Search

Your search keyword '"Liu, Guanli"' showing total 3 results
Start Over Author "Liu, Guanli" Publication Year Range Last 3 years
3 results on '"Liu, Guanli"'

1. Efficient Cost Modeling of Space-filling Curves

Author

Liu, Guanli, Kulik, Lars, Jensen, Christian S., Li, Tianyi, and Qi, Jianzhong

Subjects
Computer Science - Databases
Abstract

A space-filling curve (SFC) maps points in a multi-dimensional space to one-dimensional points by discretizing the multi-dimensional space into cells and imposing a linear order on the cells. This way, an SFC enables the indexing of multi-dimensional data using a one-dimensional index such as a B+-tree. Choosing an appropriate SFC is crucial, as different SFCs have different effects on query performance. Currently, there are two primary strategies: 1) deterministic schemes, which are computationally efficient but often yield suboptimal query performance, and 2) dynamic schemes, which consider a broad range of candidate SFCs based on cost functions but incur significant computational overhead. Despite these strategies, existing methods cannot efficiently measure the effectiveness of SFCs under heavy query workloads and numerous SFC options. To address this problem, we propose means of constant-time cost estimations that can enhance existing SFC selection algorithms, enabling them to learn more effective SFCs. Additionally, we propose an SFC learning method that leverages reinforcement learning and our cost estimation to choose an SFC pattern efficiently. Experimental studies offer evidence of the effectiveness and efficiency of the proposed means of cost estimation and SFC learning.

Published
2023

2. Learning Spatial Indices Efficiently

Author

Liu, Guanli and Liu, Guanli

Abstract

Machine learning and database management systems have been extensively researched for many years. A database management system, while supporting data persistence and stable queries, can often encounter efficiency issues when building indices for querying big data. In recent years, machine learning is gradually used in databases to solve efficiency issues such as knob tuning, index selection, cost estimation, and index building. Most problems are solved by using regression models and reinforcement learning. Here, using machine learning techniques for index building leads to a new type of index structures named the learned index, which has shown better query performance than traditional indices, e.g., B-trees and R-trees. However, the efficiency of building learned indices is still a key challenge, especially when facing large-scale datasets. The reason is that index building is based on the whole dataset, which requires a full dataset scan in each epoch, and there is at least one epoch for building a learned index. Existing learned indices suffer from this issue, which hinders the application of indices in the database management system. In this thesis, we address this efficiency issue for index learning over a special type of data, the spatial data, i.e., data associated with geographical location information, the volume of which is rapidly growing due to the prevalence of smart mobile devices, the Internet of Things, and 5G networks. We study four research problems on effective and efficient spatial data indexing using machine learning techniques. The first problem is an empirical study of two learned spatial indices RSMI and ZM, which support point, range, and kNN queries. These indices have reported better query performance than a traditional spatial index, the R-trees. However, there is no open-source code or extensive basis that supports testing and evaluating these learned indices against other traditional spatial indices for large real-world datasets. We addr

Published
2023

Catalog

Books, media, physical & digital resources
See catalog results