Nottingham Trent University
The k-Nearest Neighbor (k-NN) algorithm has been a promising classification tool. In spite of its extensive application, k-NN suffers from few inherent problems. A considerable number of the proposed approaches have exhibited quite promising results and has motivated further research on improving the k-NN method. In this paper, the authors devise a dynamic nearest neighbor classifier for data integrated via generalization. Here, they make use of OO concept generalization to integrate the data collected from the different service providers, having varying formats, into a single consolidated data unit (training instance).