Figure 3 - uploaded by Ratko Orlandic
Content may be subject to copyright.
Source publication
This paper addresses the problem of supporting region queries over a dynamically growing set of data in feature spaces with many dimensions. It formulates two principles of indexing data in high-dimensional feature spaces and presents an approach to high-dimensional indexing with these principles in mind. The combined goal of the two principles is...
Context in source publication
Context 1
... [OLS02], we proposed a new space-partitioning strategy called Γ. is statically partitioned by several nested hyper-rectangles, called generators. The low endpoint of each generator lies in the origin (low left corner in Figure 3) of the space. The space between a generator and the one it immediately encloses, if any, is called a Γ subspace. ...
Similar publications
Data warehouse has become more and more popular for an enterprise as a data repository system. Yet tools to appropriately design its conceptual model are rarely available, even though this model is known as a key for the successful of the overall design. In this paper we propose an approach and a tool to guide the decision makers in designing data...
Designing an efficient Distributed Database System (DDBS) is considered as one of the most challenging problems because of multiple interdependent factors which are affecting its performance. Allocation and fragmentation are two processes which their efficiency and correctness influence the performance of DDBS. Therefore, efficient data fragmentati...
XML is one of the premier formats for data representation and interchange. Many organizations are starting to store data in XML and using XML as an intermediate format for publication and use of these documents. Most database systems have support for storing data in XML or internally representing XML data for storage. However, XML does not have a s...
There are many issues that affect modeling and designing of realtime databases. One of those issues is maintaining consistency between the actual state of the real-time object of the external environment and its images as reflected by all its replicas distributed over multiple nodes. An efficient replica control algorithm can contribute significant...
Citations
... Since this technical document is intended to provide a complete description of the technical aspects of the D* system, we repeat here some information and algorithms already published in [Kul06], [Luk04], [Orl05], or [Orl06]. In the rest of the document, Section 2 summarizes related work. ...
... As we will see in Section 3, the first design principle deals with the grouping of multi-dimensional data in index pages, whereas the second deals with their representation within the indices. The pursuit of these principles can result in significant improvements of retrieval performance [Orl06]. The principles of clustering and cluster representation have been pursued in some contemporary multi-dimensional access methods. ...
... However, the adherence to one principle does not imply the adherence to the other. Therefore, these principles are largely orthogonal design goals [Orl06]. ...
D* is a novel system for data storage and retrieval appropriate for advanced scien-tific studies, as in high-energy physics, environmental sciences, and astronomy. The design of the D* system is based on certain principles of organizing and ac-cessing multi-dimensional data on storage, whose pursuit requires that the storage system acquire a greater knowledge about the data. This provides a basis for a tighter integration of data storage and data mining technologies. Through the ap-plication of innovative retrieval and clustering techniques for high-dimensional data, D* can support high-performance data access and provide data mining ap-plications useful insights into the data that can facilitate subsequent processes of data preparation and data mining. The basic processes of the D* system include data clustering, space partitioning, data loading, and data retrieval based on region queries and similarity searching. D* scales well with increasing data dimensional-ity and works well on incremental load of data.
