Multi-criteria decision making has been made possible with the advent of skyline queries. However, processing such queries for high dimensional datasets remains a time consuming task. Real-time applications are thus infeasible, especially for non-indexed skyline techniques where the datasets arrive online. In this paper, we propose a caching mechanism that uses the semantics of previous skyline queries to improve the processing time of a new query. In addition to exact queries, utilizing such special semantics allow accelerating related queries. We achieve this by generating partial result sets guaranteed to be in the skyline sets. We also propose an index structure for efficient organization of the cached queries. Experiments on synthetic and real datasets show the effectiveness and scalability of our proposed methods.