By Cui Yu

In this monograph, we research the matter of high-dimensional indexing and systematically introduce effective index constructions: one for diversity queries and the opposite for similarity queries. vast experiments and comparability experiences are performed to illustrate the prevalence of the proposed indexing methods.
Many new database functions, akin to multimedia databases or inventory fee details platforms, rework vital good points or homes of information gadgets into high-dimensional issues. trying to find gadgets in line with those gains is therefore a seek of issues during this function house. To aid effective retrieval in such high-dimensional databases, indexes are required to prune the hunt area. Indexes for low-dimensional databases are good studied, while every one of these software particular indexes usually are not scaleable with the variety of dimensions, and they're now not designed to aid similarity searches and high-dimensional joins.

Show description

Read or Download High-Dimensional Indexing: Transformational Approaches to High-Dimensional Range and Similarity Searches PDF

Similar structured design books

Java(tm) for S/390® and AS/400® COBOL Programmers

The ebook may still specialize in Java on AS400. additionally it makes use of visible Age that is outmoded should still use Websphere as a substitute. the code isn't transparent because it attempts to match COBOL(structure programing) with Java(Object orientated

Web Work: Information Seeking and Knowledge Work on the World Wide Web

This ebook brings jointly 3 nice motifs of the community society: the looking and utilizing of knowledge through participants and teams; the construction and alertness of information in agencies; and the basic transformation of those actions as they're enacted on the web and the realm extensive internet.

On the Move to Meaningful Internet Systems 2007: OTM 2007 Workshops: OTM Confederated International Workshops and Posters, AWeSOMe, CAMS, OTM Academy Doctoral Consortium, MONET, OnToContent, ORM, PerSys, PPN, RDDS, SSWS, and SWWS 2007, Vilamoura, Portugal

This two-volume set LNCS 4805/4806 constitutes the refereed court cases of 10 foreign workshops and papers of the OTM Academy Doctoral Consortium held as a part of OTM 2007 in Vilamoura, Portugal, in November 2007. The 126 revised complete papers offered have been conscientiously reviewed and chosen from a complete of 241 submissions to the workshops.

Dynamic Data-Driven Environmental Systems Science: First International Conference, DyDESS 2014, Cambridge, MA, USA, November 5-7, 2014, Revised Selected Papers

This publication constitutes the refereed court cases of the 1st overseas convention on Dynamic Data-Driven Environmental platforms technological know-how, DyDESS 2014, held in Cambridge, MA, united states, in November 2014.

Additional info for High-Dimensional Indexing: Transformational Approaches to High-Dimensional Range and Similarity Searches

Sample text

The generation of the signature of the range query is to map a float vector to a small number of bits, where it loses in terms of accuracy. , a signature of size 20 bits (4 bits per dimension) in a 5-dimensional data space can address 100M different data points. However, the number of bits needs careful tuning, and unfortunately, there are no guidelines for such tuning. To make the quantization more dynamic, instead of quantizing based based on quantiles, the IQ-tree [9], which is a three-level index structure, quantizes based on a regular decomposition of the page regions of the index.

Identification of partitions The aim of indexing is to facilitate and speed up query retrieval on database. For point queries, iMax is easy to implement and efficient computationally, since it only needs to calculate the iMax value of the query point and search the B+ -tree directly. To perform a range search using the iMax, the algorithm first checks the partitions that overlap with the query region, which is a hyper-rectangle in high-dimensional data space. It then computes every subquery range with respect to the iMax space, and for each subquery, it traverses the tree once.

D1 (1,1) 1 111111111111 000000000000 (0,0) 1 d0 Fig. 6. 1% for range query. 5, the query hyper-cube always intersects the central line of data space. In this case, a range query will always be transformed into d subqueries, since every partition is intersected by the hyper-cube. When a range query is conducted on a d-dimensional data space, the query is divided into d subrange queries, each of which is a single-dimensional iMax value range. iMax value only presents the largest attribute on one dimension, so, when we search based on such values, the candidate data point set is large.

Download PDF sample

Rated 4.45 of 5 – based on 29 votes