Time-aware Knowledge Graphs (TKGs) extend traditional Knowledge Graphs (KGs) by incorporating temporal dimensions, enabling improved handling of time-sensitive data in domains such as finance, healthcare, and organizational management. However, temporal data raises privacy concerns due to potential pattern recognition and semantic similarity attacks, which are often overlooked by current anonymization methods. Existing techniques inadequately address semantic context and fail to determine optimal k and l values in k-anonymity and l-diversity frameworks. To address these challenges, we propose a novel approach leveraging sentence embeddings to capture semantic nuances and applying k-anonymity and l-diversity to enhance privacy. We formulate a Multi-Objective Optimization Problem to select optimal k and l values by minimizing quasi-identifier distances, maximizing sensitive attribute diversity, and reducing information loss. Experiments on three real-world datasets demonstrate that our method achieves lower information loss and superior anonymization compared to benchmarks. Further, downstream task evaluations using KG embedding models reveal that anonymized KGs outperform non-anonymized counterparts in Mean Reciprocal Rank (MRR) and Hits@n metrics, although with slightly reduced triple confidence scores. These results highlight the effectiveness of our approach in balancing robust anonymization with practical utility.