Author:
Carsten Kropf
Affiliation:
Hof University, Germany
Keyword(s):
Database Architecture and Performance, Indexing, Spatio-Textual Indexing, Hybrid Index, Data Structures and Data Management Algorithms.
Related
Ontology
Subjects/Areas/Topics:
Data Engineering
;
Data Management and Quality
;
Data Structures and Data Management Algorithms
;
Database Architecture and Performance
;
Databases and Data Security
;
Databases and Information Systems Integration
;
Enterprise Information Systems
;
Large Scale Databases
;
Open Source Databases
Abstract:
This paper presents the optimization of the reorganisation algorithms of hybrid index structures supporting multimedia search conditions. Multimedia in this case refers to, on the one hand, the support of high dimensional feature spaces and, on the other, the mix of data of multiple types. We will use an approach which may typically be found in geographic information retrieval (GIR) systems combined of two-dimensional geographical points in combination with textual data. Yet, the dimensions of the points may be arbitrarily set. Currently, most of these access methods implemented for the use in database centric application domains are validated regarding their retrieval efficiency in simulation based environments. Most of the structures and experiments only use synthetic validation in an artificial setup. Additionally, the focus of these tests is to validate the retrieval efficiency. We implemented such an indexing method in a realistic database management system and noticed an unacce
ptable runtime behaviour of reorganisation algorithms. Hence, a structured and iterative optimization procedure is set up to make hybrid index structures suitable for the use in real world application
scenarios. The final outcome is a set of algorithms providing efficient approaches for reorganisations of access methods for hybrid data spaces.
(More)