Effectively Indexing the Uncertain Space
Ying Zhang, Xuemin Lin, Wenjie Zhang, Jianmin Wang, Qianlu Lin
Abstract
Ying Zhang, Xuemin Lin, Wenjie Zhang, Jianmin Wang, Qianlu Lin
Abstract
With the rapid development of various optical, infrared, and radar sensors and GPS techniques, there are a huge amount of multidimensional uncertain data collected and accumulated everyday. Recently, considerable research efforts have been made in the field of indexing, analyzing, and mining uncertain data. As shown in a recent book on uncertain data, in order to efficiently manage and mine uncertain data, effective indexing techniques are highly desirable. Based on the observation that the existing index structures for multidimensional data are sensitive to the size or shape of uncertain regions of uncertain objects and the queries, in this paper, we introduce a novel R-Tree-based inverted index structure, named UI-Tree, to efficiently support various queries including range queries, similarity joins, and their size estimation, as well as top-k range query, over multidimensional uncertain objects against continuous or discrete cases. Comprehensive experiments are conducted on both real data and synthetic data to demonstrate the efficiency of our techniques.
OpenAlex reports 24 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
With the rapid development of various optical, infrared, and radar sensors and GPS techniques, there are a huge amount of multidimensional uncertain data collected and accumulated everyday. Recently, considerable research efforts have been made in the field of indexing, analyzing, and mining uncertain data. As shown in a recent book on uncertain data, in order to efficiently manage and mine uncertain data, effective indexing techniques are highly desirable. Based on the observation that the existing index structures for multidimensional data are sensitive to the size or shape of uncertain regions of uncertain objects and the queries, in this paper, we introduce a novel R-Tree-based inverted index structure, named UI-Tree, to efficiently support various queries including range queries, similarity joins, and their size estimation, as well as top-k range query, over multidimensional uncertain objects against continuous or discrete cases. Comprehensive experiments are conducted on both real data and synthetic data to demonstrate the efficiency of our techniques.
Key concepts: Search engine indexing, Computer science, Uncertain data, Joins, Data mining, Range query (database), Field (mathematics), Tree (set theory)