User Opinion Prediction for Arabic Hotel Reviews Using Lexicons and Artificial Intelligence Techniques

Opinion mining refers to the process that helps identify and classify users’ emotions opinions from any source, such as an online review. Thus, opinion provides organizations with insight into their reputation based on previous customers’ regarding services or products. Automating in different languages is still important topic of interest for scientists, including those using Arabic language, especially since potential customers mostly do not rate explicitly. This study proposes ensemble-based deep learning approach fastText embeddings proposed emoji emoticon lexicon predict user opinion. For testing purposes, uses publicly available HARD dataset, which includes hotel reviews associated ratings, starting one five. Then, by employing multiple resources, it experiments generated features dataset combining shallow approach. To best our knowledge, this first create a considers emojis emoticons its prediction. Therefore, mainly helpful contribution literature related lexicons. Compared other studies found five-star rating prediction accuracy reached increase 3.21% balanced 2.17% unbalanced dataset. The work can support new direction automating unrated social media, five levels, provide stakeholders precise idea about service product quality, instead spending much time reading learn information.