中央研究院 資訊科學研究所

In recent years, the significance of data has been on the rise, and efficiently storing, accessing, and analyzing the increasing volume of data is crucial for diverse application scenarios today. In many application scenarios, managing multidimensional data faces greater challenges, such as supporting more efficient data access. While numerous existing indexing data structures have been proposed for multidimensional data, their designs have not been fully optimized for modern nonvolatile memories. Consequently, they might encounter serious performance degradation during insert or query operations, or exacerbate the memory or storage space utilization. In this talk, we will outline the grand challenges of multidimensional data management on modern nonvolatile memories, highlight the limitations of existing multidimensional index data structures, and present a novel approach called the WARM-tree for persistent memories. Moreover, we shall also present the potential future directions of development in this area.

Shin-Ting Wu is currently an assistant professor at the Department of Computer Science in National Chengchi University (NCCU). Previously, she was a postdoctoral researcher at the Institute of Information Science (IIS), Academia Sinica, Taiwan (R.O.C.). She received her Ph.D. degree in the Department of Computer Science from the National Tsing Hua University, Taiwan, in 2024. Her research interests focus on the development of space-efficient multidimensional indexing data structures for non-volatile memories (NVMs) and the optimization of data mining and machine learning algorithms to leverage the performance characteristics of emergent NVM.