Choosing Exploration Process Path in Data Mining Processes for Complex Internet Objects

摘要

We present an experimental case study of a novel and original framework for classifying aggregate objects, i.e. objects that consist of other objects. The features of the aggregated objects are converted into the features of aggregate ones, by use of aggregate functions. The choice of the functions, along with the specific method of classification can be automated by choosing of one of several process paths, and different paths can be picked for different parts of the domain. The results are encouraging and show that our approach allowing for automated choice, can be beneficial for the data mining results.