<p>Sentiment analysis has gained significant attention in the research domain of language understanding. It offers valuable insights into public opinion, customer feedback, and user experiences, but the quantum machine learning (QML) approaches used in it are still lacked and in the theoretical stages. In this paper, we focus on evaluating one of the QML approaches, which is QCNNs in movie reviews classification; this model represents a quantum neural network design inspired by Convolutional Neural Network (CNN) and exclusively employing parameterized quantum circuits. However, classical text data need to be converted into quantum states; for this, we use three quantum encoding techniques, which are angle encoding, amplitude encoding, and instantaneous quantum polynomial encoding. We investigate how different QCNN models distinguished by the number of qubits used and the structures of parameterized quantum circuits can improve the performance of sentiment analysis classification across both artificial and real datasets of movie reviews.</p>