Automatic Progressive View Maintenance in Datamining and Warehousing

A data warehouse organizes and stores consolidated data derived from distributed, autonomous data sources for OLAP (Online Analytical Processing) and data mining analyses. View maintenance involves the process of propagating the changes to the data warehouse. The duration may be too long for some systems. In this research, we applied a multi-agent approach to enable continuous updating of data warehouse views as transactions are executed at the sources. This technique, called Automatic Progressive View Maintenance (APVM), eliminates view maintenance down time for the data warehouse-a crucial requirement for Internet-based applications. Through the use of cooperation between agents, the data consistency problem usually associated with APVM is solved. In addition, a fuzzy agent scheduling system was developed to prioritize tasks for the agents. The results from this research showed that the proposed multi-agent system drastically increases the availability of the data warehouse while preserving a stringent requirement of data consistency.