Diseases and pests articles identification is an important pre-task of natural language processing in the field of diseases and pests. It is of great significance to develop a fast and accurate method for diseases and pests articles identification. In order to solve the problems of insufficient learning of semantic features and insufficient use of context information in the process of diseases and pests articles identification, a neural network model of attention pooling based bi-directional long-short term memory (AP-LSTM) was proposed, which was based on attention pooling strategy and bi-directional long-short term memory (BiLSTM). The model adopted the stacked LSTM structure, which improved the learning ability of semantic features. In the stacking operation, the input vector and output vector were concatenated to further enhance the representation of semantic information. Then, a pooling strategy based on the attention mechanism was used to assign different weights to different words, so that the model can make full use of context information while grasping the keywords. The experiments were carried out on a self annotated dataset with 2500 labeled samples, including 1439 positive cases and 1061 negative cases. The precision, recall, F1 score, and accuracy of the proposed AP-LSTM model on the dataset were 92.67%, 97.20%, 94.88%, and 94.00%, respectively. The experimental results showed that the proposed AP-LSTM model can effectively identify pest literature.