Multi-Agent Driven Rule-Based Dss In Data Warehouse

Data warehouse continues to play an important role in global information systems for businesses. While applications of data warehouse have evolved from reporting and decision support systems (DSS) to mission critical decision making systems (such as spacecrafts, flood controlling, earthquake prediction etc) this requires data warehouses to combine both historical and current data from operational systems. In a data warehousing process, the phase of data integration is hence very crucial. Indeed, many methods for data integration have been published in the literature in these recent years. However, with the development of the Internet, the availability of various types of data (e.g. images, texts, sounds, videos, databases...) have increased, the processing of these huge data and discovering related patterns to leading to knowledge representation and decision making is a difficult task one cannot ignore.The application of data warehouse technique and multi-agent systems to modern decision support system (DSS) not only that they assist supervisors of enterprise to make an analysis of strategic decision and to improve the competitive power of the enterprise but also assist in other crucial areas such as health sector. A good example is to consider cognitive models which capture clinical reasoning which can help to design medical tools whose behavior can be closer to doctor’s, physician’s or nurse’s reasoning.In this thesis we propose architecture for heterogeneous data integration based on Multi-Agent Driven Rule-based Decision Support System in Data warehouse. In accordance with Data warehouse structure, a Multi-Agent System (MAS) is designed to extract data from different sources, build rule-based model and lastly provide decisions for the end user. Our aim focus in building a model by the use of data mining technique and generate some rules which provide high degree of accuracy to help the clinicians to attain a correct decision.Lastly, we describe our multi-agent system in Breast cancer diagnosis system as our case study. The clinical datum is extracted, transformed and loaded known as (ETL) into the Data warehouse. Data mining techniques particularly Decision Tree, and how they can be applied to knowledge discovering to finally help the user to make a proper decision in the daily life is our main focused subject in this work. We actually focus on proposing an architecture which gives outputs with high degree of accuracy compared to previous architectures implemented by different authors.