earticle

영어

Data warehousing is gaining importance day by day in enterprises, as it helps them to improve their business intelligence. The process of creating a data warehouse needs to be automated so that the transactional sources are generated in least time, with maximum accuracy and with minimum dependability on users. This automation proves its worth particularly when working with small and medium enterprises, where hiring of new people just for creating data warehouses can be unaffordable. The technique presented in this paper automates entity relationship model into data warehouse logical model to generate semi-star schema by using artificial neural networks. More precisely, the step of differentiating dynamic behavior dimensions from static behavior dimensions has been automated by using feedforward back-propagation neural networks. This network ascertains dimensions which are sensitive to changes. The network is trained for all the possible values of inputs and has been tested for actual results. The performance of proposed technique is evaluated by comparing certain metrics like simplicity and minimality with existing data warehouse creation techniques.