network and CRF model increased by 3%. The recall rate saw a significant improvement of 17%, and the F1 value increased by 10%. The experimental findings suggest that the data augmentation technique presented in this paper can improve the model’s learning effectiveness on the dataset related to livestock and poultry diseases.