On Tuesday, 5th of May 2020, 09:00 AM, M.Sc. student “Refed Adnan Jalil” defended her thesis entitled:

“Design and Implementation of Intelligent Decision Support System based on Optimized Data Warehouse”

The discussion committee included:
Prof.Dr. Abdel Moneim Saleh Rahma / University of Technology, Department of Computer Science / Chairman.
Assistant Prof.Dr.Mohammed Faleh Abdel Karim / College of Information Engineering/member.
Assistant Prof.Dr. Methaq Kata’a Student / Al-Mustansiriya University / College of Science.
Prof.Dr. Talib Mohammed Jawad / College of Information Engineering/member and supervisor.

Abstract :

Various government bodies are enhancing their decision-making capabilities using a data warehouse, which acts as a middleware in Intelligent Decision Support System architecture (IDSS). It is one place for the storage and participation of data, which users can readily utilize to produce improved decisions, instead of accessing the entire various databases. The essential requirement for an efficient data warehouses is precise and timely united information along with fast and effective query response time. Due to the vast number of databases within Information and Telecommunication Company (ITPC) obtaining the data in a successful method requires a harmonious potential through the three systems of Document, finance, and human resource. Thus, This thesis presents proposed framework of IDSS for ITPC that contains five layers (Interface layer determines data sources, Extract Transformation Load Layer (ETL) determine ETL design process in SQL Server Integration Service (SSIS), Data warehouse layer determines independent data marts that conform Galaxy Schema, Quantum Particle Swarm Optimization (QPSO) Layer determine the best-materialized view, and Knowledge layer determines Naïve Bayesian (NB) and C4.5 Classification models).

For getting IDSS that have fast responding query, low cost for handling the query, low area value, and efficient decision making the QPSO has been applied for selecting best views to be materialized and for applying NB and C4.5 classifiers models on best materialized view. The Results shows the data mart building advantage by routing the same queries to the data marts and databases, queries through Materialized View are more efficient than direct access by up to 402.38%, and the C4.5 have better accuracy up to 94.46% than the NB which equal 91.07% for predicting employee performance.