Visualizing knowledge for data mining using dimension reduction mappings

In typical data mining applications we often have large amounts of data at our disposal along with knowledge often available in quite different ways such as rules, cases, analytical models or correlations among variables. Many classical machine learning methods may result inadequate in this scenario because they seldom allow to make use of all the knowledge that we might have at hand. Visualization techniques that have been used for a long time for data visualization can also be used to visualize certain forms of knowledge, resulting in a more efficient data mining process. We present a unifying approach for knowledge visualization based on dimension reduction (DR) that allows to represent rules, cases, models and correlations on a low-dimensional visualization space in a consistent way.