In the ever-evolving digital era, social media, especially platforms like TikTok, has become a primary channel for users to share opinions, experiences, and expressions. However, the increasing prevalence of hate speech in reviews on the Google Play Store for the TikTok app indicates the need for a sophisticated approach to identify and classify harmful content. This research is aimed to optimize the classification of hate speech in Google Play reviews of the TikTok app by integrating Term Frequency-Inverse Document Frequency (TF-IDF), Differential Evolution, and Word2Vec within a Recurrent Neural Network (RNN) model. The TF-IDF technique will be used to extract relevant features from a review, while Differential Evolution will be applied to efficiently optimize the model parameters. The use of Word2Vec will enhance the representation of words in the context of app reviews, whereas the RNN model will enable the recognition of temporal patterns in hate speech. The results of this research are expected to contribute significantly to the improvement of hate speech classification on digital platforms focused on app reviews.