Computing and Informatics

Sentiment classification plays a pivotal role in natural language processing (NLP), and prior research has established the efficacy of utilizing convolutional neural networks (CNNs) and long short-term memory (LSTM) in this task. However, these approaches suffer from individual performance limitations: CNNs are limited to extracting local information and fail to express context information adequately, while LSTM networks excel at extracting context dependencies but exhibit long training times. To address this issue, we propose a novel text classification algorithm based on a hybrid CNN-LSTM model that leverages the strengths of both approaches and overcomes their limitations by combining them. Our approach is evaluated on the IMDB dataset, and we present a hyperparameter optimization framework utilizing Random Search to increase the likelihood of producing an optimally performing model.