As near-infinite amount of data are becoming accessible on the Web, it is getting more and more important to support intelligent query mechanisms, to help each user to identify the ideal results of manageable size. As such mechanism, skyline queries have gained a lot of attention lately for its intuitive query formulation. This intuitiveness, however, has a side-effect of generating too many results, especially for high-dimensional data, to satisfy a wide range of user's needs. Our goal is to support personalized skyline queries as identifying "truly interesting" objects based on user-specific preference and retrieval size k. While this problem has been studied previously, the proposed solution identifies top-k results by navigating a "skycube", which incurs exponential storage overhead to data dimensionality and excessive one-time computational overhead for skycube construction. In contrast, we develop novel techniques to significantly reduce both storage and computation overhead. Our extensive evaluation results validate this framework on both real-life and synthetic data.