<![CDATA[Game descriptions on digital distribution platforms play a crucial role in conveying the characteristics of gameplay to players. However, the language complexity of these descriptions varies and may influence players' understanding of the gameplay being offered. This study aims to classify gameplay complexity based on sentence structure in game descriptions using a Natural Language Processing (NLP) approach. The dataset used is the 10k Most Popular Gaming 2025 dataset obtained from Kaggle, with a focus on the game description column. The description data is grouped into three complexity classes: simple, medium, and complex, based on the linguistic characteristics of the text. The research process includes text preprocessing, sentence-structure-based linguistic feature extraction, and data balancing using the balance rank method. Classification is performed using the Logistic Regression, Random Forest Classifier, and Support Vector Machine algorithms. Evaluation results show that the Random Forest Classifier achieves the highest accuracy of 0.85, while Logistic Regression and Support Vector Machine obtain accuracies of 0.81 each. Feature analysis reveals that word count and average sentence length are the most influential features in determining gameplay complexity. Visualization using Principal Component Analysis shows a clear distribution pattern of complexity classes, although some overlap between classes remains. The results of this study demonstrate that sentence-structure-based linguistic analysis is effective in representing gameplay complexity in game descriptions.]]>
