Natural Language Processing is a sub-field of Artificial Intelligence focusing on the interaction of human languages and computers. As the way which computers communicate and treat data is different to the way natural languages are represented, there is a range of unique challenges in this field. Text Classification is a fundamental NLP task, in which Convolutional Neural Networks (CNN) have shown state of the art performance in. However, designing an optimal CNN model architecture can often require specialist knowledge, as well as be an arduous process for each distinct problem. This paper presents two Genetic Algorithms (GA), single-view and multi-view, which can automatically learn CNN architectures for text classification problems.