The growing number of fake profiles on social media has made it necessary to develop intelligent models for identifying and removing them. This study proposes a novel approach for identifying fake profiles using temporal sentiment analysis. The model collects large datasets from various social media platforms, which are then processed using an ensemble sentiment analysis engine. This produces temporal sentiment patterns, which are used to train a 1D Convolutional Neural Network (1D CNN) for the efficient identification of fake profiles. The model is based on the theory that the data communicated by fake users always converges, and was observed to improve fake-profile detection accuracy by 5.9% with higher precision and recall compared to standard methods. This proposed model can be applied to multimodal social media interfaces.