This report describes the design and implementation of the AUBA tool, which is a web usage mining tool for AUB. The goal of the project is to give AUB a tool to analyze the behavior of the users of Auboline and to be able to establish success rates for the different functions in the system.
A post-processor has been implemented to accommodate the ETL process of extracting the data from the log files, transforming it to fit the data warehouse format and finally loading it into the data warehouse. The data warehouse consists of two star schemas that are designed to enable general as well as domain specific analysis of the data.
Audit control and other data quality assurance activities have been integrated in the ETL process to make sure that the data is not flawed.
The analyses made with the AUBA tool give the AUB staff information that they have not previously been able to learn about the use of Auboline. Analysis has been made easy with a simple user interface, and query performance has been optimized by using summary tables.
A user survey has been conducted to find out if the results found with the AUBA tool match the users' own opinions of their behavior in Auboline. The results of the two types of analysis are comparable for the most part.