KMeans-LSTM model and comment text feature extraction and false comment identification method

The invention discloses a KMeans-LSTM model and a comment text feature extraction and false comment identification method, and belongs to the field of evaluation systems, and the KMeans-LSTM model comprises TF-IDF comment text vectorization, data set balancing based on a K-means clustering algorithm, GloVe pre-training word embedding and LSTM classifier. The comment text feature extraction and false comment identification method comprises the steps of S1, data preprocessing, S2, clustering and data balancing, S3, word embedding and model training, and S4, model evaluation and application. According to the method, the problem of data imbalance is effectively solved, and the learning ability and prediction accuracy of the model are improved by using the deep learning technology. The model improves the recognition rate of false comments, also has high generalization ability, and can be suitable for comment data of different sources and types. By analyzing and applying comment data from different sources, the gen