Context Sentiment Classification Based on Improved Deep Extreme Learning Machine

In order to improve context sentiment classification accuracy, a sentiment classification algorithm based on deep extreme learning machine with the nearest neighbor and sparse representation (NS-DELM) is presented. Firstly, the extreme learning machine (ELM) is combined with the auto encoder. Secondly, the idea of sparsity and nearest neighbor is integrated into the deep network, and the data integrity is maintained through sparse representation in the projection process. Thirdly, the local manifold structure of the data is maintained by the nearest neighbor representation, and the deep features of the data are extracted layer by layer unsupervised; Finally, the least squares is solved by supervised learning for context sentiment classification. The method is applied to the context sentiment classification experiment of shopping comments. Compared with support vector machine (SVM), stacked auto-encoder (SAE) and convolutional neural network (CNN), the experimental results show that NS-DELM algorithm has higher accuracy than other existing algorithms. © 2021. All Rights Reserved.