Abstract
With the development of Internet and Intranets, the exchange and sharing of the enormous amount of information from various data sources scattered across the Web and in different organizations have become crucial issue. To meet these needs, integration solutions have been proposed by offering uniform interface to end users. These solutions concerns three main dimensions: data integration, application integration and platform integration. This talk focuses on three phases of lifecycle of data integration systems (e.g. data warehouses): (1) construction, (2) exploitation and (3) personalisation. For the construction phase, a classification of existing integration systems is given by on three criteria: (a) data representation, (b) sense of the mapping between the global and local schemas and (c) mapping automation. This classification facilitates the understanding of existing data integration systems. We also present a semantic data integration approach dealing with ontology-based database sources (OBDS). An OBDS is a source that contains a local ontology explicating its semantic. This local ontology references one or several shared ontologies. Three integration scenarios are presented based on company requirements. Once data integration built, solutions to facilitate its exploitation need to be developed. In this talk, we develop these solutions for an example of a materialized data integration system, which is data warehouse, used for business intelligence applications. A panorama of optimisation structures used to speed up complex queries is presented. Selection algorithms of two examples of optimization structures: horizontal partitioning - considered as a no redundant structure and bitmap join indexes - considered as a redundant structure are presented in isolation and joint ways. Some issues on designing parallel data warehouses are given. A tool assisting database administrators during their exploitation tasks is presented. This tool is connected to commercial DBMS. A data integration system is usually accessed by a large number of users (or decision makers), where each one has its own preferences and profile. These profiles should be included during the exploitation phase and may influence the exploitation phase. Finally, our talk ends by presenting some research directions related to the three phases of the lifecycle of data integration systems, in general, and data warehouse in particular.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have