Aspect-Based Sentiment Analysis (ABSA) plays a crucial role in understanding finegrained customer feedback, particularly in domains like hospitality where specific aspects of service often influence overall satisfaction. However, non-English languages such as German face a scarcity of readily available corpora and evaluated methods for ABSA, making it a challenging problem. This paper addresses this gap by utilizing BERT-based transformer models, known for their exceptional performance in context-sensitive natural language processing tasks, to perform ABSA in a multi-label classification setting. We demonstrate our approach on a novel dataset of German hotel reviews that we have collected and annotated from TripAdvisor, thus contributing a new resource to the field and proving the effectiveness of our methodology. With achieving a micro f1-score of up to 0.91 for aspect category classification and 0.81 for end-to-end ABSA, our approach aligns with the performance of similar methods on other German-language datasets and surpasses performance achieved on English language datasets in the hotel domain.